Direct die phase change, possible? (refrigerant actually touching die)

[xDUCK]

As the subject says... Is this possible? The only problem I see is that you would have to have a valve at the cooling head (and remember to shut it off, lest you lose all your refrigerant every time you remove the head :eek: ). I understand that it is under very high pressure, which is why I am unsure if such a system would be possible.

Anyone care to enlighten me? :)

[DaBit]

Well, I see a few problems in doing so. The first one is pressure. This is the highest during off-time, and it depends on ambient temperature and refrigerant. It could run up to about 16 bar (~240psi). That's 16kg/square centimeter, or 240 pounds per square inch. An Athlon will certainly crack. A P4's heat spreader could possibly hold the pressure, but then you are better off removing the heat spreader and using an evaporator block instead.

The second problem is to seal the evaporator (well, the shell with the suction line and cap tube) from the CPU, but clever mounting and Viton O-rings should bring you somewhere.

The third problem is the oil attacking the core. I doubt the package material withstands POE oil for a long time. The refrigerants itself are not the problem; the now-banned refrigerant R11 was used as a cleaning agent in PCB manufacturing.

[Liquide]

Yeah it's possible, I believe that Cray computers have some sort of direct die cooling, saw a movie about the latest one a while ago and it looked niiice.

Have also seen people that direct die cooled using water, didn't read to much of the article ( some website posted it ) but I remember they said they did get some nice temps ( nothing amazing though ).

[xDUCK]

Originally posted by Liquide
Yeah it's possible, I believe that Cray computers have some sort of direct die cooling, saw a movie about the latest one a while ago and it looked niiice.

Have also seen people that direct die cooled using water, didn't read to much of the article ( some website posted it ) but I remember they said they did get some nice temps ( nothing amazing though ).

Direct die watercooling is nothing special. I direct die my NB with water. ;) Crays are direct-die water cooled. Also, nothing special. :(

Thanks for the detailed reply, DaBit.

[Holst]

I dont think its practically possible, I dont see how you can get a good enough seal with the CPU in a board.

[berkut]

Im thinking about 1 more thing... the moment when a boiling drop of lets say R22 (-40C) hits the 20C core.... ouch ;]

The cpu should withstand oils etc, the biggest problem is keeping it tight and making sure theres always refrigernt on the core

[NUNEZ1980]

I don't think is possible, and if so the effort, if you have a piece of blocks, you can turn on the cpu and see the temps rise just resing if the cooling is not working, but imaging you letting some freon out and you can burn it fast,

but the main problem is pressure, it can work, and you can't use o-ring, cause they will leak when they freeze;)

[Gary Lloyd]

Direct contact isn't necessarily best. Imagine an air cooled CPU with heatsink and fan. What happens if you remove the heatsink?

[NUNEZ1980]

Originally posted by Gary Lloyd
Direct contact isn't necessarily best. Imagine an air cooled CPU with heatsink and fan. What happens if you remove the heatsink?

your cpu burn:rolleyes: