chilled water or TEC?

[Since87]

I don't want to cover my motherboard with goop and void the warranty.

My waterchiller will never cool the water below the dewpoint. (I've got an electronic hygrometer to ensure this.) Most of the time the dewpoint in my house is going to be less than 10C. I can easily maintain a supply of water at 10C with a couple 172 watt TEC's with fairly low power consumption. I will have a microcontroller controlling the voltage to the TEC's, so I can continuously vary the TEC voltage between 0V and 75% Vmax.

With the use of an insulated reservoir as Doc described, I can "store up cold" with low power consumption while I'm at work. If I game for a few hours in the evening, the TEC's only need to provide enough cooling to maintain the water just above the dewpoint.

 

[doc]

Since87 sounds like your good to go. Since your dT is less you will be operating on a better part of the COP curve. On the chiller above the dewpoint was not a problem, but we needed sub-zero temps for hours at a time. I hear ya about the goop on the MB. I like the high tech microcontroller keeping the coolant temp just right.

 

[aenigma]

paxmax:
When I was referring to using 4 pelts, blocks/heatsinks I was referring to doc's setup, and the guy with 4 226 watt pelts.I never said that is required.You could get away with 1 TEC, but your temps will be pretty bad.

I see what you mean about have a thermostat.Just to keep it at whatever temp it equalizes at under load, even when its idle.

doc:
wow you like typing

Ok you were cooling 30 watts, back in the day of low wattage processors pelts would work great.But now cpu's put out up to 150 watts.I see where your going with the whole buffer, I understood that part when you first posted.But that buffer gets ruled out if your system is on 24/7.

Since87:
Yeah if 10c is all you want, then a pelt will work for that.

 

[Paxmax]

paxmax:
When I was referring to using 4 pelts, blocks/heatsinks I was referring to doc's setup, and the guy with 4 226 watt pelts.I never said that is required.You could get away with 1 TEC, but your temps will be pretty bad.

Are you sure you know how many blocks and heatsinks DOC used?
Actually you DID say "Pelts will also require 4 individual heatsinks..." wether it was a "mistake" or "typo error" or "not writing coherently" I don't know, but I do know that you wrote it, but hey, keep claiming "I didn't.." if you want, I'm giving up on the issue right about now anyway.

Cheers aenigma and good luck to you too.
I'll try to remember mentioning my results to you when my rig gets done.

 

[aenigma]

Are you sure you know how many blocks and heatsinks DOC used?
Actually you DID say "Pelts will also require 4 individual heatsinks..." wether it was a "mistake" or "typo error" or "not writing coherently" I don't know, but I do know that you wrote it, but hey, keep claiming "I didn't.." if you want, I'm giving up on the issue right about now anyway.

Cheers aenigma and good luck to you too.
I'll try to remember mentioning my results to you when my rig gets done.

Um yeah it was obviously a typo if you can't see that, and kind of obvious I was refering to the guy actually using pelts to chill his water...

 

[doc]

Aenigma you have a very valid point, the buffer idea does not work if your computer runs 24/7. But for those who do not run 24/7 and still like to OC or run a little cooler, the system I used works well and is low cost to run.

I guess I need to clear up something else about my water chiller.
I used kind of an unusual way of cooling the res. My TECs has heat sinks on both sides, but one sink in the container and one sink in air with a fan on it. The container is polyproplene and the sinks are larger than the TECs. A square hole was cut in the container just a bit bigger that the TEC. The sink inside was drilled and tapped. A bead of silicone bathtub caulk is used seal each one. The ouside sink was just drilled out in the same places. Small diameter springs between the screw head and sink keeps constant and uniform pressure on the TEC as well as rubber seal ( with small holes for the wires) between the outer sink and the container. Which pulls the inside sink snug to the container wall. This is somewhat a passive system, but the warm return coolant is at the top of the container as are the TECs. When the system is not in use, but cooling, the cooler more dense coolant gently drifts to bottom of the container. The output line is taken from the bottom.

 

[aenigma]

Yeah if someone doesn't run all the time it would, and to increase that buffer even more you can use a big huge ice chest

I bet that 5 gallons takes awhile to get from ambient to -5c!
These new cool chips, if they are as good as they say, will be great for chilling water or just using directly on the cpu.They say they are more effecient that refrigeration.

 

[doc]

Ya, you right about that! It took 3 days to get to -5C, little bit more than a week end. But well worth it.

 

[LittlePiggie]

Uhm no, TEC waterchillers don't work for ****.

I have done some "research" and came to the conclusion that the power of the pelts used for chilling the water is not that big an issue. The reason for this is simple: a 172 watt pelt can remove X amount of heat from an area, but 4 40 watt pelts can remove X amount of heat from a bigger area. the issue with chillers is not wattage, but surface area. so, for the cost of the 4 226 watt pelts, you could get say 16 40 watt pelts and have a way more efficient chiller by virtue of the surface area.

The main thought was not formed completely. to use the surface area effectively, you need a lot of flow past the heatsinks your using for the cold side of the pelts. an efficient design might be a circular reservoir with the line into the loop at the bottom, the line out of the loop at the top with a couple rows of pelts running down the sides.

 

[doc]

This is the problem in cooling anything. How much surface area do I have to do this and how much time will it take to get there. In my case with a large ( 5 gallon ) buffer, time was on my side, so to speak. During times when the system was not in use, the pump was not on. Therefore it became a more passive system. The heat sinks that were inside the container had lots of surface area, which are coated with a thin layer of an epoxy based paint. Yes! It made the HS a little more resistive, but most of the time the TECs were pumping a fraction of Qmax. The return line from the pump was setup ( at the top of the container ) so the coolant would go thru and around the heat sinks. With the turbulence and mixing this caused, I get OK cooling . If you operate the system with constant current instead of constant voltage they become almost self regulating. In other words as the temp in the buffer rises, the TECs will pump more heat. If what you are cooling does not need a specific temp, it does work to a degree.

 

[Paxmax]

In other words as the temp in the buffer rises, the TECs will pump more heat. If what you are cooling does not need a specific temp, it does work to a degree.

Cool effect !

Well... I have two things to try b4 I settle for one of them.
My future system has: P4 1.8G, 2 pcs 226W TEC. 2 or 3 heater cores. 2 or 3 Delta 190 cfm(yep, noisy, but they, a bit of headroom is good to have!!)

Option 1.) Mill waterblock for TECs mill coldplate for P4. 720 gph pump.

Option 2.) Mill a "TEC holder" direct TEC-die cooling 720 gph pump. Direct CPU-die cooling with water from TECs coldside, pump 320 gph. Ha, this option is inspired by aenigma.