Manfred,
Thank you for this advice. I've been thinking about it ever since you sent
it. Today I got to see for myself the difference in thermal greases.
To make a long story short, replacing the original generic silver grease
(probably from Asia) between the CPU and its heat sink greatly lowered the
CPU core temps and eliminated all the problems I was having due to the CPU
overheating.
The thermal grease I used contains boron nitride flakes. It's *Arctic
Silver Premium Ceramic Polysynthetic Thermal Compound*. It's kind of an
off-white color. I even put in on thicker than I should have.
Before replacing that grease, it was not uncommon to see the CPU core
temperature over 190 degrees F. I now cannot get it over 120 F, even with
both cores at 99% for an extended period of time.
Thanks again!
73, Mike
www.w0btu.com
On Thu, Feb 5, 2015 at 11:38 AM, Manfred Mornhinweg <manfred@ludens.cl>
wrote:
> Mike,
>
> the thermal conductivity of such a paste depends more on the granulometry
> of the solids in it, than on the actual material.
>
> Thermal compound is usually a mixture of finely ground thermally
> conductive material, and some sort of oil or other thick liquid. The
> problem is that all suitable liquids have very low thermal conductivity,
> compared to even a modest ceramic material, let alone metals. So, whether
> the granules are alumina with a thermal conductivity of 18 W/(m*K), or
> silver with 419 W/(m*K), makes little difference when the voids between the
> granules are filled with an oil having a thermal conductivity of 0.21
> W/(m*K)!
>
> And between silver and copper (372 W/(m*K)) there is essentially no
> difference, when the voids are filled with that oil.
>
> But it makes a lot of difference whether all granules are balls of the
> same size (worst), or flakes of widely varying sizes having flat sides
> (that would be a good filler). With those flakes, there is far less space
> between particles that has to be filled with the liquid, and this strongly
> improves the thermal conductivity of the compound.
>
> I would think that manufacturers of premium thermal compounds have their
> own technologies to make filler powder with carefully optimized particle
> size and shape. It should be such that under pressure the particles arrange
> to squeeze out as much liquid as possible, so that the final compound layer
> is mostly solid, with only very small spaces filled by the liquid.
>
> Also it would help a lot to have liquids with higher thermal conductivity,
> that are stable enough to use them in thermal compounds. Water (0.6
> W/(m*K)) is much better than most other liquids, but of course it dries off
> quickly, so it's useless in thermal compounds.
>
> Cheap, lightly loaded thermal compounds typically have a thermal
> conductivity of only around 0.3 W/(m*K), so it's only slightly better than
> using unloaded plain grease. At least these are easy to apply and spread
> out, producing a thin layer. Premium thermal compounds fal into the range
> of 3 to 5 W/(m*K) - some manufacturers may claim even slightly more, and
> it's up to teh user to believe that. But these premium compounds have lots
> of solids and little liquids, so they are harder to spread out in a fine
> enough layer. Well applied, they are indeed much better than the cheap
> compounds, but used in a casual way they might not bring much improvement,
> because the layer ends up thicker.
>
> Instead if you solder down the parts to the heatsink, you get a thin
> layer, along with the thermal conductivity of solder, around 50 W/(m*K).
> This is very much better than any other mounting system allows, but it's
> not always possible to do, it gives no electrical insulation, and there can
> be problems coming from stress cracking due to different thermal expansion
> of the parts.
>
> Manfred
>
_______________________________________________
Amps mailing list
Amps@contesting.com
http://lists.contesting.com/mailman/listinfo/amps
|