"Our second-best prototype had 90 percent efficiency – less than 10 percent of the energy was wasted," Dean says. "Our best prototype had 95 percent efficiency. And both had component costs of about 50 cents. All other converters either cost more, were less efficient or both."

Well, I feel smug. I used a cheap low end PIC processor to do this only 3 weeks ago and got 97% efficiency using synchronous rectification.

Additionally, some low-energy processors have switch mode converters built in already. For example the nRF8001 and it's sister parts.

You can build a high-efficiency converter with any processor that has a built-in comparator. External parts needed (if you do synchronous rectification):

1x schottky diode and 1x small cap (~1nF) for bootstrapping the high-side MOSFET gate.
2x NCHAN MOSFETs
1x resistor (noise filtering on the output)
1x storage cap (>10uF)

Use the right component values and you'll get 95%+ efficiency, and it's practically free.

Haha, I forgot the most important thing, you need an inductor of course... Duh!

Most "sidekick" comparators are pretty slow, around 1us, so you'll need to use an appropriately large inductor to slow things down so the comparator can respond quickly enough. At least 100uH.

Oh, and don't use the darlington transistors they show in most application notes, they are horrible and utterly ruin the efficiency (<75%). Just use nice, high side bootstrapped MOSFETs.

Nice! What operating point (Vin, Vout, Iout) got you that 97% efficiency? We were targeting 5 V in, 1.8-5.0 V out at < 1A. Which MOSFETs did you use?

We were creating a system runs a multithreaded application (GPS data logger) and three buck converter control loops on a 32 MHz 16-bit MCU, with all of the code meeting its deadlines using a real-time kernel and design methods. So once we hit those efficiency numbers we didn't try to raise them any more. We use burst mode at low current to get there.