Does the type of thermal grease make a difference?

  • Thread starter Thread starter altcomphardware
  • Start date Start date
altcomphardware said:
Apart from a few degrees Celcius difference, is there much point in getting
some fancy thermal compound costing 10x more than a cheapo one?

Nope, that is all the difference it makes, a few degrees at most.
Does the performance of thermal compounds degrade over time as they dry out?

Nope, because its sandwitched between the top of the cpu and the heatsink.

Worth cleaning it off and replacing it if you remove and replace the heatsink tho.
 
altcomphardware said:
is there much point in getting some fancy thermal compound costing 10x more
No.

Does the performance of thermal compounds degrade over time as they dry
out?

I have an audio amplfier and TV from the 1970s that have never been
regreased, but I don't expect them to fail any day now from this gross
negligence.
 
I have an audio amplfier and TV from the 1970s that have never been
regreased, but I don't expect them to fail any day now from this gross
negligence.


The thermal density (discrete component die to it's package
spreader to heatsink) on any properly designed audio
amplifier (particularly output pairs of BJT or Mosfets) is
far lower than on a modern CPU. A system can easily have
full load for extended periods of time where single point
heat load is over 100W. How often does your amp see (at a
bare minimum, 2 pairs of output devices for stereo sound) 4
x 100, 400W RMS?
 
Apart from a few degrees Celcius difference,

Usually the largest difference in temp results if comparing
a compond too thick to one of proper viscosity, or trying to
mate irregular heatsink (base surface) to the part. In this
latter case, the obvious solution is to reject heatsinks
with a poor base finish or if you consider it reasonable to
lap a heatsink, to do so. However, many heatsinks with poor
surface finish aren't very good in general, there may be
diminishing return in trying to polish a poor one to a
mirror shine (and in general, mirrorlike surface is not
needed).


is there much point in
getting some fancy thermal compound costing 10x more than a cheapo one?

Don't get the cheapo one. Why do people always try to
contrast and consider extremes? Of course if you consider
extermes, on the one end you get junk and on the other you
pay more than it's really worth- same as with just about any
product.
http://www.frostytech.com/articleview.cfm?articleID=803
http://www.dansdata.com/goop.htm

Does the performance of thermal compounds degrade over time as they dry
out?

It is useful that:

- Compound is thinner (but this need not necessarily be
high-end product BUT if very low end it means a lot of
silicone oil which will separate from the solids that much
faster). This is for good mating surfaces, if your heatsink
is very rough, the compound viscosity needs be slightly
higher so it doesn't run out. Such 'sinks should be avoided
altogether (in general) but there are cases where this isn't
so easy, for example on a video card that has a special
proprietarily shaped 'sink that is either difficult to lap
or you don't want to void the warranty visually by lapping
it. If the only problem is lapping a surface that is not
entirely on the same plane (staggered such that it makes
contact with different height PCI Express to AGP bridge chip
or memory chips, etc) one possibility is using a small scrap
of heatsink with a flat base as a sanding block.

- Less is applied (goes back to prior point unless one has
plenty of experience applying, and later examining, prior
similar heatsink interfaces

- Synthetic base oil, not silicone so the solids stay in
suspension better, not pumping out and drying as much over
time and repeated thermal cycles.


I suggest Arctic Alumina. Not as expensive as Arctic
Silver, but still synthetic based, reasonably thin,
non-capacitive, and cleanup is easier as you don't have to
get every last trace of it off of parts (as with Arctic
Silver, which leaves a silvery mess unless parts are
thoroughly wiped or washed).
 
altcomphardware said:
Apart from a few degrees Celcius difference, is there much point in
getting some fancy thermal compound costing 10x more than a cheapo one?

http://www.frostytech.com/articleview.cfm?articleID=803
http://www.dansdata.com/goop.htm

Does the performance of thermal compounds degrade over time as they dry
out?

Assuming a "decent" thermal compound you will probably find more
temperature difference attributable to the flatness of the heatsink than
to the brand of the compound. I've found that lapping the copper base
that contacts the CPU can easily yield 2 degrees of improvement under
full load. Of course some compounds are supposed to last longer than
others but I've never found any, even the cheap stuff that comes with
off-the-rack heatsinks, that could actually be called bad.
 
kony said:
On 8 Aug 2006 13:21:14 -0700, "altcomphardware"

The temp differences between the various different thermal compounds
are around the 1C mark, which makes no real world difference to 99.9%
of computer users. There is therefore no reason to buy anything but
basic zinc oxide in grease.

If your HS base is damaged, rubbing it on very fine sandpaper on a
piece of glass is a quick way to remove any sticky-out nicks. If youre
building PCs, its quicker to just relagate damaged sinks to cpus that
dont get hot enough for a problem to occur.


Toothpaste will, though it performs just as well as proper compounds
when applied.


NT
 
The temp differences between the various different thermal compounds
are around the 1C mark, which makes no real world difference to 99.9%
of computer users. There is therefore no reason to buy anything but
basic zinc oxide in grease.

If your HS base is damaged, rubbing it on very fine sandpaper on a
piece of glass is a quick way to remove any sticky-out nicks. If youre
building PCs, its quicker to just relagate damaged sinks to cpus that
dont get hot enough for a problem to occur.



Toothpaste will, though it performs just as well as proper compounds
when applied.


You are quoting things I did not write.

However, it is known that silicone based grease compound
(ALL of them) do degrade over time when used at the thermal
densities of modern CPU... but far moreso the flipchips than
those with heat spreaders as the latter is a lower thermal
density endured by the grease. I have personally pulled off
heatsinks where the CPU was beginning to overheat from this
degradation of otherwise decent silicone oil based grease,
and it was a properly applied grease in the beginning, the
system had ran for 18 months or so then began crashing.
 
kony said:
On 9 Aug 2006 14:57:33 -0700, (e-mail address removed) wrote:
You are quoting things I did not write.

indeed :)
However, it is known that silicone based grease compound
(ALL of them) do degrade over time when used at the thermal
densities of modern CPU... but far moreso the flipchips than
those with heat spreaders as the latter is a lower thermal
density endured by the grease. I have personally pulled off
heatsinks where the CPU was beginning to overheat from this
degradation of otherwise decent silicone oil based grease,
and it was a properly applied grease in the beginning, the
system had ran for 18 months or so then began crashing.

AIUI it is mainly the solid component in the compounds that conduct
heat, whether its zinc oxide, copper or silver. In which case the
oil/grease degrading would have little effect on thermal conductivity.
I thus have to wonder if the cause of the problem was cpu overheating
_not_ as a result of the thermal compound.


NT
 
indeed :)


AIUI it is mainly the solid component in the compounds that conduct
heat, whether its zinc oxide, copper or silver. In which case the
oil/grease degrading would have little effect on thermal conductivity.

A nice theory but wrong.

You end up with caked islands of solids that don't even
cover the whole core anymore. It is vital for grease
performance that it not dry out.

I thus have to wonder if the cause of the problem was cpu overheating
_not_ as a result of the thermal compound.

Wonder whatever you like, I'm 100% sure.
 
altcomphardware said:
Apart from a few degrees Celcius difference, is there much point in
getting some fancy thermal compound costing 10x more than a cheapo one?

http://www.frostytech.com/articleview.cfm?articleID=803
http://www.dansdata.com/goop.htm

Does the performance of thermal compounds degrade over time as they dry
out?


It is if you plan to overclock your CPU and a few degrees makes a
big difference. Otherwise it doesn't make much difference.

--
I used to have abs. Now, I've just got ab.
One big ol' Ab. - BigSkiff www.titanspot.com

Pyongyang sounds more like the sound effect an ACME catapult makes
as it goes off at precisely the wrong moment for Wile E. Coyote. -
Cadbury Moose
 
Back
Top