The Great Thermal Interface Material Con

Don
OP
Joined
7 Aug 2003
Posts
44,276
Location
Aberdeenshire
The final results are in!

First up AS5:-

liquidpro1.JPG


Load 48 (and maybe pushing 49)
Chipset 32

Next up, the new challenger for the crown, Coollaboratory Liquid Pro:-

liquidpro2.JPG


Load 47
Chipset 32

Liquid Pro pips AS5 to the post, despite being ever so slightly warmer (chipset reached 32C a lot quicker) it held load temp at least 1 (!) C cooler than the AS5.

As regards the curing of AS5, I'm not overly convinced of the gains that can supposedly be had as I've seen little to back it up.

Jokester
 
Soldato
Joined
28 Mar 2006
Posts
4,379
Location
Jarrow, Tyne And Wear
thats stupid since solids are naturally better conductors of heat than liquids, i tried my CNPS9500 without any TIM and it worked fine, 2*C below what i was getting using ceramique, was as close to perfectly flat as one can get (no lapping, came that way) so contact was very good :) give it a try sometime, remember solids are better conductors than liquids so copper is gonna conduct heat quicker and more efficiently than some layer of goop and then copper
 
Soldato
Joined
5 Jan 2003
Posts
5,001
Location
West Midlands
Gashman said:
thats stupid since solids are naturally better conductors of heat than liquids, i tried my CNPS9500 without any TIM and it worked fine, 2*C below what i was getting using ceramique, was as close to perfectly flat as one can get (no lapping, came that way) so contact was very good :) give it a try sometime, remember solids are better conductors than liquids so copper is gonna conduct heat quicker and more efficiently than some layer of goop and then copper

But AIR is a lousy conductor of heat. The whole point of the thermal compound is to allow metal to metal contact where possible, BUT to make sure any minute air pockets are avoided, and are instead filled with the compound as the 'next best thing' to metal > metal contact.
 
Associate
Joined
8 Sep 2005
Posts
303
Location
Leixlip, Ireland
Gashman said:
thats stupid since solids are naturally better conductors of heat than liquids, i tried my CNPS9500 without any TIM and it worked fine, 2*C below what i was getting using ceramique, was as close to perfectly flat as one can get (no lapping, came that way) so contact was very good :) give it a try sometime, remember solids are better conductors than liquids so copper is gonna conduct heat quicker and more efficiently than some layer of goop and then copper

Solids do conduct heat better then liquids... at least stationary ones.

AS5 is a suspension though... mostly liquid with disolved or suspended mettalic particles. Not suprising that a 100% liquid metal in between would give better conduction.

Running without any TIM and getting better results is either down to their being some residule TIM on one of the contact surfaces or there being too thick a layer of TIM when testing with that, or both. It is possible you have very finely polished surfaces, but it should still benefit from a little TIM.

According to my good friend wikipedia, after testing some liquid silver alloy we should get to work on tracking down some liquid diamond paste :D http://en.wikipedia.org/wiki/Thermal_conductance#Examples

Im wondering though... considering that the chip being tested has an integrated heatspreader...

Im betting that the direct contact heatsink test would be different if it werent for that rough heat spreader... but you would neet a little toothpaste to get contact to all the chip.

Seriously though, the toothpaste is partially evaporating as the test is going, so your actually using a phase change cooling system vs air cooling. I know toothpaste burns, and burns at a pretty low temperature so i would be carefull with any tests over an hour at over 50 C
 
Last edited:
Don
OP
Joined
7 Aug 2003
Posts
44,276
Location
Aberdeenshire
*Update*

Took the waterblock off my CPU and the Liquid Pro had dried out and welded itself to the both surfaces in an uneven distribution. Took 20mins of elbow grease and Brasso to get the worst of it off before reapplying it again.

Jokester
 
Soldato
Joined
18 Aug 2005
Posts
13,170
Location
Shropshire
Jokester said:
*Update*

Took the waterblock off my CPU and the Liquid Pro had dried out and welded itself to the both surfaces in an uneven distribution. Took 20mins of elbow grease and Brasso to get the worst of it off before reapplying it again.

Jokester


Ouch :(
 
Back
Top Bottom