Subject: Peltier idea, need knowlagable opinions
By: Knut (IP: 67.21.53.*)
Written on: 03-04-2006 20:11

I have given a lot of thought to applying a peltier (Ill call it a TEC) to my system for years. After doing a bit of testing I found that a good TEC drew a lot of power risk (back when a 250W PS was the norm), and the benefits didnít outweigh the potential safety risks. Well these days I have a high end 520w (620w peak) power supply, that supplies 28A@12V.

Anyway, I have never been very comfortable with applying a TEC directly to a CPU. There are too many variables not the least of which is what happens if the fan dies and the TEC heats up to 300F on a CPU that can reach 300F on its own. Even if the CPU shuts down the TEC can get hot enough to melt the motherboard

After revisiting the use of the TEC I thought to ask why not put the hot side outside the case and the cool side in the case, but not in direct contact with any components?

I have a pair of old PII heat sinks. They are roughly 1" tall, 2" wide, and 4" long. I have some aluminum tape and I have taped all but the 2" ends and bottom, effectively making a 4" tunnel for the cool side. I put fairly slow (about 15cfm) and quiet 60mm fan on one end to blow through the cool side. If the fan is too fast then the cold side wont have time it needs to remove the heat from the air as it passes over.

The hot side is outside the case and I have the advantage of being able to use a fairly quiet fan and a fairly large heat sink to cool the hot side (even though I donít have one at the moment).

Now I used materials I had on hand.

I have a VERY old TEC (probably 12-15years old), that doesn't work very well at all that I did some testing with. Indecently, Iíll be able to update this post with some good info soon as I just ordered a new TEC from Melcor. I was able to move about 15CFM at 10F lower than ambient at room temp. This is on a TEC that the cool side only gets to the mid 50s when the hot side heat sink is totally submerged in cold water (read: crappy TEC).

The goal is be to be able to drop the case temps below the normal ambient for the inside of a case. In my instance I run about 36C/104F. Also to be able to do this with limited modifications and minimal fan noise.

Any drop below that would be welcome. I guess the question is how much heat could I dissipate? If I canít dissipate the heat from the entire case (I suspect that that will be too much), perhaps I could build a small tunnel from my 'unit' that would vent the cold air straight to the CPU and GPU heat sinks, increasing its efficiency. Since I have a hole for an 80mm fan on the side of the case directly above the CPU, this would be pretty easy.
Seems that I could effect overall performance best by lowering the operating temp of these two items, thus increasing the potential for overclocking and It should have the added benefit of lowering the case temps as the two main producers of heat would be cooled.

Ideally the hot side heat sink could be pretty big, say 10" square about a 1" deep with a 5" copper plate spreading out the heat. The cooling area would be great enough that I could probably use 1 maybe 2 slow 120mm fans. On the inside a 6-10" long, 2" tall by 3" wide heat sink would be bust for the tunnel. Again, this would be the ideal setup if these parts even exist.

I expect that I should be able to get the cold side HS around the freezing mark in an open room, and 15-25F once in the case.

Course all this is purely speculation and based on a number of factors Iím not fully educated in. I donít know how much heat in watts the entire system is putting out now, nor do I know what the ideal sizes for the cold side heat sink as well as the best speed for the cold side fan to blow over the heat sink.

Here are the benefits of a system like this vs. applying the TEC directly to the component to be cooled.

1 - Worst case if the hot side fan fails the TEC is not in direct contact with anything. If the hot side HS is large enough, passive cooling would be enough to keep case temps from getting high enough to cause any damage.

2 - I wouldnít expect condensation to be a problem as the fan on the cold side will help evaporate any potential water vapor. Even if you had a little, because the unit is mounted on the removable door, water drops would fall harmlessly into the bottom of the case, at least in my case.

The down side

1 Ė TECís always use a lot of juice for the amount of heat they dissipate

2 Ė If I manage to drop case temps by 10-20F its no where near the potential of dropping the CPU itself to 20F or more.

Thoughts, ideas, comments?

By: Knut (IP: 67.21.53.*)
Written on: 03-04-2006 21:19

EDIT first post:

This sentence:
"I expect that I should be able to get the cold side HS around the freezing mark in an open room, and 15-25F once in the case. "

15-25F should be 35-45F, because the cool side will get warmer not colder....
By: Tillmann (IP: 146.140.7.*)
Written on: 26-05-2006 13:35


there is an interesting thread about this subject here:

Back to index | Reply to this message

Forum software based on code ©