Thermo-Electric Generator - TEC1-12706

The manufacturers sell Peltier junctions as heat transfer devices, and char acterise them for that application. Exploiting them as Seebeck generators i s perfectly possible, but not popular.

Internal resistance depends heavily on the size of the device and the way t he junctions are assembled. Lots of junctions in series have a higher serie s resistance than the same junction is parallel.

The tendency is to characterise them at the highest temperature difference they can sustain (while transferring no heat) and at the higher heat flow t hey can offer (while sustaining no temperature difference). Turning that da ta into Seebeck parameters isn't easy, but if you think about the thermodyn amics you can get some of the way.

Price typically depends on the size of device you buy. My guess would be th at you will have to buy one and measure it before you can get anywhere.

Farnell/Newark stocks a fair number of Marlow parts, at intimidating prices - $50 each and up.

Yes, I have two and have done some characterization as I said in my OP. My initial measurement was a bit on the crude side. I used clip leads, but when dealing with mV levels and 1 ohm or less loads that isn't the best way to go. I improved my connections and used some resistor values closer to the expected peak and found an internal resistance in the 5 to 6 ohm range which is not far off from the page I found that showed 10 ohms. Even though we were both testing the same capacity module, it may not be from the same maker and so may vary a bit in the specs.

I'm just surprised the internal resistance is so low, but I guess I shouldn't be, it is rated for 6 amps. At that current it would be dissipating over 200 Watts just in the internal resistance. I should be surprised the internal resistance isn't lower!

I'd like to connect it to one of the LT chips designed to turn this output into useful power. But their eval module is $150. I may make my own board and do some testing. My initial measurements indicate I will need more than 2 of these TEG units to work with the LT devices at just a couple of degrees delta.

characterise them for that application. Exploiting them as Seebeck generato rs is perfectly possible, but not popular.

ay the junctions are assembled. Lots of junctions in series have a higher s eries resistance than the same junction is parallel.

nce they can sustain (while transferring no heat) and at the higher heat fl ow they can offer (while sustaining no temperature difference). Turning tha t data into Seebeck parameters isn't easy, but if you think about the therm odynamics you can get some of the way.

e that you will have to buy one and measure it before you can get anywhere.

ices -

It might make more sense to buy a single bigger unit. That usually means mo re junctions in series, thus more volts for a given temperature differentia l.

The bigger units are more expensive, but you get more junctions per dollar that way. I doubt if you will be drawing enough current for the series resi stance to be of anything but academic interest.

For heat transfer, Peltier effect is measured in watts per amp, while the w aste heat you generate in the junction is proportional to amps squared, so the series resistance defines your operating range.

Before you go too far down that route check that you cannot more sensibly power the device for years with a button cell.

I have done candle powers a white LED once using a high temperature TEC in series with a low temperature TEC and then a DC to DC converter from that combined output. It was a lot harder to make it work than I expected and I had a huge temperature difference to play with. To make it work really well the lower plate had to be at the melting point of solder and the upper cold heatsink fresh out of the deep freeze.

You can get rather more junctions on a larger plate but check first that the same volume of batteries won't be a *LOT* more cost effective!

We did this a year or so ago... someone had a nice link to old Russian TEG' s that had some efficiency numbers. As I recall they had a few percent (5% ?) of the theoretical max. efficiency. (The max is Thot-Tcold/Tcold) where all T's are in degree's K. You can see right away that a few degrees of t emperature difference is paltry. I think you will also struggle with a low voltage. Putting my hand across a ~128 element TEC I get ~100 mV of open circuit voltage. (My hands are always cold though.) Do you have a way to gain up such a low voltage?

George H.

I've already looked at devices to capture the energy from a TEG. Energy harvesting is not a brand new area and there are a number of chips available. LT has a couple that seem to work well and one even has a pair of inputs which allow it to capture TEG power independent of polarity. So you can get energy on both the ebb and the flood tide.

I think I am going to take the LT eval board design and turn it into a unit that I can make up cheaply using oshpark.com. I'll need to ramp up on KiCAD. I've been using FreePCB which works just fine, but I have wanted to move toward running Linux for some time and working with dual platform tools is the first step. So this can be my test/learning project.

The LT parts use a 100:1 transformer to step up the voltage and claim to work with voltages as low as 30 mV. That may mean the loaded voltage which is a lot lower than the open circuit voltage of course. Even so, my measurements show there is "significant" power. Significant is relative of course. I'm looking for a few hundred uW.

Along with that will be my first rPi effort. I need to measure the

doesn't cut it. So I'll be adding electronics to interface some thermistors and display the temperatures.