Pelt Powered Chiller

[superart]

I got an idea to make a chiller that cools water with an array of low powered pelts. It would be a cube-like design, with the pelts being mounted on the outside of the 4 walls of the cube. Then have big heatsincs attached to that.
Please refer to the picture.

I am currently considdering useing 4 smaller 36W pelts (1 on each wall of the cube). Is this a good amount of cooling power? (too much? too little?)

Also, any issues with condensation? This will be mounted to the top of the case with 9 bolts. I will sandwitch a layer of neoprene in between for insulation, and noise dambening.

Also, anyone have an idea for a cover to this? I cant think of anything in terms of mountng or clasping it shut.


Thanks in advance for any help.

[killernoodle]

Would work, but pelts really arent effecient for chilling. They work best mounted directly to the coldplate and CPU. Plus, you dont have to worry about condensation as much. I too have thought of this idea, but it really isnt as great as you may think.

[superart]

do you have an idea of how the performance would be of such a system? Do you know of someone who has tried this before?

just out of cureosity

[superart]

Also, I forgot to mention, one goal of this is to do away with the radiator and fan. If I put hte TECs on the chip(s) directly, I would still need to have a rad/fan in the system.

right?

[kibbles]

Originally Posted by superart

Also, I forgot to mention, one goal of this is to do away with the radiator and fan. If I put hte TECs on the chip(s) directly, I would still need to have a rad/fan in the system.

right?

I don't remember where, but probably linked to by this forum, someone tried to use pelts to cool water and the efficiency was found to be terrible. Realize that you're adding another water-copper interface for heat to travel across.

[superart]

Originally Posted by kibbles

I don't remember where, but probably linked to by this forum, someone tried to use pelts to cool water and the efficiency was found to be terrible. Realize that you're adding another water-copper interface for heat to travel across.

Well the res would be outside the case, so the heat would travel into the ambient air via big heat-sinks, and the cold part would cool the copper res, which would in turn cool the water within the res.

I realize it isnt the most efficient methood, but am I right in thinking that it would get the water/coolant to AT LEAST the temps that a normal rad/fan setup would achieve?

[killernoodle]

You would still need a radiator and fan, it will just be larger and more bulky, less effecient, and shaped like a resevior.

You are still thinking inside the box my man

Move that peltier directly to the cpu and you will have far better cooling. Anyways, why not just get a large passive radiator?

Oh, and you may end up heating the system up with pelts because the cold side can only be as cold as the hot side is cold.

Plus, tecs use LOTS of power and really the added complexity, the decrease in performance, the problems with condensation, possibly added noise (you will need a very large power supply for multiple tecs, possibly requiring a fan anyways), and not to mention the added money just arent worth it.

A copper box will not absorb the heat from the water very quickly anyways, there just isnt enough surface area.

A radiator is really the easiest way to go.

[superart]

yea, i guess you're right. Too bad, it would've looked pretty cool having that thing bolted to the top of my case. I guess that's just how the laws of physics are....sometimes they work with you, and sometimes they work against you.

Thanks for your help, guys.