My foray into TEC cooling

[WoodButcher]

After seeing the threads here on TECs I decided to finish this project. Here is the thread I created on XS, I took a year and a half to finally complete but it was a learning experience. Tecs are not somethig I would consider for any 24/7 application but I did run this setup for the past week so I thought I'd share.

Here are most of the pics but for more detailed explanation and final conclusion you'll need to visit the original thread,

Thermal Enterprises model CP1-12726
50mm x 50mm x 3.64mm
Maximum power consumption 400 Watts
Operates from 0-15.4 volts DC and 0-26 amps
Operates from -60 deg C to +180 deg C
Each device is fully inspected and tested Fitted with 6-inch insulated leads Perimeter sealed for moisture protection

and these,

INTEL XEON
Socket 771
QUAD-CORE 5400 SER. 3U ACTIVE/(1U Passive) FAN AND HEATSINK

and these are what resulted,

my "nightstand" don't mind the carpet, We had a little flood recently,,,

top removed

side open

back, tecs installed

and my thermo control unit.

 

[Schmide]

That's a cool block. Kudos.

 

[daw123]

Nice work :thumbsup:

What tools did you use for the copper sheet fabrication?

 

[WoodButcher]

Nice work :thumbsup:

What tools did you use for the copper sheet fabrication?

Offset snips, red and greens, the yellow/ straights are worthless. A good pair of shears and seamers. I own a 10' brake for big stuff but I did most of this sitting watching the boob tube. Even the soldered joints are done with a jewelers torch, the real tools stay outside.

 

[daw123]

Offset snips, red and greens, the yellow/ straights are worthless. A good pair of shears and seamers. I own a 10' brake for big stuff but I did most of this sitting watching the boob tube. Even the soldered joints are done with a jewelers torch, the real tools stay outside.

Thanks Woodbutcher

 

[Rubycon]

That's a pretty large TEC!

I still say the best way to feed them is using a simple PID (Proportional Integral Derivative) controller so they get a continuously variable input. Of course you'd need a beefy PSU (24V DC at 20-25A) to avoid current clipping, etc. If you really want to avoid condensation an enthalpy function could be also added with the input of a humidity sensor. This would keep the exposed surfaces at just above the ambient dewpoint.

The hardest part about getting this working would be the initial calibration, rough and fine tuning, etc. Any PID system is like this. The end result is a very stable temperature across a wide difference in CPU load. If you allow the controller to sample realtime CPU load it can even compensate for overshooting and keep temperatures shockingly tight. (not really a requirement but fun for control freaks - literally!)

Are you soldering the copper with low temp solder? Soldering copper is fun especially with the right tools. 🙂

 

[WoodButcher]

That's a pretty large TEC!

I still say the best way to feed them is using a simple PID (Proportional Integral Derivative) controller so they get a continuously variable input. Of course you'd need a beefy PSU (24V DC at 20-25A) to avoid current clipping, etc. If you really want to avoid condensation an enthalpy function could be also added with the input of a humidity sensor. This would keep the exposed surfaces at just above the ambient dewpoint.

The hardest part about getting this working would be the initial calibration, rough and fine tuning, etc. Any PID system is like this. The end result is a very stable temperature across a wide difference in CPU load. If you allow the controller to sample realtime CPU load it can even compensate for overshooting and keep temperatures shockingly tight. (not really a requirement but fun for control freaks - literally!)

Are you soldering the copper with low temp solder? Soldering copper is fun especially with the right tools. 🙂

"enthalpy function"??? rofl! Sent me to wiki again, you did...
My thought was to control the temp keeping a few degrees above dew point. I have in my room a thermometer w/ a humidity reading so a quick calculation gave me a rough idea of dew point.
Of course my PSU has no variable control other than hard wiring so my options were 12, 6, and 5v. 6v worked best given the components capabilites but not enough to deal with a loaded CPU. The TECs running continuosly would keep the water just under 22c, no where close to dew point, then 12-14C.
I built the mercury switch box to control the water temp but sadly the setup is inadequate to lower temps enough for the switch to be tested let alone calibrated. I suppose I may have played around with it at higher temps, off at 25c or something but I rewired to 12v Sat morning. The air cooling on the hot side failed miserably even with Delta high speed 120 x 38 fans. My concern with the merc switch was the possibilty that it would sense the temps to slowly, "reaction time" as it were. Like you said this is where fine tuning would come into play. I have a led temp readout w/ probe at the pump outlet for water temp and an IR gun to measure surface temps in other locations but until such time as I pick this up again I'm back to my humble water loop.
I've done copperwork for many years and always used 50/50 tin lead with a flux paste on the roof and a 90/10 for plumbing. I've never been formaly trained or done any real research on the goodies I use because I never needed to, I know what works and how to use it. Ignorant bastard, eh?🙂 When working sheet copper I've always preferred an acetylene "B" tank and torch head w/ fine tip for most applications, not a turbo tip like most plumbers use today. For this I used mapp gas with a small hobby torch like this using electronics flux core solder and paste flux if needed. One day maybe I'll spend the time to learn how to do it right.

A bit O.T. but "Happy V-day Ruby!" your probably working hard to make it a good day for others so I hope you have a good day too. You deserve it,,,,:thumbsup:

 

[Rubycon]

"

A bit O.T. but "Happy V-day Ruby!" your probably working hard to make it a good day for others so I hope you have a good day too. You deserve it,,,,:thumbsup:

Thanks and yes it's a bit crazy today. The parties are about to begin. 😀

B tanks are good as they are a manageable size and have a good supply of gas. Turbo torches are definitely way overkill for 50/50 or even 95/5. Silfos brazing is awesome but WAY overkill for a water block. Now if your block happens to be an evaporator on a cascade system... 😉 I can hear the sound of a turbo torch like a train coming for miles and always know someone is working on refrigeration! :biggrin: