Can I make a homemade Peltier module with constantan wire?

Question

I'm trying to make my own Peltier module.

I ordered constantan wire, which I used as N-type semiconductor and iron wire as P-type.

I made this and powered it with 3V, but without response.

What am I doing wrong?

Answer 1

Iron and constantan make a good measurement thermocouple, where thermoelectric efficiency doesn't matter. The metal properties of ability to be drawn into wires, flexibility, and be used over a wide temperature range make them good, practical sensors.

While they can be run backwards as a heat pump, the efficiency is so bad that not only are they completely impractical as a device, you would have difficulty even measuring any heat pumping effect.

Peltiers didn't take off as practical cooling devices until we started making the thermocouple junctions from semiconductors, which materials have a several order of magnitude times better FOM (figure of merit) than metals.

The FOM \$zT\$ depends on the material's electrical conductivity (\$\sigma\$) (good), thermal conductivity (\$\kappa\$) (bad), and Seebeck coefficient (S) (good).

$$ zT = \frac{\sigma S^2 T}{\kappa}$$

A high thermal conductivity conducts heat across the device without being useful. A low electrical conductivity generates a lot of heat in the device due to current flow. Unfortunately, in metals, the thermal conductivity and electrical conductivity tend to both vary together, as they are both mediated by the 'sea' of conduction electrons. They also have a very low S.

When you go to non-metals, the two conductivities use different mechanisms, so there's the opportunity to make materials with better ratios, and better S.

Even today, the materials used for Peltiers are only just good enough to make devices for niche applications where nothing else will do. Compressor systems are an order of magnitude better.

The above equation is from the Thermoelectric Materials article on wikipedia. It also gives a comprehensive list of the materials presently being used, and those being researched in the pursuit of better \$zT\$.

Answer 2

Have you tried putting about 1,000 junctions in series electrically, and in parallel thermally?

Constantan-Iron has a thermal EMF of less than 300uV for 5°C.

The thermal leakage will be huge, I doubt you get much temperature difference but it should be measurable, with care.

P.S. you may not be able to get actual cooling, but even if not there should be a measurable difference in the heating of the two sides that is polarity dependent. You could detect it with a (slow) bipolar square wave current and synchronous demodulation of the differential temperature.