Does it mean that 1/6 of the energy is turned into heat when charging?

Really interesting. Have you also seen http://www.bootstrapsolar.com/ which has open sourced a lot of the design and seems to have access to cheaper components.

That's a good point, Jiansheng -- the panel is rated at 3.4 Watts, but Adafruit said they usually got around .5 Amps (~ 3 Watts, for the 6 Volt solar panel) ... maybe the way that manufacturers rate these devices is rather optimistic?

Gregor -- thanks for that link! I hadn't seen bootstrapsolar -- they're using almost precisely the same circuitry (though they went ahead and combined the two separate open hardware projects into one, nice board) -- and, as you say, they located cheaper components. Great find! I'll contact the fellow, point him to what we've done, and look to collaborate / get advice ...

The Adafruit circuits are fairly inefficient, and in practice solar panels are already only like 10% efficient. Story time:

Over two years ago I was working on a Maximum Peek Power Charging (MPPT) circuit for lithium ion batteries based around the LT3652. The first set of boards I printed had a missing wire, and by the time I'd fixed it I was 1) busy writing grants for Public Lab in my free time rather than doing that, and 2) the dude who was letting me use his surface mount soldering gear turned out to be a crazy bank robber (true story).

so I dropped the project... I figured I was fighting uphill against the innovator's dilemma and that someone would have built such an OS MPPT circuit by now, but I can't find any such thing.

The files are on the Yahoo Charge Controller Group, and, using my admin privileges, attached to this research note. I hear by bequeath them to you!

They include the circuits, parts lists and item#'s at Digikey, datasheets, and some other stuff. I actually don't know if the circuits really work, but the background work I did should give you a good start-- I built a spreadsheet around the parameters of the datasheet for picking components.

So rad you're working on this! the world needs a good, non-proprietary field charger for small electronics!

Mathew --

You're like a combination Lucius Fox (Batman) / Q (Bond) / Merlin (Arthur)! But, in addition, you actually employ in the field the tools you create, so you're also, e.g., the Bond to your own Q. The most parsimonious explanation I can think of is that, like Merlin, you are living backwards in time and delivering us the fruits of your future research.

Incredibly helpful post. I'm going to share all of your info -- circuits, parts lists, datasheets -- with Ben, and have a sit-down with him when I return to Amherst on the 28th. Will also look into the Yahoo Charge Controller Group.

And I'm going to want to hear the bank robber story at the next meetup.

First off, nice to see I'm not the only one with this kind of idea. Really rooting for this, the benefits are just so endless. I have been trying to piece together a small solar to LiPo usb charger myself. Still working on the BOM (Bill Of Materials) list and getting everything to make a first test circuit. (seems alot like what mathew did, will take a look after this)

In this time, will I've been searching the web I know it seems easy at first sight but for someone taking it seriously the first challenge is in the optimization. What to watch for regarding input, middleware, output. For my solar cell I'm gonna try a chip with MPPT (Maximum Power Point Tracking) to get the maximum amount of energy out of it, independent of the load. (@Jiansheng Feng: this should make it possible to get the advertised output power. These are measured in ideal conditions like alignment to the sun, outside, load (which depends on the internal resistance). Wiki: Module performance is generally rated under standard test conditions (STC): irradiation of 1,000 W/m², solar spectrum of AM 1.5 and module temperature at 25°C.) The middleware is just trying to keep the losses to a minimum when converting voltages between in- and output. The battery is an out- and input, both states having its limitations regarding voltage and current depending on the used battery technologie. Since these battery technologies keep evolving recent evolution is to include some logic in the battery pack, call it battery management. This is e.g. applied in the green cars. It can do many advanced things but concerning to us the idea is to just make the battery ask an adjustable PSU (Power Supply Unit) the input it needs. The next big challenge is how to make it universal. To make something universal we need standards. This holds in it to simplifying things and think ahead. For example how do we tackle the growing amount of renewable sources with even more diverse outputs. I propose a modular approach. It seems best suited here. It makes it more generally deployable, future proof and more manageable. And I think (re-)usability tops extreme optimization for yield. This modules should best be classified in input, middleware and output classes. Optionally each subdivided in power ranges, I think this is most important for the middleware. For the renewable sources I would suggest we apply the similar idea as used for the batteries. Just make it possible to add the related technology to the source by an individual module which can than be connect to a middelware module of a certain power range. Dividing the middleware in power ranges according to their input should make optimal design easier. Their output should be already universal voltages making most things already possible. Making it really analog variable is not the core of what we are trying to do here. First this can be included in battery technologie depend output modules and can then later on be made in a separate output module that listen to what the battery management has to say.