Try to follow me a bit and see if i didnt make a error:
If i take the Zalman CNPS7700-Cu as a ref:
http://www.zalman.co.kr/eng/product/...45&code=005009
It has a thermal resistance of 0,19-0,24 °C/W
(surface coils=0,3268m2 !!!!, if you use the 0,19 and calculate backwards you get a Length of your coils totalling 25 meters!! insane)
With 150 watt generation by the CPU it would mean you need to cool the surface of the coils with 28,5-36°C to keep the CPU at a stable temperature.
Based upon AMD64 Athlon:
Pnew=Pspec x Vnew/Vspec x Fnew/Fspec => 150=89 x 1,9/1,5 x 3500/2600
(just to calculate with something)
Gallium:
CPU watt generation = 150 watt
Surface heat transfer = 0,0014m2
a) Heat transfer coeff = 2,4x1E5 W/m2.K (flow rate 2 cm/s)
b) Heat transfer coeff = 5x1e4 W/m2.K (flow rate 0,1 cm/s)
A)
150=240000x0,0014xdT => dT=0,45°C
B)
150=50000x0,0014xdT => dT=2,14°C
So again if you can cool down the flow of Gallium with more then 2,14°C then you can get ride of the 150 watt and keep the Tcpu stable.
(click here to see the calulations of the Heat transfer coeff.)
http://img.photobucket.com/albums/v2...k-Yu/gal01.jpg
http://img.photobucket.com/albums/v260/Fook-Yu/gal2.jpg
I will continue with the calulation of a thermal resitance number for the heatsink needed to facilitate this in a moment. I will not spend time on the convectivity of the Gallium flow => heatsink as i can pretty much make the surface several times larger then that of the CPU, thereby assuming that the thermal resistance of this part will be of no or extremly little influence.
Assumption:
- Energy los in piping is minimal => 150 watt remains
- heat transfer coefficient turbulent air through finned plate 50 W/m2.k (this is high!! and definitly not soundless)
150 = 50 x A x (dT)
Option 1) CPU temp 30°C air 20°C
A => 0,3 m2 (that is the heatsink of the Zalman and we already noticed the problems with the thermal resistance in that unit)
Option 2) CPU temp 40°C air 20°C
A => 0,15 m2 only halved, not enough.
So working with a liq. metal will enable you to transfer almost any rate of heat development (watt) from your cpu at soundless level. But you get in troubles cooling the gallium down with air inside your PC. Making a system that leaves your PC will be to expensive with a price of 3$/g of Gallium (remember the density of 6 kg/L).
Then you could compine it with something else. However as you can't go below the 16°C you can forget using phase cooling to directly cool the gallium, because you will freeze your medium.
What remains is a combined system of watercooling and liq. metal. Guess how the do it in nuclear plants were the idea of using liq. metal for cooling was taken from
. But A system with only water will be enough to pull the heat It is simply a matter of flowrate and water temperature.
No advantage then what so ever. i really think not for a CPU. Simply because the temp levels are not extreme enough. Gallium has one huge advantage not that it start to melt at 30°C but it first starts to boil at 2200°C.
Why use it in a graphics card? I guess because you can reduce your overall thermal resistance by enlarging your contact surface. AND because a GPU produces less watts then a Cpu that make it possible to have a small air cooled circuit.
If i made a mistake in my assumption please comment!!!?
Linear Mode
