Building a 12 tec chiller to decrease power usage. Yep, decrease.

gruntledweasel · 04-13-2004, 10:57 PM

I picked up one of those thermotek chillers off ebay for cheap a while back. I'm still modding it to suit my needs (flow restrictions galore in that thing!), but one thing bugs me a bit. The darn thing's rated to draw more current than my full size fridge, and I'm going to be cooling a loop dumping 170-190 watts or so into it. I think that might put it fairly close to its peak power draw. I don't pay for electricity, but that kind of inefficiency seems a bit silly if I can do something about it, you know?

I had an idea when I saw an auction like this on ebay for a bunch of 72 watt pelts. (Don't be fooled by his numbers, qmax != max watts drawn). I got 15 for about three dollars each. I'm building a "prechiller" into the unit using some parts I've just got sitting around. I've got a few of those koolance hard drive cooling blocks sitting around, and it turns out six 40mm TECs will fit quite nicely on each side of one. Playing around with kryotherm, I can see that twelve TECs of roughly similar specs (HAIL module) running at 2-3 volts can cool a pretty hefty heatload down to below ambient without drawing much current at all (~20W at 2V, ~50W at 3V, entire pelt array), even with a crappy C/W for the coolant system and TIM joint.

So I'm going to build a prechiller out of three of those hard drive coolers and a small radiator and pump I've got sitting around (previous crappy WC setup, I've long since upgraded). The two "hotside" blocks are in their own loop, cooling one chiller block sandwiched between them. The assembly ought to fit in the thermotek box where the dehumidifier assembly used to sit. Dumping 50-100 W of heat out via this second loop should let the actual chiller unit run its pelts at a lower voltage and let me push the water temp farther below ambient.

I've done some preliminary testing to see exactly what voltage I want to run the peltiers at (the thermotek unit's still in pieces pending the purchase of some stuff, so I can't do a test with the final setup). It's looking like 3V might be the way to go. I did a quick test with two pelts stacked, but that didn't yield much of a deltaT improvement.

Just as an aside...I'm well aware of the problems involved with inline chillers. Bear in mind that I'm only augmenting a setup that's already "chiller-only" (no radiator inline). The only way this'll hurt me is if the coldside temp of the peltier is actually greater than what the equilibrium temperature of the water would be without the additional chiller inline. This would only happen if the thermotek unit were running all-out on a heatload significantly smaller (80W less or so) than what I will be putting into the loop.

I know much better could be done (koolance equipment =

), but this is a "I had it sitting around anyway" kind of mod. Thoughts? I promise I won't make scatological remarks about others' neural equipment if my idea needs improvement.

bigben2k · 04-14-2004, 12:10 AM

I have a 252. I'll be using it for my testbench.

The power draw actually depends on the heatload that you're going to run through it, so I wouldn't worry about power consumption. The ratings on the machine are Max, not actual.

Hey, do you have the software for your unit?

gruntledweasel · 04-14-2004, 01:07 AM

Quote:

Originally Posted by bigben2k

I have a 252. I'll be using it for my testbench.

The power draw actually depends on the heatload that you're going to run through it, so I wouldn't worry about power consumption. The ratings on the machine are Max, not actual.

Hey, do you have the software for your unit?

The breaker in my bathroom thought I was showering with a hairdryer when I first plugged it in, so I figured it was drawing a fair bit of juice.

But even if I'm not helping with current draw, I'll still do this little project to get coolant that's farther below ambient.

No software here, I'm sorry to say. I've got a 252 as well. The dude I bought it from assumed I'd be using it for an aquarium. Any idea why this particular model's seen such widespread use for non-lab work?

Groth · 04-14-2004, 08:29 AM

No neuro-scatology? What a shame.

I'd say you're well on your way to serious chillin'. If you can find the details on the TEC's used in your t253, model them in Kyrotherm as well. That way you can pick voltages/temperature such that the CoP (Coefficient of Productivity) of all your bits is similar and the heat load is properly distributed.

I wish I had free electricity.

murray13 · 04-14-2004, 09:51 AM

I've got three of them! One ripped apart (completely).

It's time for me to get some sleep, but I WILL get back with info and some ideas.

Just off the top of my head the Thermotek uses 10 pelts in series. Will look and get the exact specs later.

bigben2k · 04-14-2004, 01:04 PM

Quote:

Originally Posted by gruntledweasel

The breaker in my bathroom thought I was showering with a hairdryer when I first plugged it in, so I figured it was drawing a fair bit of juice.

But even if I'm not helping with current draw, I'll still do this little project to get coolant that's farther below ambient.

No software here, I'm sorry to say. I've got a 252 as well. The dude I bought it from assumed I'd be using it for an aquarium. Any idea why this particular model's seen such widespread use for non-lab work?

That's normal: it's going to draw a lot of power on startup, as it adjusts to the set temperature.

As far as I've been able to find out, some company (Chrysler?) had a lot of them, then decided to dump them all, and they ended up on eBay.

Murray13: got the software?

murray13 · 04-14-2004, 02:50 PM

The Thermotek I have torn apart had ten Marlow DT12-6 modules wired in series. There is about 60V across the modules when cooling at max.

Yes for an additional 50 or 60 watts (electrical) you will get an additional 4 to 7 deg C lower temps.

Ben: No I do not have the Thermotek software to monitor/control the unit.

Cptn. Foo Foo · 04-14-2004, 05:45 PM

Quote:

Originally Posted by bigben2k

Hey, do you have the software for your unit?

I do not have one of these units - I just read this thread.
After browsing around the Thermotek site I came across this software:

http://www.thermotekusa.com/download/ttk_rs232v623.zip

Its called "TTK RS232 Interface Ver 6.23"

Is this what you guys need?

gruntledweasel · 04-14-2004, 06:48 PM

Rawk! Certainly looks that way. Can't say for sure as my unit's still in bits and pieces right now, but I think we have a winner.

...I also think that the full contents of that directory aren't really supposed to be public?

edit: and let me just say I feel really stupid for not thinking to browse around their /download directory.

bigben2k · 04-14-2004, 07:44 PM

Downloading... installing... (DOH! I'm at work!)

Yep, it sure looks like it. I'll give it a go from home. Thanks, Cptn. Foo Foo!

Cptn. Foo Foo · 04-14-2004, 08:14 PM

Quote:

Originally Posted by bigben2k

Yep, it sure looks like it. I'll give it a go from home. Thanks, Cptn. Foo Foo!

You're welcome

gruntledweasel · 04-14-2004, 08:50 PM

Quote:

Originally Posted by bigben2k

Thanks, Cptn. Foo Foo!

Heh. Why is common courtesy like common sense?

Thanks Cap'n.

gruntledweasel · 04-17-2004, 07:30 PM

Minor update to correct something I said earlier that was inaccurate...I don't know what I was smoking when I first tested out peltiers stacked (one pelt on top of the other).
A few recent tests I ran gave me a very nice improvement in coldside temperature with a two-pelt stack, basically a little better than increasing the voltage running through the standard non-stacked pelts by 1V, with the improvement starting to degrade around 4V. I suspect that this degradation was due in large part to the crappiness of the heatsink setup (coolant temperature rose by a little less than 2 degrees C for every 1V increase in voltage, which I didn't see with a non-stacked assembly). I suspect that if one were to make a nice custom waterblock and use a decent radiator/fan setup, the performance enhancement of using stacked peltiers in a chiller (at these low voltages) would be worth consideration.

As it stands with my setup, the power drawn by 24 pelts at 2V would not be that much better than 12 pelts at 3V, and the setup would not be able to handle an increased heatload as well. So stacking peltiers is out. Again, I'll note that if one were designing a chiller system from the ground up, stacking the peltiers would definately be worth looking at.

Anybody know of kryotherm-ish software that can calculate with peltiers in a stacked configuration? As it stands, I've gotta run two copies of kryotherm, and play balance-the-numbers until the coldside temp of the top set = ambient T of the bottom set, watts of bottom set + heatload = heatload of top set, etc. Rather tedious.

murray13 · 04-17-2004, 08:05 PM

The Aztec program from melcor will let you play around with their standard stacked modules. Of which there is one of them that has specs close enough to give you some idea of what you would get. I used it before I found Kryotherm. I know a standard stacked module won't be quite the same as stacking modules yourself but it should be close enough for what your looking at.

04-13-2004, 10:57 PM	#1
gruntledweasel Cooling Neophyte Join Date: Dec 2002 Location: Irvine Posts: 63	Building a 12 tec chiller to decrease power usage. Yep, decrease. I picked up one of those thermotek chillers off ebay for cheap a while back. I'm still modding it to suit my needs (flow restrictions galore in that thing!), but one thing bugs me a bit. The darn thing's rated to draw more current than my full size fridge, and I'm going to be cooling a loop dumping 170-190 watts or so into it. I think that might put it fairly close to its peak power draw. I don't pay for electricity, but that kind of inefficiency seems a bit silly if I can do something about it, you know? I had an idea when I saw an auction like this on ebay for a bunch of 72 watt pelts. (Don't be fooled by his numbers, qmax != max watts drawn). I got 15 for about three dollars each. I'm building a "prechiller" into the unit using some parts I've just got sitting around. I've got a few of those koolance hard drive cooling blocks sitting around, and it turns out six 40mm TECs will fit quite nicely on each side of one. Playing around with kryotherm, I can see that twelve TECs of roughly similar specs (HAIL module) running at 2-3 volts can cool a pretty hefty heatload down to below ambient without drawing much current at all (~20W at 2V, ~50W at 3V, entire pelt array), even with a crappy C/W for the coolant system and TIM joint. So I'm going to build a prechiller out of three of those hard drive coolers and a small radiator and pump I've got sitting around (previous crappy WC setup, I've long since upgraded). The two "hotside" blocks are in their own loop, cooling one chiller block sandwiched between them. The assembly ought to fit in the thermotek box where the dehumidifier assembly used to sit. Dumping 50-100 W of heat out via this second loop should let the actual chiller unit run its pelts at a lower voltage and let me push the water temp farther below ambient. I've done some preliminary testing to see exactly what voltage I want to run the peltiers at (the thermotek unit's still in pieces pending the purchase of some stuff, so I can't do a test with the final setup). It's looking like 3V might be the way to go. I did a quick test with two pelts stacked, but that didn't yield much of a deltaT improvement. Just as an aside...I'm well aware of the problems involved with inline chillers. Bear in mind that I'm only augmenting a setup that's already "chiller-only" (no radiator inline). The only way this'll hurt me is if the coldside temp of the peltier is actually greater than what the equilibrium temperature of the water would be without the additional chiller inline. This would only happen if the thermotek unit were running all-out on a heatload significantly smaller (80W less or so) than what I will be putting into the loop. I know much better could be done (koolance equipment = ), but this is a "I had it sitting around anyway" kind of mod. Thoughts? I promise I won't make scatological remarks about others' neural equipment if my idea needs improvement.

04-14-2004, 12:10 AM	#2
bigben2k Responsible for 2% of all the posts here. Join Date: May 2002 Location: Texas, U.S.A. Posts: 8,302	I have a 252. I'll be using it for my testbench. The power draw actually depends on the heatload that you're going to run through it, so I wouldn't worry about power consumption. The ratings on the machine are Max, not actual. Hey, do you have the software for your unit? __________________ WBTA (Water Block Testing Alliance) Nordic Hardware WC 101

04-14-2004, 07:44 PM	#10
bigben2k Responsible for 2% of all the posts here. Join Date: May 2002 Location: Texas, U.S.A. Posts: 8,302	Downloading... installing... (DOH! I'm at work!) Yep, it sure looks like it. I'll give it a go from home. Thanks, Cptn. Foo Foo! __________________ WBTA (Water Block Testing Alliance) Nordic Hardware WC 101

04-14-2004, 08:29 AM	#4
Groth Cooling Savant Join Date: Mar 2003 Location: MO Posts: 781	No neuro-scatology? What a shame. I'd say you're well on your way to serious chillin'. If you can find the details on the TEC's used in your t253, model them in Kyrotherm as well. That way you can pick voltages/temperature such that the CoP (Coefficient of Productivity) of all your bits is similar and the heat load is properly distributed. I wish I had free electricity.

04-14-2004, 09:51 AM	#5
murray13 Cooling Savant Join Date: Sep 2002 Location: Omaha, NE USA Posts: 216	I've got three of them! One ripped apart (completely). It's time for me to get some sleep, but I WILL get back with info and some ideas. Just off the top of my head the Thermotek uses 10 pelts in series. Will look and get the exact specs later.

04-14-2004, 02:50 PM	#7
murray13 Cooling Savant Join Date: Sep 2002 Location: Omaha, NE USA Posts: 216	The Thermotek I have torn apart had ten Marlow DT12-6 modules wired in series. There is about 60V across the modules when cooling at max. Yes for an additional 50 or 60 watts (electrical) you will get an additional 4 to 7 deg C lower temps. Ben: No I do not have the Thermotek software to monitor/control the unit.

04-14-2004, 06:48 PM	#9
gruntledweasel Cooling Neophyte Join Date: Dec 2002 Location: Irvine Posts: 63	Rawk! Certainly looks that way. Can't say for sure as my unit's still in bits and pieces right now, but I think we have a winner. ...I also think that the full contents of that directory aren't really supposed to be public? edit: and let me just say I feel really stupid for not thinking to browse around their /download directory.

04-17-2004, 07:30 PM	#13
gruntledweasel Cooling Neophyte Join Date: Dec 2002 Location: Irvine Posts: 63	Minor update to correct something I said earlier that was inaccurate...I don't know what I was smoking when I first tested out peltiers stacked (one pelt on top of the other). A few recent tests I ran gave me a very nice improvement in coldside temperature with a two-pelt stack, basically a little better than increasing the voltage running through the standard non-stacked pelts by 1V, with the improvement starting to degrade around 4V. I suspect that this degradation was due in large part to the crappiness of the heatsink setup (coolant temperature rose by a little less than 2 degrees C for every 1V increase in voltage, which I didn't see with a non-stacked assembly). I suspect that if one were to make a nice custom waterblock and use a decent radiator/fan setup, the performance enhancement of using stacked peltiers in a chiller (at these low voltages) would be worth consideration. As it stands with my setup, the power drawn by 24 pelts at 2V would not be that much better than 12 pelts at 3V, and the setup would not be able to handle an increased heatload as well. So stacking peltiers is out. Again, I'll note that if one were designing a chiller system from the ground up, stacking the peltiers would definately be worth looking at. Anybody know of kryotherm-ish software that can calculate with peltiers in a stacked configuration? As it stands, I've gotta run two copies of kryotherm, and play balance-the-numbers until the coldside temp of the top set = ambient T of the bottom set, watts of bottom set + heatload = heatload of top set, etc. Rather tedious.

04-17-2004, 08:05 PM	#14
murray13 Cooling Savant Join Date: Sep 2002 Location: Omaha, NE USA Posts: 216	The Aztec program from melcor will let you play around with their standard stacked modules. Of which there is one of them that has specs close enough to give you some idea of what you would get. I used it before I found Kryotherm. I know a standard stacked module won't be quite the same as stacking modules yourself but it should be close enough for what your looking at.

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