Sort of a waterchiller.....
 

I was thinking about making a waterchiller, but putting a twist on it. Basically, I would have a 80w peltier cooling the water as well as a standard radiator. The 80w peltier would be cooled by a heatsink. I am doing this because I want my temps to be as close to ambient as possible, but without creating condensation. Second, when I turn the fans down on the radiator, I want it so that the water will still be cold. I will have a concept art very soon.

You arent going to get where you want to go with a single 80W peltier.

Yeah, the 80W Peltier just doesn't cut it. Basically, you will be using electricity with next to nothing for noticeable gains.

I was playing around with this idea earlier, though with a larger pelt. It's a good idea, or at least I thought so. The pelt will only be drawing a part of the heat from the water loop, so it can draw the temp of the loop down a good couple degrees.

I thought up of a couple problems with it, but can't remember what they were anymore.

It at least seems feasible.

Alchemy

how are u going to cool the water? sticking the pelt on a block with a fan on the backside? that wont work cuz it will create ice crystals in the block and will put condensation in the tubing around the block....u need to mount something waterproofed into the water resivior or install it to the front of your radiator with a fan attached.......

heres a tool i mentioned earlier in my post....its called the iceprobe it is actually a 50 watt pelt with an extension on the end that gets inserted into your resivior...it forms an iceball around it and drops temps considerably......a guy over at hardocp tested it w/o a radiator and used his homemade compact resivior giving him a drop to down in his water temp to 15C w/o any condensation on the tubing!!http://hardocp.com/image.html?image=...dfMV8xX2wuanBn

That seems pretty unlikely, considering the fact that the processor would put out more that 50W of heat.

With a radiator you have only a few degrees over ambient on a non peltier cpu block. That would puch it under ambient. If the cpu had a peltier, the added 80w would definately help. I thought about something like that, but cant afford it. Make a metal res and put a peltier aircooled on it. That would most likely lower the water temp quite a bit. Helping the water block with colder water, less of a strain on the cpu peltier. The Ice Probe just seems like alot of money. If you want to run multiple pelters and already have a second psu you might as well use another seperate pelt on the res. I imagine you need a big rad to cool a cpu peltier.

the procooling article seems to think different.
: : link : :

seemed like quite a flop.

Right, you are cooling water that is only sitting in the water block for a short amount of time. It wont be able to cool as effectively. If its in the closed loop and has water in a res then its just cooling water that cant go any where. If he had a metal res that had a gallon of water or whatever and it wasnt going through a pump with no time to cool down, then it would do something. The reason why we pump water through a block so fast is to make sure to get alot of water to cool a processor. But if you are trying to run the entire sink through that, then it wont cool down. It will just keep cycling to fast to do anything effectively.

This is completely wrong. The same principles apply when heating water with a waterblock, or chilling water with a TEC.

If the TEC is, "just cooling water that cant go anywhere", then all heat removed by the TEC will have to conduct through that 'stagnant' water. Water is not a particularly good conductor of heat.

Ok, I am guessing that it must be extremely diffucult to push water to below ambient. Plus having it run back through the radiator might warm the water back up if not cool it. Anybody know how warm there water gets from peltier cooling a cpu? Say 226w peltier. This would help.

4.186j/gK assuming ambient is between 14.5 and 15.5C ;)

Obviously you won't be able to use a radiator with a water chiller.

It adds 226w (assuming you actually ran it at 226w) + the heat outputed from the CPU.

I believe that this is incorrect. The 226w is the max heat it can absorb from a heat source. The actual heat generated is usually much much higher.

It adds the amount of power it consumes, eg. VoltagexCurrent.

So, if it is a 226W TEC that has Vmax = 15.2V and Imax = 24A, and you are running it at 12V:

P = 12V * (12V/15.2V) * 24A = 227W

Add the CPU's heat generation to that number.

It's not quite that simple.

For a given voltage input, the current draw will depend on the temperature difference between the hotside and the coldside. (Seebeck effect) The greater the temp difference, the lower the current draw.

Get the Kroyotherm calculator. It helps the understanding of how tec work. Unfortunely there is no 226w tec available for the calc's

Damn those TECs.... why can they not behave like resistors ;).

I mean a TEC will give off the power is consumes + CPU power. I'm assung it uses 226w, of course it will probably use something else.