Peltier (TEC) + Liquid Cooling Project

[myom]

Hello,

I am working on a project (not related to PC cooling) that involves modulating the temperature of a certain region of interest (involves both heating and cooling). So far, I've got a Peltier (TEC) system set up, and I'm looking to implement a liquid cooling system to transfer heat/cold.

The heating/cooling will take place in a small enclosed space (specifically, at the fingertips of a model baby), and I'm trying to circulate heated/cooled liquid into this space using tubes and a pump. The TEC will be heating/cooling the liquid.

I have 1/4'' ID, 1/8'' OD tubing, and I plan on purchasing this water block (http://www.shop.customthermoelectric...m1plqscsfapp06) so the TEC can heat/cool the water.

Now, here are a few questions I have:
- Will this fitting (http://www.mcmaster.com/#5694t11/=snv896) be appropriate?
- Are there really small pumps available? (To fit in a model baby--space is of the essence)
- Is there a generic water block that I could use to heat/cool the region of interest (fingertips)?
- Will a T-line (instead of a reservoir) be plausible?

 

[Witchdoctor]

Would need a complete design criteria to comment on this

 

[myom]

Would need a complete design criteria to comment on this

What do you mean by complete design criteria? I can describe the system more thoroughly I suppose:

I will need to heat AND cool a substance and need to modulate temperature at the fingertips of a model baby. I figured using tubes and liquid cooling would be plausible (and would sort of mimic the circulatory system).

The loop will consist of the tubes, water block(s?) at the fingertips, a pump, a T-line, and a TEC+water-block unit, which will be used to provide the heating/cooling energy. The TEC will be electronically controlled via an Arduino board.

Is something like this achievable?

 

[Mouse McCoy]

Doesn't look possible but I'm no expert. TECs are much more efficient at producing heat than cold. What are the temperature ranges you need?

 

[myom]

Doesn't look possible but I'm no expert. TECs are much more efficient at producing heat than cold. What are the temperature ranges you need?

Which component makes this seem implausible? The TECs, the liquid cooling, or both? Need to cool to approx. 25C and heat to ~35C.

 

[Bobnova]

That's a very doable range with a TEC. The difficulty with TECs is that they are woefully inefficient. If you need to move 10 watts of heat out of the loop you need to use 50-100w to do it. That 60-110w then has to be removed from the hot side of the TEC.
Getting the TEC, blocks, pump, and TEC cooling into a baby sized/shaped thing is going to be very difficult.
Powering it without external cables is going to be just this side of impossible, if not impossible.

 

[myom]

That's a very doable range with a TEC. The difficulty with TECs is that they are woefully inefficient. If you need to move 10 watts of heat out of the loop you need to use 50-100w to do it. That 60-110w then has to be removed from the hot side of the TEC.
Getting the TEC, blocks, pump, and TEC cooling into a baby sized/shaped thing is going to be very difficult.
Powering it without external cables is going to be just this side of impossible, if not impossible.

Would TECs not be the best option then? So far, it seems like TECs would be small enough to fit into the baby, but the issue is the excess dissipated heat.

That's why I was considering liquid cooling to compensate for the heat. Is there an alternative choice for cooling down the TECs?

 

[Bobnova]

What are you going to power this operation with, how long does it need to run, and where are you going to put the heat from the TECs?

 

[myom]

What are you going to power this operation with, how long does it need to run, and where are you going to put the heat from the TECs?

Will be powered by an ArduinoUNO (for control) + a 12V power supply (for TECs). The operation will likely involve maintaining the temperature at 37C (body temp), cooling down to ~25C (however long that would take) and maintained for a minute or two, then heated back up to the body temperature. The problem here is the cooling (too much dissipated heat from TECs).

Here's the TEC spec sheet: http://www.customthermoelectric.com/tecs/pdf/00711-5L31-03CA_spec_sht.pdf

 

[Bobnova]

So you do have an external power source?
Can you run a couple water lines out with the wires and have an external radiator?

 

[myom]

So you do have an external power source?
Can you run a couple water lines out with the wires and have an external radiator?

Yes, I will have an external power source. It may be possible to run the water lines out to a different part of the baby's body, but keeping everything contained within the baby is a priority.

 

[freakdiablo]

I still think TEC is the better way to go. As you said in your OP, you can use it to both heat and cool the liquid. With a traditional watercooling set up, you're still going to need some sort of heating element, which I feel is only going to complicate it more than simply slapping a large heatsink on it.

 

[Bobnova]

I'm not sure that a 1.5w tec is going to do much. It's only rated for 0.85v as well, so you'll need some regulation going on. With an arduino controlled H-bridge you can flip the hot side and cold side for heating / cooling, so that's a plus.
A 1.5w tec can move maybe 0.2 to 0.8w of heat in a meaningful way, are you sure that's going to be enough?

 

[freakdiablo]

I'm not sure that a 1.5w tec is going to do much. It's only rated for 0.85v as well, so you'll need some regulation going on. With an arduino controlled H-bridge you can flip the hot side and cold side for heating / cooling, so that's a plus.
A 1.5w tec can move maybe 0.2 to 0.8w of heat in a meaningful way, are you sure that's going to be enough?

Oops, missed the spec sheet. I was thinking along the lines of 5-10w, not 1.5w. Yeah, depending on the amount of water in the system that will take some time, if it can even keep up with the water heating/cooling to match the ambient temps. I've seen reviews of those USB powered TEC based drink cooling coasters that say they don't do jack, and they may be even more than 1.5w.

 

[myom]

I still think TEC is the better way to go. As you said in your OP, you can use it to both heat and cool the liquid. With a traditional watercooling set up, you're still going to need some sort of heating element, which I feel is only going to complicate it more than simply slapping a large heatsink on it.

Hm, I see. So do you think slapping on a "small" heatsink to one side of the TEC will do the trick (no radiator)? I'd put on some thermally conductive grease like Arctic Silver 5 then use thermally adhesive tape to attach a heatsink such as this: https://www.sparkfun.com/products/11510. Or maybe I could just attach a sheet of aluminum.

Then I'll be able to transport both heat AND cold out of that side depending on whether I've flipped on heating or cooling? Not sure how transporting cold energy works, bc I suppose "cold" is simply lack of heat.

 

[ShrimpBrime]

Will be powered by an ArduinoUNO (for control) + a 12V power supply (for TECs). The operation will likely involve maintaining the temperature at 37C (body temp), cooling down to ~25C (however long that would take) and maintained for a minute or two, then heated back up to the body temperature. The problem here is the cooling (too much dissipated heat from TECs).

Here's the TEC spec sheet: http://www.customthermoelectric.com/tecs/pdf/00711-5L31-03CA_spec_sht.pdf

TECs this small will do you no justice and will not cool your loop at all.

You want to run 2 water loops. 2 pumps. Minimum radiation on the cpu side (don't want to dissipate you cold from the loop, but in case of TEC failure be able to handle the Cpu load. Leave fans off to keep cold on this loop.) so basically a single 120.2 on a fan controller, or single 120.1.

The second loop is going to be crucial and very important how it's set up. In order to remove big heat from big TECs (cause that's what it takes to chill water being warmed by a cpu) The heat dissipation needs to be super fast and have lots and lots of radiators for heat dissipation. This will increase ambient room temps, so keep the AC on or a window open.

For every 125w of heat, you want to chill about 3x that much. So 3x 12715 tecs would probably be enough for a stock 125w cpu on a loop.

These can chill at 136w http://www.hebeiltd.com.cn/peltier.datasheet/TEC1-12715.pdf and can handle 15v no problem.

Typically, I use these right on top of the cpu, then only needing one loop. Can take 65w chips sub zero and keep sub ambient while fully loaded.

Need any more TEC support, I'd be glad to help. But chilling liquid with TECs is very difficult, I've tried many many different configurations. The most recent would be FX-4300. I have a small thread about it HERE

5.3ghz FX-4300 TEC cooled pictures in case peeps don't wanna click the linky from which these pictures came.

Fun stuff.

 

[Bobnova]

While the above post is very useful for CPU/GPU cooling uses, I think you might have missed the part about what he's cooling

 

[ShrimpBrime]

While the above post is very useful for CPU/GPU cooling uses, I think you might have missed the part about what he's cooling

RIGHT!! I was lost at this statement.

modulating the temperature of a certain region of interest

Figured we'd be chilling or heating a certain region of interest? Or pehaps better translated as a spot or place??

Not sure, I didn't read every single word from every single persons posts.....

Sorry if I made a fool of myself again

 

[ShrimpBrime]

I figured using tubes and liquid cooling would be plausible (and would sort of mimic the circulatory system).

I kinda see whats up here.

This would all depend on time. How much time does he need the liquid (blood) to be chilled or heated. This would determine the size of TECs needed, voltage used, and all the Delta math involved. Also needing to take into consideration the Tmax Qmax Vmax and Imax.

But in a nut shell, this could be done really easily I think as long as there's no major changes in the liquid (or blood) needed at a fast speed or major jump in temps in either direction. I'm thinking this based on the fact the human body regulates and maintains a cool and breezy 98.6f all the time, I imagine the outside of the tubing would be cased in skin perhaps a glove to mimic skin and then pretend we are getting frost bit at the finger tips.

I dunno

 

[myom]

But in a nut shell, this could be done really easily I think as long as there's no major changes in the liquid (or blood) needed at a fast speed or major jump in temps in either direction.:

What would you say is "a fast speed"? Temps might have to jump 10-12 degrees C within at most 30s-1min

Also, perhaps the liquid cooling is a little excessive/too complex. Wondering if a quick and dirty heatsink can do the trick..