This setup lives on my desk now (click to enlarge):
It shows the potmeter I'm using to find the optimal input resistance to get the maximum of power out of the Peltier module; I'm using the 3105 because it's capable of delivering far more power than the 3588 from the Peltier element.
This setup is something I created few days ago; I suddenly got the idea to make a shower radio based on energy harvesting; the hot side of the tap remains hot, the cold side remains cold while you're showering, and you could power a small FM receiver (bungling at the side) out of the temperature difference, or let a shower timer beep when you've passed 5 minutes hot shower to make it a true 'green' solution. Speaking for it is that the shower is not a place where you want to go with batteries (need all kinds of sealing rings) or AC power (forbidden near showers). It would be a good opportunity, but I couldn't find a good way to mount it; the idea stays alive in my head though....
You can also see my problem with the 'demo board', as I described earlier. It IS a great demoboard, but the possibilities for tuning it to your own needs are limited. One of the legs of the potmeter is now creatively soldered directly to the QFN housing of the LTC3105 after having an accident where I wiped off a small test pad.The circuits are designed to show how packed (and efficient) you can make them, not to allow easy breadboarding. Requires a bit of creative soldering 