I've been trying to build a small cooler using 2 stacked peltier devices, as per Youtube video.

Attached is a diagram of the setup. Results have been poor at best. After several attempts, it seems the TEC-12715 module goes open circuit. (as measured on meter is diode mode)

The TEC-12706 is rated at 15.4v 6A max, and the TEC-12715 is rated at 15v 15A max. I'm using an old 250w ATX power supply which is rated at
12v 9A and 5v 25A.

The TEC-12715 is connected through a temp controlled module(W1209) , whose output relay is rated at 14v 10A. The TEC-12706 is connected
directly to the 5v ATX rail.

Why the TEC-12715 blows, and the TEC-12706 doesn't is confusing at best.

 

Why the TEC-12715 blows, and the TEC-12706 doesn't is confusing at best.

You roasted the 12715 by supplying it from the 12V rail, and by dumping all the heat from the 12706 into it. The current and voltage were technically within the specifications, but the resulting current and heat likely took the temperature far above the maximum rating. Those TEC ratings are generous and assume an "infinite" heat sink on the hot side and a low temperature in the cold side, so that the maximum ∆T doesn't make the hot side too hot. It's very hard to make that happen in the real world.

TECs are also much more efficient if run well below their maximum ratings, where they are very inefficient. Of course you lose ∆T at less-than-maximum operation.

 

I would also consider thermal damage caused by on/off control.

I also don't see any thermal paste between the "15" and its heat sink in the PDF.

Be sure to only use a thin coat of compound and be sure to apply proper compression.

There isn't nearly enough information on the usage of TECs in that video.

 

I believe the thinking was that by running the 12706 at 5v it would run at about 40% capacity and not overheat the 12715, and that the stacking would allow the 12715 to not work as hard.

The heat sink on the 12715 is a pretty good size, although during testing, it wasn't on the hot side all the time. I tried monitoring temp on the stack with an IR probe, and my finger, but just for a few seconds. It got hot for sure, but put fan on it as it got to 100F. In fact when I tried my spare 12715, I put the assembly together without the finger test, and it fried as well.

There is paste on the 15/heat sink side. My bad on the pic. Indeed the video is vague at best. One thing is his heat sink is bigger
and has dual fans, that mine. Despite the video issue, somehow he got to minus -40c with direct connections to the ATX supply. I'm using exact same peltier devices. My paste layer may not be thin enough, so I can try to get it thinner next try.

Any ideas how to throttle back on the current/voltages to get this thing to cool? I'm not looking for -40c but 32F, or close would be great.

I can add a bigger heat sink and larger fan if needed as well.

 

You want to get the hot side heat sink to as close to ambient as possible. There are formulas for that sort of thing involving adding up the thermal resistances between TEC and heatsink and air (expressed in degrees C/watt) and multiplying by the watts of heat flow.

Or, better, consider using PWM speed controllers/dimmers to reduce the duty cycle of the power to the TE modules. Then you can adjust the power inputs to what the heatsink/fan are capable of removing. (You can find pwm modules on ebay in various current ratings.)

For what it's worth, real world performance in a camping cooler is about 20 C below ambient, with about 4 amps and a fan-cooled heatsink.

 

Thanks for your inputs. Is this the type of PWM you are referring to ? :

I might be able to control the 12v TEC directly from this, adjusting the voltage into 12715 as needed. I guess I'd have to bypass
the temperature controller however in which case I won't have a temperature control feedback. I can live with a 20 C
below ambient delta.

 

A good control scheme will be helpful but can't solve the basic thermodynamic problems.

It's a chore but if I were you I'd set up a spreadsheet to model your design. Your inputs would be the voltages to each TEC (or the time-average voltage if you're applying PWM). You need to refer to the technical specifications of the TECs to convert those applied voltages to current, power consumption, ∆T of the TEC and heat flux. That's the chore part - it's not that simple for one TEC let alone two. But it is in theory doable since all the parameters are known and predictable.

This model would allow you to play around with different settings and get an idea where the weak links are, the workable operating area and so on. In the long run this approach will get you to a working device faster and with less toasted TECs than the plug-and-chug, one step at a time approach. Of course that's assuming you're handy with a spreadsheet.

If you're not, then I'd suggest using temperature sensors on the TECs so you can have an idea what's happening in real time. If the hot side of TEC Start with both TECs at 30% say. If #1 is getting too high but #2 is OK, turn #1 down a little and turn #2 up. That sort of thing.

 

Yes, that's typical of the PWM dimmer/speed control modules. I expect they could be adapted to closed-loop control if you replaced the PWM signal from the 555 chip with one from an Arduino (might need a level converter chip).

 

TEC & PWM aren't a good idea. The heating from the current in the TEC goes as the square of the current, so the
PWM peaks really hurt the cooling. PWM also causes thermal stresses in the TEC -- it can break.

 

Although using a PWM would give me better control of input voltage, the loss of the closed loop temperature control is an issue. With the inputs from Michael8, that pretty much puts PWM out of the circuit. I've attached the spec of the 12715. At 12v 9A I be running it a bit too 'hot', however how much of that is due to the 5v device being stacked makes it more difficult to assess.

I may have to gather a handful of power adaptors, since I don't have access to a variable supply that can source 4-9A and see if I can't monitor the temperature on the hot side better before destruction.

 

PWM is ok to regulate the current just smooth out the current before running in through the TEC. Think of a
buck switching regulator with a nice DC filtered output.

 

The temp control module requires 12v and something greater than 1.5A. I tried a DC adaptor at that level, and the temp module went crazy !

Do you think I can use a PWM into the temp module? I know the PWM's allow for output voltage changes, but can I regulate the current independently? I'd like to get to 12v and adjust current from 4-9A. Perhaps 9V with same current range.