Water Chiller!

thezfunk · 03-06-2004, 06:48 PM

Well,

I got one of those TEC 12v car cooler things. Took it all apart and it had a 40w pelt inside. All the other ones I have seen have the exact same specs so I am guessing they all pretty much come with 40 watters. My plan is to replace that 40 with a 172 watt pelt and fill the aluminum tub its attacted to with water and use it as a water chiller for my CPU. I am almost 100 percent sure that it will work its just a matter of how well. With water inside the water would get down to about 42F and thats with the stock 40 watt pelt (the thermometer I had only did F). Just by redoing all the thermal connections with AC5 cut the water temp down to 38F. I would do a bigger pelt but it is only 40mm x 40mm and the highest wattage pelt that size I could find was 172 watts.

I tried talking about this at the Hardforums but the children over there had nothing but negative to say. My roomate also has very negative things to say like its a waste of time/money and I am like what is it to you? Its my time and money. I think he is afraid it will work.

The only other problem I forsee is that the tub is aluminum and the waterblock is going to be copper. Probably a Maze 4. I am afraid of galvanic corrosion. Because of the interaction with the copper and aluminum. I have talked to some professional engineers who work everyday with metal finishing that say aslong as they don't touch and I use plastic tubing with distilled water to cut down on the available ions in the water I should be fine. Putting alittle antifreeze in will help too. One of my other options would be to get some silver nitrate and run it through the copper waterblock which would give me a silver coated inside of a waterblock and copper nitrate. But I was told thats alittle extreme and I shouldn't have to do that.

Thoughts and opinions? Suggestions for improvements?

Oh! Just thought of another question. Inside this thing it has a aluminum coldplate. Where can I get a copper one made exactly the same way?

feathers · 03-08-2004, 03:58 AM

Hi,

I have worked on several prototype water-chillers. It's definately a worthwhile project! The idea of warm water running through a system sucks big-time! I use a peltier to chill the CPU but still have warm water flowing over the peltier to keep it cool. It would be much better to have chilled water flowing through the cpu waterblock because this would increase the performance of the CPU peltier!

The only negative thing I will say is this: Chilling the water is a waste of time if you don't have a peltier over the CPU as well! A peltier produces very low-temps but only across a small surface area (50mm max).. The most effective place for a peltier will always be directly over the CPU first and foremost. If you can then chill the water as well then you will see a dramatic increase in the performance of the CPU peltier!

Regarding corrosion and different metals: Yes, it can be a problem and you will always need some corrosion inhibitor. I've used anti-freeze but I don't like it. It's an extremely nasty poison and you don't want it evaporating into your computer room! I've also used a de-toxified glycol coolant which the manufacturer claimed was extremely safe! I became ill within 24 hours of using the stuff!

Antifreeze reduces the ability of the water to remove heat although it does help to prevent corrosion. Car windshield washer would be better in terms of cooling performance. It does contain antifreeze (which is why it comes with a warning of serious consequences for exposure) but it also contains a surfactant which reduces the water tension. When you reduce the surface tension of water you will see your CPU temp drop!

I would recommend redline water-wetter. I'm using it in my system now. It's a pink liquid with corrosion inhibitors. Redline water wetter has made my noisy water-pump silent! (I have an old Eheim 1048 with a damaged shaft). The water-wetter is less toxic than antifreeze and it's viscocity is much lower than glycol/antifreeze.

Going back to your water chiller:

If you can chill a thermally insulated container with water and have the water freeze solid.. You could then have the coolant running through this block of ice and this would give you a much better chilling performance. It doesn't take much effort for the CPU to raise the temp of chilled water.. It does require an enormous effort for the CPU to melt a solid block of ice!

Finally.. There is rule you might want to remember with regard to chillers..

The amount of chill you get will be proportional the power of the peltier that's chilling the water versus the waste heat in watts from the CPU.

In other words: If you CPU is generating 90 watts of heat and you use an 80 Watt peltier to chill the water: Don't expect any chilling performance! You really need to double the ratio at least - so the peltier would be 180 Watts for a CPU generating 90W.

If you need an especially high-powered peltier (280Watts) then I can acquire it from our German supplier. If you'd like some pics of the chillers I worked on then email me.

:-)

|kbn| · 03-09-2004, 09:04 PM

Why would you have to use the mini fridge? There are otehr ways to do it...
I also have a minifridge which had a 28w TEC. If you were to do this I think using more than one TEC would be best, 4 or 5x80w TECs would be much easyer to cool than one large 300w TEC. Remember you have to cool the hot side too, how would you do that?

For the heat exchanger haveing an ice block as said above might be a good idea, but with one change to make it better - get some 1/2" copper pipe, and loads of T connectors. Then cut the pipe and have 10 1/2 pipes running in parrelell - This would give you a nice large surface area. Using small lenghs would probably fit in your minifridge. you oculd use 2 or 3 of these in parrelell aswell if you have space in there! Using it in parrellell wil give you less flow restriction. A pipe used as a heat exchanger between your loop and the ice would seperate the chiller from the main loop, giving you no corrosion if the rest of your system is only copper/plastic.
I aviod aluminim compleatly, although I am using some silver, and also some solder (tin + lead) I dont get any niticable corrosion. I use pure de-ionised water. Im not and have never made a chiller. I might soon, after Ive tryed my heatercore outside in the cold

thezfunk · 03-10-2004, 02:24 AM

That was the other idea!

You reminded me. Yeah I think what I forgot to add was just like you were saying. Have the cooling loop run closed through the chilled water in the mini fridge tub. Run some copper tubing all through it. Fill it with water let the pelt cool the water in the tub which cools the copper tubing and the closed loop to the CPU. The only thing I was concerned about though was that having that copper loop would waste cooling abiltiy because you are adding another material you have to transfer the heat through...the copper tubing. Instead of just running the cooled water to the CPU block right out of the tub.

You guys think it would be that big of a difference?

My hope is to make something in the end that will cool better than just a heatercore or rad.

feathers · 03-10-2004, 03:09 AM

"Im not and have never made a chiller. I might soon, after Ive tryed my heatercore outside in the cold " - I have this recurring idea to buy a 2x120 heatercore and freeze it in a liquid bath. The bath would be water-filled and then frozen by TECs. The computer coolant obviously then flows through the frozen radiator block.

thezfunk · 03-10-2004, 01:50 PM

You guys seem to be 2 steps ahead of me :-). In the aluminum tub I could just put a heatcore from a car like they normally do but instead of having fans blowing air through the core have the core sitting in the TEC chilled water. Then I have a closed loop system and solve alot of my problems.

Is there somewhere online that sells good heater cores I can check out.

|kbn| · 03-10-2004, 11:44 PM

what you could do, is get some liquid nitrogen, build a water chiller that works like the way ive described above, exept instead of using tecs, build a system to drip 1ml or less of LN into the water every so often. using a lagre quantity of water would spread out the effect - this is so that it doesnt instantly FREEZE your main loop, instead the water will get very cold, but not enough to damage anything. you will need lots of antifreeze in both loops this way.
I admit it may not be good enough for long term cooling, but it would make a nice machine for doing a one time bench, and should last longer than the limited normal ln benhces, where the ln freezes everything and damages it with moisture. imagine using sub -30 degree water/antifreeze/other mix, sounds good doesnt it

GLHF finding a suitable pump though :P

redleader · 03-11-2004, 01:11 AM

Chilling water with TECs is a waste of time IMO. Way too little cooling power for way too much work. It'd be easier and probably no more expensive to do it right and get a compressor.

feathers · 03-11-2004, 06:14 PM

I suspect the redleader is right. TECs are great for chilling a small area but not so good for chilling a large volume (as in the case of water). Having said that: there are some companies using TECs in air-conditioning systems now and I know of at least one company in china who specialise in TEC based waterchiller blocks. By careful design of the 'cold-sink' it's possible to maximise the chilling effect of the peltier.

Our first chiller design worked so well that we had to use anti-freeze to preven the coolant from freezing solid. It was based on 8mm tubing system though. Slow moving water works better for a chiller. The rules change when you're dealing with a water-chiller.

We will probably upload some pictures of our Mk-1 and 2 water chillers. I would certainly like to work on a new design based on the information gathered from those two projects. There are some tricks you can use to get effective chilling from even a single 200 Watt peltier.

Ultimately though - phase-change inline chillers are more effective at this point in time. Perhaps when we have a new generation of TECs then this will change?

thezfunk · 03-12-2004, 12:36 AM

Hey thanks guys,

You guys have been very informative and thought provoking. I am still going to give it a shot though :-). You will hear back from me with pics and review and things and I probably will ask more questions before I am done. The hardforum can kiss off as far as I am concerned. I refuse to get bashed by 16 year kids who haven't had a thermo class let alone chem. I do see the cards are stacked against me but thats what makes it fun and challenging.

feathers · 03-13-2004, 01:17 PM

Thinking of holding a challenge over at my website for thermoelectric waterchiller. We can show pictures of the various designs people send in as well as show performance data.

Interested?

www.icetec-uk.com

03-06-2004, 06:48 PM	#1
thezfunk Cooling Neophyte Join Date: Mar 2004 Location: Wouldn't you like to know Posts: 33	Water Chiller! Well, I got one of those TEC 12v car cooler things. Took it all apart and it had a 40w pelt inside. All the other ones I have seen have the exact same specs so I am guessing they all pretty much come with 40 watters. My plan is to replace that 40 with a 172 watt pelt and fill the aluminum tub its attacted to with water and use it as a water chiller for my CPU. I am almost 100 percent sure that it will work its just a matter of how well. With water inside the water would get down to about 42F and thats with the stock 40 watt pelt (the thermometer I had only did F). Just by redoing all the thermal connections with AC5 cut the water temp down to 38F. I would do a bigger pelt but it is only 40mm x 40mm and the highest wattage pelt that size I could find was 172 watts. I tried talking about this at the Hardforums but the children over there had nothing but negative to say. My roomate also has very negative things to say like its a waste of time/money and I am like what is it to you? Its my time and money. I think he is afraid it will work. The only other problem I forsee is that the tub is aluminum and the waterblock is going to be copper. Probably a Maze 4. I am afraid of galvanic corrosion. Because of the interaction with the copper and aluminum. I have talked to some professional engineers who work everyday with metal finishing that say aslong as they don't touch and I use plastic tubing with distilled water to cut down on the available ions in the water I should be fine. Putting alittle antifreeze in will help too. One of my other options would be to get some silver nitrate and run it through the copper waterblock which would give me a silver coated inside of a waterblock and copper nitrate. But I was told thats alittle extreme and I shouldn't have to do that. Thoughts and opinions? Suggestions for improvements? Oh! Just thought of another question. Inside this thing it has a aluminum coldplate. Where can I get a copper one made exactly the same way? Last edited by thezfunk; 03-06-2004 at 07:13 PM.

03-10-2004, 02:24 AM	#4
thezfunk Cooling Neophyte Join Date: Mar 2004 Location: Wouldn't you like to know Posts: 33	That was the other idea! You reminded me. Yeah I think what I forgot to add was just like you were saying. Have the cooling loop run closed through the chilled water in the mini fridge tub. Run some copper tubing all through it. Fill it with water let the pelt cool the water in the tub which cools the copper tubing and the closed loop to the CPU. The only thing I was concerned about though was that having that copper loop would waste cooling abiltiy because you are adding another material you have to transfer the heat through...the copper tubing. Instead of just running the cooled water to the CPU block right out of the tub. You guys think it would be that big of a difference? My hope is to make something in the end that will cool better than just a heatercore or rad. Last edited by thezfunk; 03-10-2004 at 02:31 AM.

03-08-2004, 03:58 AM	#2
feathers Cooling Neophyte Join Date: Mar 2004 Location: UK Posts: 81	Hi, I have worked on several prototype water-chillers. It's definately a worthwhile project! The idea of warm water running through a system sucks big-time! I use a peltier to chill the CPU but still have warm water flowing over the peltier to keep it cool. It would be much better to have chilled water flowing through the cpu waterblock because this would increase the performance of the CPU peltier! The only negative thing I will say is this: Chilling the water is a waste of time if you don't have a peltier over the CPU as well! A peltier produces very low-temps but only across a small surface area (50mm max).. The most effective place for a peltier will always be directly over the CPU first and foremost. If you can then chill the water as well then you will see a dramatic increase in the performance of the CPU peltier! Regarding corrosion and different metals: Yes, it can be a problem and you will always need some corrosion inhibitor. I've used anti-freeze but I don't like it. It's an extremely nasty poison and you don't want it evaporating into your computer room! I've also used a de-toxified glycol coolant which the manufacturer claimed was extremely safe! I became ill within 24 hours of using the stuff! Antifreeze reduces the ability of the water to remove heat although it does help to prevent corrosion. Car windshield washer would be better in terms of cooling performance. It does contain antifreeze (which is why it comes with a warning of serious consequences for exposure) but it also contains a surfactant which reduces the water tension. When you reduce the surface tension of water you will see your CPU temp drop! I would recommend redline water-wetter. I'm using it in my system now. It's a pink liquid with corrosion inhibitors. Redline water wetter has made my noisy water-pump silent! (I have an old Eheim 1048 with a damaged shaft). The water-wetter is less toxic than antifreeze and it's viscocity is much lower than glycol/antifreeze. Going back to your water chiller: If you can chill a thermally insulated container with water and have the water freeze solid.. You could then have the coolant running through this block of ice and this would give you a much better chilling performance. It doesn't take much effort for the CPU to raise the temp of chilled water.. It does require an enormous effort for the CPU to melt a solid block of ice! Finally.. There is rule you might want to remember with regard to chillers.. The amount of chill you get will be proportional the power of the peltier that's chilling the water versus the waste heat in watts from the CPU. In other words: If you CPU is generating 90 watts of heat and you use an 80 Watt peltier to chill the water: Don't expect any chilling performance! You really need to double the ratio at least - so the peltier would be 180 Watts for a CPU generating 90W. If you need an especially high-powered peltier (280Watts) then I can acquire it from our German supplier. If you'd like some pics of the chillers I worked on then email me. :-)

03-09-2004, 09:04 PM	#3
\|kbn\| Cooling Savant Join Date: Feb 2004 Location: W. Sussex, UK Posts: 329	Why would you have to use the mini fridge? There are otehr ways to do it... I also have a minifridge which had a 28w TEC. If you were to do this I think using more than one TEC would be best, 4 or 5x80w TECs would be much easyer to cool than one large 300w TEC. Remember you have to cool the hot side too, how would you do that? For the heat exchanger haveing an ice block as said above might be a good idea, but with one change to make it better - get some 1/2" copper pipe, and loads of T connectors. Then cut the pipe and have 10 1/2 pipes running in parrelell - This would give you a nice large surface area. Using small lenghs would probably fit in your minifridge. you oculd use 2 or 3 of these in parrelell aswell if you have space in there! Using it in parrellell wil give you less flow restriction. A pipe used as a heat exchanger between your loop and the ice would seperate the chiller from the main loop, giving you no corrosion if the rest of your system is only copper/plastic. I aviod aluminim compleatly, although I am using some silver, and also some solder (tin + lead) I dont get any niticable corrosion. I use pure de-ionised water. Im not and have never made a chiller. I might soon, after Ive tryed my heatercore outside in the cold

03-10-2004, 03:09 AM	#5
feathers Cooling Neophyte Join Date: Mar 2004 Location: UK Posts: 81	"Im not and have never made a chiller. I might soon, after Ive tryed my heatercore outside in the cold " - I have this recurring idea to buy a 2x120 heatercore and freeze it in a liquid bath. The bath would be water-filled and then frozen by TECs. The computer coolant obviously then flows through the frozen radiator block.

03-10-2004, 01:50 PM	#6
thezfunk Cooling Neophyte Join Date: Mar 2004 Location: Wouldn't you like to know Posts: 33	You guys seem to be 2 steps ahead of me :-). In the aluminum tub I could just put a heatcore from a car like they normally do but instead of having fans blowing air through the core have the core sitting in the TEC chilled water. Then I have a closed loop system and solve alot of my problems. Is there somewhere online that sells good heater cores I can check out.

03-10-2004, 11:44 PM	#7
\|kbn\| Cooling Savant Join Date: Feb 2004 Location: W. Sussex, UK Posts: 329	what you could do, is get some liquid nitrogen, build a water chiller that works like the way ive described above, exept instead of using tecs, build a system to drip 1ml or less of LN into the water every so often. using a lagre quantity of water would spread out the effect - this is so that it doesnt instantly FREEZE your main loop, instead the water will get very cold, but not enough to damage anything. you will need lots of antifreeze in both loops this way. I admit it may not be good enough for long term cooling, but it would make a nice machine for doing a one time bench, and should last longer than the limited normal ln benhces, where the ln freezes everything and damages it with moisture. imagine using sub -30 degree water/antifreeze/other mix, sounds good doesnt it GLHF finding a suitable pump though :P

03-11-2004, 01:11 AM	#8
redleader Thermophile Join Date: Jun 2001 Location: The deserts of Tucson, Az Posts: 1,264	Chilling water with TECs is a waste of time IMO. Way too little cooling power for way too much work. It'd be easier and probably no more expensive to do it right and get a compressor.

03-11-2004, 06:14 PM	#9
feathers Cooling Neophyte Join Date: Mar 2004 Location: UK Posts: 81	I suspect the redleader is right. TECs are great for chilling a small area but not so good for chilling a large volume (as in the case of water). Having said that: there are some companies using TECs in air-conditioning systems now and I know of at least one company in china who specialise in TEC based waterchiller blocks. By careful design of the 'cold-sink' it's possible to maximise the chilling effect of the peltier. Our first chiller design worked so well that we had to use anti-freeze to preven the coolant from freezing solid. It was based on 8mm tubing system though. Slow moving water works better for a chiller. The rules change when you're dealing with a water-chiller. We will probably upload some pictures of our Mk-1 and 2 water chillers. I would certainly like to work on a new design based on the information gathered from those two projects. There are some tricks you can use to get effective chilling from even a single 200 Watt peltier. Ultimately though - phase-change inline chillers are more effective at this point in time. Perhaps when we have a new generation of TECs then this will change?

03-12-2004, 12:36 AM	#10
thezfunk Cooling Neophyte Join Date: Mar 2004 Location: Wouldn't you like to know Posts: 33	Hey thanks guys, You guys have been very informative and thought provoking. I am still going to give it a shot though :-). You will hear back from me with pics and review and things and I probably will ask more questions before I am done. The hardforum can kiss off as far as I am concerned. I refuse to get bashed by 16 year kids who haven't had a thermo class let alone chem. I do see the cards are stacked against me but thats what makes it fun and challenging.

03-13-2004, 01:17 PM	#11
feathers Cooling Neophyte Join Date: Mar 2004 Location: UK Posts: 81	Thinking of holding a challenge over at my website for thermoelectric waterchiller. We can show pictures of the various designs people send in as well as show performance data. Interested? www.icetec-uk.com

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