Originally Posted by RnRollie
Actually, maybe iso folding cupcakes out of the sheet, maybe using it as a strip to carry the heat away to the box walls might be an idea. Make the box into a large thermal buffer. Thou, probably those strips would be too fragile.
Perhaps not quite cup cake cases but yeah you have the general idea.....no need to try and extend the graphite sheet to the box....it would be too far to be an efficient form of heat transfer...and there's no need when there' s 20 liters of liquid butane surrounding the cpu at -20 to -30c....ain't that enough buffering for ya?
Originally Posted by RnRollie
AFAIK, the panasonic procedure creates graphite
sheets which act as a 'special' IHS by transporting the heat away from the source to the edges of the sheet (hence the application in mobile phones - transporting the heat to the enclosure); and would therefore be a terrible TIM. As it actually blocks the heat from going straight up/through.
Although I'm not using this as a TIM and you are right that it is not intended for this use, none the less the heat tranfer along that perpendicular plane is not absolutely terrible at 10 - 26 w/m/k......its actually comparable to a very good TIM in that respect.
This straight up/down transportation is still however important, the heat doesn't actually come out of the edges....even if it did say that in the semi accurate article...I'd like to see how they make an edge connection to the 10 micron sheet edge...lol.
The way it works is that the huge lateral conduction spreads the heat over a large area which then dissipates in the more poorly conducting direction vertically through the surfaces, but because of the large area the heat becomes spread over then the poorer vertical conductivity makes little difference as the total heat transferred depends on the conductivity the heat gradient and the AREA of the transfer interface, so all those unit areas each with 10 -26 w/m/k add up to a lot of heat transfer.
My main concern with using this is actually that the lateral conduction will not be enough, because just like we have to consider the AREA of the surface you also have to consider the cross sectional area of the film as even though it has large conductivity it has very little thickness......as its 70microns and has about 2.5X the conductivity of copper then it would be like using a 70x2.5 = 175 micron or 0.175mm thick copper sheet to conduct the heat laterally.
From my understanding of studies on water blocks there is an optimal thickness for the base plate, too thick and it doesn't transfer heat well and too thin and it doesn't distribute the heat adequately.
It is this latter scenario I fear may be the case....but again that is why it is important not to disregard the vertical heat dissipation....as stated it has the same conductivity as a good TIM in this direction so I'm hoping that an equivalent amount of heat that would normally be passed into a heat sink will just pass vertically up and be dissipated into the liquid, this is actually crucial that this happens (and I'm making the assumption here that the IHS actually spreads the heat well over its entire surface?), any further dissipation from the fins I'm going to place over the center of the cpu and further dissipation from the sheet that extends beyond the IHS will really just be a bonus compared to what a standard heat sink would do.
It is experimental...just hope it works.
Think I've just wasted my money on this graphite sheet......although it may have great lateral conductivity if you start plugging in some figures to work out how much temp delta you get for a given heat energy input then it works out very poor.....simply because it is so thin it can't transfer much energy.......may look great on the demo's with an IR camera showing the heat spread but that doesn't show the quantity of energy being moved...just the temp distribution. Edited by technogiant - 8/20/13 at 9:39pm