Thermoelectric (Seebeck) solar power at reasonable cost

Interesting project featured on Hack-a-Day today:

http://www.instructables.com/id/S6KN8N2FYTCMBPS

A Fresnel lens, some black spray paint, a Peltier cooler run backward, a heat sink, and a cooling fan, makes a reasonable amount of power at a reasonable up-front cost.

I'd have gone with liquid cooling, and done a CVD coating of porous amorphous carbon (i.e., used a candle flame to apply soot) rather than use spray paint.

The author says it's similar in $/kW, but I'd take that with a grain of salt because he didn't compute power correctly (the right way is actually very complicated), and doesn't include the cost of a solar tracker.  It's still an interesting result for such an early effort.  It would be most interesting to those with a cheap source of Seebeck/Peltier devices.

The author speculates about an insulating aperture: I might use three sheets of Al foil at (and near) the focus point.

So .... optimizing the general design could lead to more watts per dollar than PV cells? 

paul wheaton wrote:
So .... optimizing the general design could lead to more watts per dollar than PV cells? 

I think so.

Certainly better EROEI.

But that isn't exactly an apples-to-apples comparison.  With liquid cooling and a good concentrator/tracker setup, a small section of ordinary PV cell can put out much more power than if it is in normal sunlight and cooled by natural air flow.  Given a well-cooled homebrew concentrator setup, it may well be that PV is cheaper than Seebeck.

The fact that this technology is fundamentally thermal opens possibilities for energy storage, for co-generation of hot water, and for alternative heat sources powering the same electrical setup when sun isn't available.

Ultimately, I think it's most interesting for those who have Peltier devices laying around already.