Cooling System with Peltier element

[knauf]

Hi guys, need a little help with an understanding of how Peltier element works in a real life. I've designed a cooling system based on TEM module MCPE1-07106NC-S, consisting of:
*) PCB plate 20 x 20 x 1,6 mm with a heat source;
*) one copper plate (Cold Side) where PCB is fitted with screws;
*) one copper part that designed as a heat exchanger; it's made as a "T" shape, where upper cap is used to hold TEM element (Hot Side), and bottom stem is rounded, with a hole inside for connecting heat pipe;
*) heat pipe SF-08-300-S;
*) heat sink LA28;
*) two clamping plate for connecting heat pipe to the heat sink;
*) polished aluminium cylindrical shell ("hermetically sealed chamber") with ID = 37 mm and OD = 40 mm; the overall height is 135 mm.
*) round PCB plate D44 x 1.6 mm that works as a feedthrough between two separate environments;
*) aluminum ring that clamps the round PCB to the shell.
Now I cannot tell all the details, but the presented information should be sufficiently descriptive to make some conclusions on the system assembly. Anyway, to cut to the chase, I've tested this system without any thermal load, creating vacuum inside the "chamber" (constantly pumping out air, pressure inside was p = 6*10^-4 mbar), at ambient temperature T = 26~28d Celsius. According to the preliminary calculations and powering Peltier with I = 4 A, the temperature on the Peltier's cold side should be somewhere -24d Celsius (or even lower!), on the hot side - +34d Celsius. During the tests the hot side temperature has risen to +38d C, but at the same time the cold side temperature has dropped only to -11d C! I've tested this system also having an Air inside the "shell". And without shell at all, placing instead of upper PCB a rigid foam cap. But nothing has changed! The results stays the same. How can it be? I mean, yes, I admit I haven't included into my calculations the heat loss from radiation. But they shouldn't have such an effect at temperatures around ~-20d C! And then again, nothing has changed when I've removed the shell completely. The heat transfer through the conductive wires? Negligible, less than 0,3 W. Heat transfer through the clamping screws (between hot side and cold side)? Special bushings have been made from POM to insulate steel screws. I even removed half of my screw during one of my experiments and it only improved the cooling temperature by 1d C. My final assumption is that I have slightly defective Peltier, that has max dT ~ 60d C instead of claimed dT = 70d C. That, of course would explain everything. But is it really so? Or am I missing out something?

 

[3DDave]

It's a bit curious as a vacuum doesn't have an ambient temperature. If I had to guess it's that there is a flaw in how the temperature is being measured. It also doesn't seem that tough to set the Peltier device in the chamber all alone like the data sheet says and check it to see if it's working right. The other consideration is the device just hasn't the capacity for the amount of heat to be moved or there is a path back from the hot side to the cold side.

 

[davefitz]

I do not have any real life Peltier wheel experience, but in theory if one substitues an isentropic expansion ( turbine, peltier wheel) for a throttling valve , then the exhaust of that engine will be much colder than the inlet, regardless of the working fluid. The work performed by the engine can be used to generate electriciy and be synched to the grid using the same technology as used by microturbines ( eg Capstone). The same thermodynamic theory is used in the Linde refrigeration cycle for cyrogenic liquids.

"...when logic, and proportion, have fallen, sloppy dead..." Grace Slick

 

[Rputvin]

The last company I was with used peltier chips (can't recall if series or parallel electrically) mounted between two plates - the hot side was finned for a cooling fan, the cold side had an additional heat exchanger manifold on it for cooling the fluid passing through the manifold. Monitor outlet and vary the chips as needed in heating/cooling mode to maintain a constant fluid output temperature. We had different OEMs making the main unit, we slapped on a cover, wired the terminals, bolted on the heat exchanger with some thermal grease, and away we went.

They were nothing but headaches if we tried to get them to run to calculated/theoretical performance levels. Inefficiencies, differences in ambient air temp used to reject heat, restrictions in airflow once in the field, etc. They still worked and were fairly trouble free, and thankfully the bulk of the applications were never sized right at the edge of performance. I don't know if the end users saw any fluctuations based on seasonal temperature changes, or if they worked well enough to be consistent year-round.

We ended up "derating" them to the performance levels we could actually achieve. Didn't stop sales from promising the moon and blaming engineering for a product that didn't meet expectations.