Once an exotic component, solid state heat pumps or Peltier devices are now pretty mainstream. The idea is simple: put electricity through a Peltier device and one side gets hot while the other side gets cold. [DroneBot] recently posted a video showing how these cool — really cool — devices work. You can see the video, below.
Many things in physics are reversible, and the Peltier is no exception. The device is actually a form of thermocouple, and in a thermocouple a temperature difference causes a voltage difference. This is known as the Seebeck effect as opposed to the Peltier effect in which current flowing between voltage differences causes a temperature difference. It was known for many years, but wasn’t very practical until modern semiconductor materials arrived.
The problem is that all the heat has to go somewhere, so there is need for some way to move heat away from the hot end. The experiments in the video use a TEC1-12706. That number may look cryptic, but it actually identifies the parameters of the module as explained in the video.
These coolers are not terribly efficient but for places where you need to get an area cold with very little extra volume and mass, these are just the ticket. We love seeing these used in odd contexts. For example, using a peltier instead of dry ice in a cloud chamber. While it isn’t very efficient, you can use one to make a kind of air conditioner.
if you cool the side you’re not heating with a heat sink does the output of the device go up?
If you put a heatsink to the cold side you are heating it, not cooling it. You are allowing more heat to go into the cold side, because the temperature difference is negative: the cold side is colder than ambient.
[I didn’t see the video] Did it explain why you shouldn’t use PWM to drive one of these, particularly if you want to cool something?
it did not.
I don’t know for certain, but can imagine that each one-off cycle causes thermal stress which leads to shorter device life. There are a large number of dissimilar metal joints in these. With each cycle of heating and cooling, these joints expand and contract at slightly different rates and this eventually leads to fatigue.
Thermal cycling matters if using on/off thermostat control (and manufacturers recommend you don’t do that), but PWM is usually run at too high a frequency to produce those effects.
The issue I was referring to was that that whilst the heat pumping effect is proportional to the average current, the I2R heating losses are proportional to the RMS current. A steady DC supply has these two the same, whereas an unfiltered PWM supply has an RMS/average ratio larger than 1. Sometimes much larger, depending on the duty cycle. This needlessly produces extra heat just where you don’t want it.
its because in the off time the heat from the warm side will go back to the cold side – its only transfairing heat while its on – during the off time it would work like a heat generator! i am not 100% sure but that would be my explaination
that is part of it, the other part is that they have some resistiv loss and P=I^2. So half the current all the time, is less loss that twice the current half the time
This is the best answer!
Wrong. You have to assume you compare it with the same Power, so half the time the current would be square root of two.
Or said differently:
“Half the current all the time” has not the same power as “twice the current all the time”. They are different by a factor of square root of two. And so is the power loss.
The reason behind not using pwm is the one from Daniel.
I wonder how inefficient is not very efficient, IIRC at a particular temperature difference, these devices can have a COP of 2.5 or so, I can’t be bothered checking, and can be ganged so as to keep one’s beer drinkable even on a hot day and still with COP greater than one. Significant advantages over the alternatives in certain applications. I’m not planning on watching the video.
COP is useful for comparing heaters. If you want to use peltier as a heater, you will have much better heater than a simple resistor. COP of 1.1 means that pumping 10W into peltier will let you extract 1w of power from cold side (assuming perfect cooling of hot side). That’s not very much. As for temperature difference, with peltiers you will have COP of 2.5 maybe when your hot side is colder than your cold side.
The higher COP might not be what you want though. A device pulling 45W of heat from the cold side is also generating 140W of waste heat while maintaining a 10C differential. It gets worse the bigger the differential gets. That makes stacking particularly difficult, unless the first stage is very small.
Oh, yes, you want the highest possible COP also for cooling. It means less electricity to move the heat. Generating 140W of heat and moving 45W means you need 95W to move 45W, that is a COP of only 140W / 95W = 1,47. If you have a COP of 2 you need only 45W_el. to move 45W and generate 90W of heat.
yes it would work but inefficient as in a case of beer worth of power to cool a bear. Wildly inefficient. Cheaper to compensate wife or child to run you beers from the frig.
That’s not bad since bear’s are pretty big. I’m not sure I’d give one an entire case of beer though… :)
I experimented with these devices a bit in the past, and there is definitively a sweet spot for how much current you should pump in. They did cool, but over some current threshold they became less cool.
Of course, because I got cheap peltiers from Ebay, there wasn’t a lot of data on those.
I’ve heard they’re most efficient at around 50% of their maximum rating. So rather than using one at 12 volts use a pair in series so they each get 6 volts.
So if you’re going to try something like a solid state air conditioner for an old pickup, use four in a series parallel arrangement. I’m thinking a dual liquid coolant loop with the hot side run through an old air conditioner condenser in front of the radiator and the cold side running through the heater core. Equip the system with valves that can be switched in cold weather to route the coolant to the heater core normally.
To power it, insert a piece of square tube into the exhaust pipe and clamp TEGs and heat sinks to all four sides. Of course the faces should be machined flat.
Even if the cooling power only amounts to a few watts, it would put no extra load on the engine, and in a standard pickup cab there’s less space to cool.
That is wrong in many ways:
1) There is no free lunch. Even if it has only a cooling power of a few watts, it loads the engine with at least double, probably 4 times, that amount. Effiency of the alternator system.
2) You would not feel a few watts of cooling. Typicel vehicle A/C systems have a few Kilowatts of power.
3) Heat exchangers for liquid cooling need some 5 to 10K of temperature differential to work. A typical peltier delivers not more than 20K difference, if it has to move any heat power. So the tiny cooling power of peltiers will be sucked up mostly by the heat exchangers.
4) The temperature differential which can be achieved by peltier elements is also limited by it’s own thermal conductivity. If you parallel two underdriven elements, you also parallel (double) the heat loss through the thermal conductivity of the elements themselves.
So all together this is useless.
tech ingredient and Matthias Wandel have made very explanatory videos on the subject.
I used one in the 80’s to keep a small cooler cold. Kept it at work, in a drawer. People freaked out when they discovered my hidden “refrigerator”. Worked great and fit 6 cans of pop, some lunch meat, and fruit. Totally silent too
These are used in fans on top of woodburning stoves in Europe to disperse the heat further into the room.
This is a datasheet for the module which includes COP info.
http://www.thermonamic.com/TEC1-12706-English.PDF
For 10K difference the COP can be high, 3 or 4, but most applications want at least a 20K difference. Maximum COP drops to 1.5 or 2. To make a fridge that amounts to a “70W module” actually providing 9W of cooling with 6W of input power getting rid of 15W waste heat. That 9W includes the load from leakage through the insulation and the 20K difference includes the load the waste heat puts on the heat sink system.
My experience with these some years ago (trying to make a tiny fridge for a single carton of milk in the office) was that they excel at making one side hot and the other side even hotter. :-)
If all you want is a temperature difference they are great, but to make something actually cold you need a lot of heat-sinking on the hot side.
I made a Peltier Snapple cooler for my office, and I can get the temperature down to about 1.5 degC (35 degF). You’re correct that you need heat sinking or everything just gets hot, but I’m using a 90mm case fan attached to the heatsink and it works well.
Pics of Snapple cooler:
https://imgur.com/gallery/1QJ6zzR
I’ve had coolers that use these but they just don’t seem to last.
They seem extremely reliable when you don’t stress them by switching on and off with a Thermostat. Pwm steering at more than 2000 khz is key and surpressing even the smallest voltage ringing of the powersupply that causes spikes going over the max the Peltier can handle.
Wrong. As long as there is a thermal difference there is heat transfer,. The time to ‘cool down’ is much longer than the PWM cycle time. Your ‘off’ time would have to be very long for what you believe to be true to matter, or your heat-sink thermal mass small.