Fairy tales… perpetual motion machines… Not all fairy tales are perpetual motion machines, but all perpetual motion machines are certainly fairy tales. Before I get into the specifics of thermoelectric coolers, however, it seems appropriate to set the stage for this particular category of fairy tales.
There are two classical types of perpetual motion “machines,” called (not so creatively) “type 1” and “type 2” machines (or equally creatively, machines of the “1st kind” and “2nd kind”). Type 1 machines are the ones most likely immediately familiar to you. They violate the First Law of Thermodynamics, which states that energy cannot be created or destroyed, only transformed from one form to another. Typically, Type 1 machines involve some sort of rotating mechanism that through apparently clever design manages to always have torque generated in a constant direction (or perhaps alternates direction, but with an average favoring one direction). In the absence of friction (or a load), they would move forever without any addition of energy. Type 1 machines are so easy to come by that the U.S. Patent Office won’t accept applications for machines of this type without a working model. In the rare cases one is provided, the “cleverness” invariably lies in hiding a small energy source somewhere, and the patent officer’s job is to be smarter than the inventor and find it! The most blatant examples of Type 1 machines are where the inventor actually claims to be driving a load even though there’s no energy source for the machine. Sneakier examples don’t hide the fact that they have an energy source, they merely claim to deliver more energy out than they take in. For instance, a few years back I was asked to evaluate the “free-energy zero-cogging generator” which claimed to deliver more electrical power out, than the driving wind-turbine put in. (In this case, I believe the inventor wasn’t intentionally deceptive, but he was woefully ignorant of how to measure electrical power!)
Type 2 machines are more subtle. They violate the Second Law of Thermodynamics, which states that entropy cannot be reduced (in a closed system). Entropy is a concept a bit difficult to grasp, let alone quantify, but very often it can be boiled down into the simple observation that heat can never passively flow from a colder place to a hotter place. If that appears to be happening, you’ve either missed something crucial, or else you’ve got a bona fide Type 2 perpetual motion machine. I recall (embarrassingly) an exam on my first undergraduate thermodynamics course. We were asked to evaluate a curious (and fishy-sounding) thing called a “vortex tube.” In a vortex tube, compressed air is supplied into the base of a T-shaped pipe, and, amazingly, cold air comes out one branch of the T, and hot air comes out the other branch of the T. I was suspicious enough to realize that this implied that somehow some energy was moving “uphill” from the temperature of the incoming stream to the hotter output branch. The problem statement was very specific, and included mass flow rates and temperatures and pressures, so I proceeded to do the calculations showing that even though no net energy was being created, the net entropy of the outflowing air streams was less than the entropy of the incoming air stream, thus proving its impossibility. Turns out, vortex tubes are a real thing! I’d made a calculation error, though the professor was generous enough to grant me partial credit for at least thinking of looking for a violation of the 2nd Law. My point here is that the 2nd Law needs to be considered whenever you’re trying to “pump” energy from a cold place to a hotter place.
Enter Thermoelectric Coolers (or TEC’s). These are clever little gadgets that use the well-established Peltier effect. They’re kind of like reverse thermocouples. You’ve probably seen them somewhere yourself in the form of a beer cooler or something similar. They obviously work (and have been patented). One of the niftiest things about them is that they have no moving parts and can be totally silent. You apply electricity at the terminals of the device, and one “side” of the gadget gets cold (the “inside” in the case of an RV refrigerator), while simultaneously the other side (or outside) gets hot. Obviously, if your surrounding environment’s temperature is somewhere in between those two temperature extremes, heat will necessarily flow out of the hot side to the environment, and heat will flow into the cold side of the device from the environment (or whatever it’s touching, e.g. your beer). If you’re paying attention, you’ll conclude two things: 1) this might be a really clever way of cooling electronics without having to use fans or liquid coolants; and 2) if this isn’t violating the 2nd Law, there’s some critical item we haven’t yet bothered to consider (and it may bite us in the end).
Here’s this thing: It’s called the Carnot efficiency of a heat engine. In application, it gives you a quick assessment, based on the temperatures involved, of the amount of extra heat you’ll have to add to a cooling system in order to move some of that heat from a colder place to a hotter place. (In fact, it’s what allows you to avoid violating the 2nd Law). For the sake of argument, it might turn out that to move 1W out of a junction, you have to add an additional 1 W, meaning that your final heatsink has to reject 2 W to the environment instead of the original 1 W. Whence does the extra energy come? Through those nice, quiet, electrical terminals. Volts applied times amperes supplied equals extra energy that wasn’t there before.
Aye, there’s the rub! Sure, you can create a miniature Peltier cooler and lower the junction temperature (Tj, the “inside” of an electronic component) to something cooler than the surrounding environment, or even – let’s not be greedy – just make it lower than it was without the cooler! The problem is, when you turn on the cooler you’ll be adding energy to the overall system to get that lower Tj. From a macro-scale thermal analyst’s perspective, this is usually the wrong thing to do, because more often than not, you were already having trouble getting all the heat out of your system in the first place. (Indeed, that problem is why your Tj was hotter than you wanted to begin with.) For instance, your PC board resistance might have to be 2x lower than it was before (bigger heat spreader, larger fan, etc.), to reject the heat added by the cooler in order to get the lower Tj. But if you could do that, then you should have just done that – in other words, without adding the cooler – and you’d have lowered your Tj a bunch anyway!
Now I can think of a couple of situations where a TEC might be an excellent choice, but you need to be very sure of your calculations. The first is, when you have a very small, localized, concentration of heat and you can afford to drive down the temperature of that spot at the expense of heating up everything else around it just a bit. The second is, when you actually need to control the temperature of a specific device within an electronics system, for instance, an image sensor (where so-called “dark current” is a serious problem and goes up rapidly with temperature). In this latter case, you have to have some margin in your system’s “thermal budget,” because from a system perspective you’re going to have to get rid of some extra heat.
My advice is to think very carefully about whether a TEC is really the right thing for your electronics cooling problem. And using it to cool your beer may not be the best choice, either, if you’re going to be trying to think carefully about cooling your electronics! You be the judge!