TEC to cool the water

Dr. Jones · 06-18-2003, 12:25 AM

I understand how a peltier can be used to cool off the cpu. However i've been doing a lot of reading on several different forums, and have come across a few people who talk about using multiple peltiers (4+) to cool the water. It sounds like they have some sort of block setup (and i'm guessing a stack of power suplies) that puts the water at sub zero (must be using anti-freeze) temps before it even gets to the cpu.

Can anyone shed some light on how this works? It sounds like they have some large resovior that they just submerge some petliers into that are attached to copper blocks... but that doesn't seem like a smart solution. What concepts are behind this sort of cooling? I couldn't find any pictures or drawings of this kind of setup.

mrmoustache · 06-18-2003, 01:12 AM

putting a tec into water would just make it hotter assuming it didnt short out or anything.

im pretty sure they have the water going through block kinda thingies with tecs on them which are water cooled by another loop.

Lmandrake · 06-18-2003, 07:10 AM

Like the previous poster said, its done by putting the pelts on a large block that water runs through and having a separate means of cooling the hot side of the pelts.

I have done it and in my experience it is just too inefficient. It really takes a lot of energy to cool a sufficient quantity of water.

There is a pelt driven device out there with a copper rod that forms an ice ball when cooled by the pelt. If you have sufficient periods when your pc is off and the ball does not erode too much, you could probably make something like that work.

The best multiple pelt idea i have seen was on some web site in italian where 4 or 5 pelts were arranged around the perimeter of a huge copper slug that sat on the cpu. Certainly a ton of effort went into that, and I wonder how well it worked.

Even a single "pelt on the cpu" setup is a fair amount of work.

When you get into these multi-pelt setups, you are spending so much time and money that you might as well go to phase change..

Cathar · 06-18-2003, 08:48 AM

I've done it somewhat successfully using two water loops. I never got the time to see it through to sticking it onto a real system since though.

Just had a single 172W 40x40mm pelt driven at 12V from a regular ATX PSU, sandwiched at high pressure between two micro-channel blocks. One cooling loop cooled the hot side of the peltier, and the other side just had water looping through it.

I esimated that around 100W of cooling power was being applied to the water, but this drops off as the water gets colder, as peltiers lose efficiency the greater the difference between the hot and cold sides.

It managed to freeze 3 litres of water solid in a little under two hours. The temperature rapidly dropped for the first 10C below room temperature, but after that it slows down fairly quickly.

Driving the peltier at 12V was a bit of an issue. Really it should've been driven with at least 16V to get good performance out of the single peltier, which is nominally rated for a 24.6V peak.

Dr. Jones · 06-18-2003, 10:32 AM

Does anyone happen to have a picture of such a setup? I'm thinking that it might be an interesting project to take on. One idea i had was doing it all outside the tower (no room inside) and then running the cold water around the edges of the tower where the intake fans are. Maybe then going through a small radiator or something. The basic concept is that not only would the cold water give the cpu colder temps, but also finding a way to use the 'coldness' of the water to cool the entire inside of the case by cooling the air that is put inside. Not sure if it would work in this way, but i'm sure there is a way to do it. If it was cooling all my components a little, and the CPU a lot, I think it'd be worth the time & effort.

Dr. Jones · 06-18-2003, 10:57 AM

Quote:

Originally posted by Cathar

Just had a single 172W 40x40mm pelt driven at 12V from a regular ATX PSU, sandwiched at high pressure between two micro-channel blocks. One cooling loop cooled the hot side of the peltier, and the other side just had water looping through it.

If I undestand this right, you had a setup like

hot water block (full cooling setup)
pelt
cold water block (just looping water)
CPU

So what was the function of the loop on the cold side? It sounds like you didn't have anything but a pump circulating the water around. Wouldn't you need some kind of radiator to get that to work? It seems like putting water through the cool side would only add heat to the system (from the water that is room temperature), since the water is warmer than the block. Rather than taking the heat from the block and puting it on the water, you'd be taking the heat from the water and putting it on the block. At least I think this is the case...

mrmoustache · 06-18-2003, 01:08 PM

you would only be dumping heat onto the cpu if the water was warmer than the cpu.

a radiator would hurt performance as soon as your water temps drop below ambient.

he said the water was frozen solid and i'd wager frozen solid is below cathars ambient temps.

edit:
also to answer your question the cold side doesn't really have a function when you're only using one pelt, you're probably much better off with the pelt on a coldplate, on the cpu.

but if you have a bunch of pelts i guess the idea is you can get the water temp lower maybe?

it sounds to me like by the time you got it working you'd be using up so much electricity you'd save money by just buying faster cpus as soon as they come out.

redleader · 06-18-2003, 02:43 PM

Pelts are really ineffcient. A 1/4 HP compressor setup might get you well below zero, but you'd need several times as much pelt to reach that. Probably more then is worthwhile.

Razor6 · 06-19-2003, 03:39 AM

Quote:

Originally posted by Dr. Jones
If I undestand this right, you had a setup like

hot water block (full cooling setup)
pelt
cold water block (just looping water)
CPU

So what was the function of the loop on the cold side? It sounds like you didn't have anything but a pump circulating the water around. Wouldn't you need some kind of radiator to get that to work? It seems like putting water through the cool side would only add heat to the system (from the water that is room temperature), since the water is warmer than the block. Rather than taking the heat from the block and puting it on the water, you'd be taking the heat from the water and putting it on the block. At least I think this is the case...

I think he meant something like this. Pardon my leet paint skills.

mrmoustache · 06-19-2003, 04:30 AM

it's kinda cool how all you do is add a second pump, it's a shame no one really wants to run a bunch of high wattage tecs.

Cathar · 06-19-2003, 08:20 AM

Razor6 has it exactly right - that's how the loop looks.

Quote:

it's kinda cool how all you do is add a second pump, it's a shame no one really wants to run a bunch of high wattage tecs.

One extra pump, and two extra blocks.

It's actually more energy efficient to run multiple peltiers at say 2/3'rds peak voltage, than to run one at full voltage.

3 x peltiers @ 2/3'rds peak voltage >> 2 x same peltiers at peak voltage

They'd use less power, generate less heat, and give a greater cooling effect.

06-18-2003, 12:25 AM	#1
Dr. Jones Cooling Neophyte Join Date: Jun 2003 Location: Denver, CO Posts: 7	TEC to cool the water I understand how a peltier can be used to cool off the cpu. However i've been doing a lot of reading on several different forums, and have come across a few people who talk about using multiple peltiers (4+) to cool the water. It sounds like they have some sort of block setup (and i'm guessing a stack of power suplies) that puts the water at sub zero (must be using anti-freeze) temps before it even gets to the cpu. Can anyone shed some light on how this works? It sounds like they have some large resovior that they just submerge some petliers into that are attached to copper blocks... but that doesn't seem like a smart solution. What concepts are behind this sort of cooling? I couldn't find any pictures or drawings of this kind of setup.

06-18-2003, 07:10 AM	#3
Lmandrake Cooling Neophyte Join Date: Nov 2001 Location: D.C. Posts: 26	Like the previous poster said, its done by putting the pelts on a large block that water runs through and having a separate means of cooling the hot side of the pelts. I have done it and in my experience it is just too inefficient. It really takes a lot of energy to cool a sufficient quantity of water. There is a pelt driven device out there with a copper rod that forms an ice ball when cooled by the pelt. If you have sufficient periods when your pc is off and the ball does not erode too much, you could probably make something like that work. The best multiple pelt idea i have seen was on some web site in italian where 4 or 5 pelts were arranged around the perimeter of a huge copper slug that sat on the cpu. Certainly a ton of effort went into that, and I wonder how well it worked. Even a single "pelt on the cpu" setup is a fair amount of work. When you get into these multi-pelt setups, you are spending so much time and money that you might as well go to phase change.. __________________ The technology is within the reach of even the smallest nuclear power.

06-18-2003, 01:12 AM	#2
mrmoustache Cooling Neophyte Join Date: Oct 2002 Location: columbus, ohio Posts: 44	putting a tec into water would just make it hotter assuming it didnt short out or anything. im pretty sure they have the water going through block kinda thingies with tecs on them which are water cooled by another loop.

06-18-2003, 08:48 AM	#4
Cathar Thermophile Join Date: Sep 2002 Location: Melbourne, Australia Posts: 2,538	I've done it somewhat successfully using two water loops. I never got the time to see it through to sticking it onto a real system since though. Just had a single 172W 40x40mm pelt driven at 12V from a regular ATX PSU, sandwiched at high pressure between two micro-channel blocks. One cooling loop cooled the hot side of the peltier, and the other side just had water looping through it. I esimated that around 100W of cooling power was being applied to the water, but this drops off as the water gets colder, as peltiers lose efficiency the greater the difference between the hot and cold sides. It managed to freeze 3 litres of water solid in a little under two hours. The temperature rapidly dropped for the first 10C below room temperature, but after that it slows down fairly quickly. Driving the peltier at 12V was a bit of an issue. Really it should've been driven with at least 16V to get good performance out of the single peltier, which is nominally rated for a 24.6V peak.

06-18-2003, 10:32 AM	#5
Dr. Jones Cooling Neophyte Join Date: Jun 2003 Location: Denver, CO Posts: 7	Does anyone happen to have a picture of such a setup? I'm thinking that it might be an interesting project to take on. One idea i had was doing it all outside the tower (no room inside) and then running the cold water around the edges of the tower where the intake fans are. Maybe then going through a small radiator or something. The basic concept is that not only would the cold water give the cpu colder temps, but also finding a way to use the 'coldness' of the water to cool the entire inside of the case by cooling the air that is put inside. Not sure if it would work in this way, but i'm sure there is a way to do it. If it was cooling all my components a little, and the CPU a lot, I think it'd be worth the time & effort.

06-18-2003, 01:08 PM	#7
mrmoustache Cooling Neophyte Join Date: Oct 2002 Location: columbus, ohio Posts: 44	you would only be dumping heat onto the cpu if the water was warmer than the cpu. a radiator would hurt performance as soon as your water temps drop below ambient. he said the water was frozen solid and i'd wager frozen solid is below cathars ambient temps. edit: also to answer your question the cold side doesn't really have a function when you're only using one pelt, you're probably much better off with the pelt on a coldplate, on the cpu. but if you have a bunch of pelts i guess the idea is you can get the water temp lower maybe? it sounds to me like by the time you got it working you'd be using up so much electricity you'd save money by just buying faster cpus as soon as they come out.

06-18-2003, 02:43 PM	#8
redleader Thermophile Join Date: Jun 2001 Location: The deserts of Tucson, Az Posts: 1,264	Pelts are really ineffcient. A 1/4 HP compressor setup might get you well below zero, but you'd need several times as much pelt to reach that. Probably more then is worthwhile.

06-19-2003, 04:30 AM	#10
mrmoustache Cooling Neophyte Join Date: Oct 2002 Location: columbus, ohio Posts: 44	it's kinda cool how all you do is add a second pump, it's a shame no one really wants to run a bunch of high wattage tecs.

Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)