Hybrid/toggle-able TEC system

I have been interested in TECs for a while now, namely how to integrate them into pc watercooling while minimising the hassle attached to them.

The main issues i see with TECs are; a) their power draw, b) expandability/size, and c) condensation controlling them.

Most people on the internet seem to agree that for every 1w of "cooling" power, 1w of power from the wall must be expended. This means that under full load, my computer would likely use well over 100W, just on cooling; even at idle a noticeable amount of power would be used just to cool it. Contrasting this with a traditional water cooling loop with the same amount of radiators, the only part using power is the pump, the radiators can easily passively dissipate the heat load at idle. While you could simply cut power to the TECs at idle, i do not think the water alone could effectively dissipate the heat without any form of radiator.

The likelyhood of failure in a simple pc water cooling loop is less than that of a TEC cooling solution, moreover, fixing a water cooling loop is more straight-forward than a TEC loop. Moreover, a TEC system physically takes up more space than a watercooling loop, thus in the even of a failure in the TECs, i would likely have to tear the entire system out of the computer or the computer would be unusable until the issue has been rectified.

Condensation and control go hand in hand. Obviously you do not want water settling on your motherboard, thus you would want the water to be above the dew point. I have never been able to come to a conclusion on how to control a TEC such that the fluid never drops below the dew point. I have had thoughts of using the motherboards PWM signal to control the cooling power of the TEC, however, i have been unable to figure out a circuit design. I have thought of using an arduino board to monitor the humidity of the atmosphere in the case and the fluid temperature, and then defined a target temperature based on that, again, i was unsure how to design a circuit to do that. I understand that when using PWM you would likely need a MOSFET, but i am not confident in my circuitry skills.

2/3 concerns i think can be resolved with a design I thought of today; what if you used two (or more) remote irrigation valves to connect and disconnect the CPU cooling loop and the TEC cooling loop? When the computer is at idle, the valves could open and the TECs could turn off, thus the heat produced by the computer would be dissipated by the radiators which would usually be dedicated to cooling the TECs. Moreover, this would reduced the risk of condensation damaging the computer, as the TECs would not be turned on constantly; also if the TECs fail, then the system could be used as a traditional water cooling loop.

My main questions are A) What is the best way to control a set of TECs based on fluid temperature (or something else if fluid temperature isnt the best metric to base it off)?, and B) would a system like this be feasible?

Attached Thumbnails

 

yes, it can be done and yea their are people that do something like what you have described with controllers keeping the fluid above the condensation point.


https://www.overclock.net/forum/62-p...build-log.html


this is a functional, affordable TEC build.


https://www.overclock.net/forum/62-p...c-cooling.html


this is a extreme and expensive TEC build.


I would guess that you are looking to build something in between the two?

I've seen your build before, it's a marvel hahaha.

I think i would definitely be on the cheaper end. I do not want to spend too much, however, i know i will spend more.

Both of these builds used relays to control the TECs, right? Do these control the TEC's power draw based on temperature? If so where could i get such a relay, and how can you program them? To me using a arduino seems like the most logical method of control, however, i am aware that pwm isn't the most efficient method of control.

yes, and both of them are using PWM control. voltage control is the best way and it is also the most expensive. my build actually uses both forms to control the TEC's. my 331watt TEC's are PWM controlled, and my 224 watt TEC's are voltage controlled. at this point I can explain the differences, as far as the actual amount of amperage being used for each. with my 224 watt voltage controlled TEC's, the circuit, not the TEC's, uses a total of one amp to control the TEC's. I can switch these TEC's from 13.2 volts, or 19 volts. with my 331 watt TEC's, with PWM control. this circuit uses two amps to control the TEC's, I can control them from 1%, to 100%, or what is basically, 1 volt, to 26.4 volts. yes you could use a arduino, but that may be a little expensive as I gave up on that idea. the other build has a temperature sensor that keeps the water at or around the set temperature. he still needs to use a humidity detector, and figure out what the dew point will be manually. than he sets the controller to keep the water temp above that point. so I do not have a good answer for you on that point.

Has anyone tried controlling the TECs with the motherboard's 4pin pwm fan headers? Would it be safe to do something like that?

My initial idea was to use a relay switch to 'transfer' the 5v PWM signal to the 12v signal, but that doesn't seem feasible (due to the expected life span of a switch which is ~10e4 cycles), so i decided using a MOSFET to control the 12v signal would be best. Did you make your own circuit to control the TECs, or did you buy a PWM controller? I would probably use maximum 2 100w TECs, at 12v that would be ~17A, could i get away with DIYing a simple MOSFET-based PWM circuit?

yes a simple circuit could be made, here is a build that did that.


https://www.overclock.net/forum/62-p...build-log.html


there is also another building a more complex controller.


https://www.overclock.net/forum/62-p...modulator.html


as for my build I got a 100amp 30volt PWM controller to control two TEC's using only 36amps total. I did this because I want the controller to last, I got a 40amp PWM controller that failed rather quick, it lasted 3 months.


I do not think that any motherboard controller would be able to handle the amount of amps needed by TEC's.

You can use the PWM signal from the motherboard as a method of control. It can be used in the same way as the Arduino PWM output (Pin B5) in this circuit:


Basically you want to use the PWM output as the input to a MOSFET driver, and then the output of that to drive the actual MOSFETs. This way you don't need to rely on the motherboard PWM output to provide high current. You could directly drive small MOSFETs with it, but the big ones you're likely to use for TECs take more power to switch.

This schematic is for the Arduino based PWM controller PCB that I'm using in my TEC build.

The only issue I can see with the motherboard controlling the TECs is that there isn't a way to implement dew point aware control. A lot of high end boards these days come with temperature sensor inputs, but the fan/PWM control options are lacking. For example, I don't know of any way to have the PWM shut off if one temperature falls below another.

Attached Thumbnails