This thermal paste has "only" Thermal conductivity (W/mK): 8.5
I have seen some thermal plates in our Office with 220w/mk
I think, as soon as I have my S3 I will replace the original paste with those ...
What is it...220W/mk...
Could you post the brand of it, specs...where to buy...more details, please...
Thanks,
ZiG
I have to ask our EE in the Office for more info on Monday.
@ aztecminer: It IS electricity conductive .. BUT since it is NOT a paste, there should be no Problem with that
as long it is cut the right size
If it's a pad (as opposed to a copper shim which is kind of useless as a TIM) it's almost certainly a graphite sheet like this one.
http://www.digikey.com/product-detail/en/A10462-03/926-1008-ND/2268505240W/mK, but only in the x-y direction. If you look at the datasheet, it has 5W/mK, which will be much worse than a standard paste as a thermal interface between chip and heatsink. They're used as heat spreaders.
8.5W/mK is fine for a thermal paste, the bulk conductivity is almost a meaningless measurement. Keep in mind that the bulk thermal impedance (in degrees C (or K) per watt) is distance / (area * conductivity). If your surfaces are pretty small and you have a bit of pressure, the bond line (distance between the heatsink and chip) can be on the order of 1 mil (0.0254mm). Even at 0.1mm and for something like a VRM chip with a 1~cm square chip, 8.4W/mK is a temperature rise of 0.12C per watt. Even if you had a paste that was rated at 1W/mK it'd only be a 1K/W temperature rise, which when you're only looking to dissipate a few watts is essentially meaningless.
Even for the main board (assuming it's around 120mm x 250mm or so like the S1 board) it doesn't make a difference there. Even if your gap is 1mm (which is huge), that's still 0.001m/(.12m*.25m*8.4W/mK) = 0.004K/W. Sure there's some simplifications assuming a fairly even distribution of heat in the board, but for 200W per board you'd still only be looking at a 0.8C rise over the thermal compound.
tl;dr, the bulk conductivity of a paste is almost meaningless. What really matters is how well it wets the surface and how small a bond line it can give you. How well it resists pumping out is important too, but not so much in mining equipment that should run at temperature 24/7. For something like the interface from the board to the heatsink for an S3, you'll want something that has a fairly low viscosity and spreads really well, since you'll have pressure points at the screw heads instead of nice even pressure, and getting a consistent amount of TIM across the whole board would be a challenge.