With all the talk these days about global warming and such, I felt this would be a worthwhile project to take advantage of all that heat. The intent of this project was to build a device that would convert heat energy to electricity using solid state components. The finished product works, but like all of my projects it comes short of working as well as I'd hoped. Never the less, I do know where I went wrong on a few accounts, so a V2 version should be much better. I'll walk you through the construction and point out my mistakes so that if you'd like to try this yourself, you can start with a better design than I.

First, I started off with the cold-side heatsink.



Right off the bat I made a mistake. You see, it is hard to find a one square foot heatsink that is cheaper than a million dollars, so I thought I'd be clever and use an aluminum honeycomb panel that I found at McMaster-Carr. I bought it because it was strong, light, and white. This was a dumb move because heat gets trapped in the air pockets of the honeycomb and soon the cold side gets heat-soaked. That's a problem because the device relies on the temperature differential between the hot and cold sides to generate the electricity.

LESSON 1: Buy a real heatsink and possibly add a fan to the cold side (you can power it off of the device).

Next, I added an insulator between the two sides.



I used high temperature silicon rubber, again from McMaster-Carr. I bought it in a one square foot, 1/4" thick sheet. This was mistake number two. The peltier chips, which I'll describe in a moment, are only 1/8" thick. They need to be in good physical contact with the hot and cold sides of the device. I thought the rubber would be of a low enough durometer (squishability) that it would easily compress to half it's thickness. Wrong!

LESSON 2: Use insulation that is the same thickness as your peltier chips, or that is very, very squishy.

I had to do some soul searching at this point. Should I order a heatsink and the right thickness foam, or press on regardless. I pressed on. But how would I overcome the thickness problem? I decided to stack the chips, two-high. In some scientific tests involving my electric stove and a glass of frozen margarita mix, stacking them was almost as effective as having them side-by-side. Of course, I was drinking another margarita at the time, so my findings might have been off.

Regardless, I pressed on. I had to decide on a layout for the chips.



I settled on a pentagram, because it's funny. Again, not the most scientific reason. Damn margaritas.

Next, I cut out squares where the chips would fit into.



The chips that I keep talking about are called thermoelectric coolers, or peltier coolers. The peltier effect is when a current passes through two dissimilar metals (or semiconductors), it causes one metal to heat while the other cools. Check it out on Wikipedia. We are actually dealing with the reverse of that, the seebeck effect. Luckily, the same device can work both ways. These devices can be bought on eBay pretty cheaply. I paid about $50 for a lot of 10 (137 watt) peltier chips.

Anyway, on with the building. The next step was to scrape away the white coating where the chips are going to sit, and apply thermal grease. Both these steps are vitally important as the device needs very good heat conductivity to function properly.



Then I placed the chips in place, making sure to smear the heat conductive grease between the stacked chips.



I then soldered all of the chips in series.



This may or may not be another mistake. In my early stove-margarita tests, I felt that a reasonable output for one of these chips was around 1.5 Volts. I wanted a 12 Volt output, so I felt that 10 of them should put me near that if I hooked them in series just like one would do with batteries. I now think that there may be some internal resistance problems which negate the usefulness of hooking these chips in series, but I have not been able to verify my hypothesis as I am out of margarita mix.

LESSON 3: Drinking + Science = Bad Assumptions

The last steps were to smear the tops of the chips with more grease, put a black anodized aluminum sheet on top, and hold it all together with aerospace fasteners.



Binder clips can be substituted for aerospace fasteners and are easily stolen from work.

After assembly, I ran the unit through some tests. I was disappointed to find that the output voltage at a reasonable operating temperature was only about a quarter of what I expected, although the dead-short current was pretty nice. I won't get into specific performance numbers, because they make me cry. I feel the lack of efficiency is due to the comedy of build errors. Improvements for version 2.0 would be to not stack the chips, and to use a finned heatsink for the cold side. Those two fixes alone should bring it up to snuff. Also, if you intend to expose your project to very high temperatures (like recapturing energy from your car's exhaust), I would highly advise you to slice the insulating foam with a razor knife along the path of the wires, and push the wires into the foam. Otherwise, they will be pressed right against the hot side of your project, and that's a bad thing. You could also build the unit from the hot side up, so the wires would be on the cold side when you finish.

LESSON 4: Run your wires on the COLD side.

As a final construction tip, if you would like to use this thingy as a solar generator, add a sheet of Plexiglases (or real glass if you're careful) to the hot side. This will act as a sort of greenhouse effect, letting in the sun's rays but trapping the heat.

Happy constructing, everyone!