Notes from the R&D Department
When I started working on the next iteration of the BootstrapSolar charger, I had a few objectives in mind:
Lower overall cost — The primary reason for lowering cost was to lower the retail price, to make the product even more affordable to more people. In particular, I wanted it to be affordable to people in developing countries, where phones (and soon smartphones) are, or will become, ubiquitous.
Lower barrier to small-scale distributed manufacturing — This objective was also in response to interest I received from the developing world. Even though the first Chiqoo is Open Source, the barrier to actually manufacturing copies is quite high. The SMT components on the circuit boards require automated assembly, which is cheap at volume, but requires a fairly large up-front investment to achieve that economy of scale. I wanted the design to be easily and affordably replicable, even in quantities of 10.
Use standardized battery — This was to give consumers greater flexibility, give ourselves greater flexibility in sourcing, and also for upgradeability. The case of the current version is designed around a specific size Li-Po battery, but Li-Po battery sizes are not standardized. So, this effectively locks our end users into buying replacements from us, and similarly, we’re locked into buying from the same manufacturer. This gives everybody less flexibility.
These 3 goals led to the prototype you see above. The single-board design reduces costs from the Chiqoo’s two board design, and it also uses through-hole components that can be hand soldered. One reason that wasn’t possible in the original was because the MCP73871 charge controller chip only came as a BGA package, but the new version uses a LT3652, which is available in MSOP. MSOP isn’t exactly easy to hand-solder, but by making the pads a little bigger, it can be done consistently with steady hands. Even in low quantities (10+), the new version can be made for ~$30 each. The LT3652 is also better (IMO) than the MCP73871 because it’s a switching regulator, and as such, can accept a much wider range of input voltages. I’ve seen the prototype work fine from 6V all the way up to 20V, which means it can be hooked up to not only a 12V system, but even a large roof-top “12V” solar panel that outputs 20V without load.
As for the battery, I decided to go with the 18650 cylindrical battery. They’re fairly well supported, and cost as little as $3, depending on the capacity. The capacity isn’t quite as high as LiPo, mostly due to its physical size, but high-end 18650s are up to 3000mAh or higher, which is good enough for a phone charger.
While testing 18650 batteries, I got some LiFePO4 batteries, and started doing some reading on them. The main advantage, in my opinion, is that LiFePO4 is more chemically stable than LiCoO2 (the more common Li-ion chemistry), making it less susceptible to thermal chain-reactions (i.e. less likely to blow up), and is more tolerant of over-current and over-voltage situations. LiFePO4 also supposedly loses capacity more slowly and can withstand more recharge cycles than LiCoO2. What’s not to like?
This led to the experiment in the picture above. Do we really even need a charge controller? In this picture, input from a 3.5V or 4V 5W panel is going straight into the battery with nothing more than a Schottkey diode in between. Since the panel produces no more than 5W (but generally less), over-current isn’t an issue. And since the panel also won’t output more than ~4.5V, over-voltage isn’t an issue either (a fully charged battery seems to float at around 4.2V with a 4V panel, which is perfect). In fact, I’ve read that some cheap solar garden lamps are starting to use LiFePO4 batteries in a similar setup, without any complex charge control circuitry. This seems like a great way to reduce cost and complexity.
That led to the Chiqoo v3 that you see below. Basically I ripped out the charge controller circuitry, and simply put in a fuse to protect against over-current, and a couple of zener diodes to protect against over-voltage (I might do 3, to protect against common voltages like 6V, 9V and 12V). And as long as it’s used with LiFePO4 batteries, it’s probably pretty safe. The only downside of this design is that the input has to be ~4V, which is uncommon (though I’m reading reports of LiFePO4s getting charged at up to 10V with little ill effects, so this might warrant some investigation).
So, right now, we have a couple of prototype designs. The “v2” which is more capable in many ways than the original, but is potentially lower cost. And the “v3” which is significantly simpler, and also lower cost. I don’t yet know what a production run will look like, but stay tuned…