Smoking crack, or good idea?

[DorianGrey]

I've looked at a lot of case cooling solutions. All of them seem to have drawbacks. Water cooling, while better than air, is generally average because the water is run through a radiator that is air cooled, usually by air that has already passed through the case.

Peltier junctions will cool much better, but thier problem is they are TOO good. The low temperature has a tendency to cause condensation which can cause corrosion, and potentially, shorts. The methods of trying to alleviate this leave you with an almost unremoveable cpu.

Enter my idea. Take a normal water cooled system. Resevoir, pump, cooling blocks. Normally you would have water in the resevior that feeds your pump, goes into the cpu/gpu/memory cooling blocks, into a radiator, back into the resevoir. For my change, lets take the hose out of the pump, out of the case, and feed it into a cpu/gpu cooling block, back into the case, then onto the rest of the system (cpu, gpu and or memory). The first cpu block is mated onto a peltier air cooling system. This would cause the first cpu block to be very cold, and chill the coolant. This would seem to me that it would provide a much more efficient water cooling system, yet not create the condensation problems that an extreme solution like peltiers give you.

I haven't seen this solution anywhere, nor have I seen it discussed on any boards. From this, I've either stumbled on a Eureka! like solution, or I REALLY need to cut back on the crack. Think about it, and advise me which one it is

 

[Mallow]

probably not a market for such a set up, but it sounds like a good idea never the less for those hardcore OC'ers. I'm happy getting a moderate OC w/ my air cooled tower

 

[GtPrOjEcTX]

eh, fans are good enough.

 

[mastay]

Unless you are trying to hit 4ghz, I fail to see the actually need for this setup. FYI, when you setup a peltier correctly, there should not be any condensation...

 

[SuperPickle]

I think I understand your idea, but you still have to put the heat somewhere. A modest-sized TEC generates about the same heat as a modern CPU and larger ones much more. To keep the cold side of the pelt cold, you have to take the heat away from the hot side with which you purpose an air-cooled pelt. I don't think a normal HS would do well cooling the TEC while taking moderate calories from the coolant as the heat from the coolant plus the heat from the pelt would have to be drawn through the air HS and dispersed to the room. Now, if you were to use a monster pelt on your external unit and then (probably) watercool that, you may see some drop in computer coolant temperature which is the ultimate goal. This would probably work although it's not something I'd call very efficient--power usage, or cost-wise.

kudos for abstract thinking--it's those types of ideas that can shake up 'normal' ways of doing things and may soon become the 'normal' way.

 

[EdipisReks]

there are already chilled water systems which, if i understand wht you typed, basically do the same thing that you are suggesting.

 

[cheapgoose]

crack good.

 

[LazyBastard]

Let me get this straight. You're suggesting that one replace their radiator with an external peltier-cooled waterblock?

What about the potential condensation problem? You may still have to deal with condensation if this setup happens to cool the water, and waterblocks, below ambient. Condensation occurs when you cool something below the dew point temperature, which, based on the relative humidity, can be near, if not the same as, the ambient room temperature. This may not be a problem in Arizona, but if you lived in New Orleans, for instance, which frequently gets to 95% to 100% relative humidity in the summer, you'd have a problem.

 

[PhoenixOfWater]

Originally posted by: cheapgoose
crack good.

Crack bad

(from what they tell me)

 

[Derick]

yeah, If the water is too cold when it hits the CPU there will still be some condensation.

there are already chilled water systems which, if i understand wht you typed, basically do the same thing that you are suggesting.

yep, either by keeping the resivior and radiator in a small fridge or using phase-change (refridgeration compressor) with the evaporator submerged in the resivior.

Both of these are much more efficient at cooling the water than your proposal. I don't know if the water would be in the block long enough to truely get cold. Then again it might cool it down some but not enough to cause condensation. Well, i guess the only way to know is if you try it.