Jeff Crystal from Voltaic Systems was nice enough to send me a 3.5W and 6W solar panel to experiment with. Since these are bare panels, I was pretty sure that connecting them directly to the LiFePO4wered/USB charger was not going to work well. For one thing, the unloaded output voltage of the panels is 7V, and the MCP73123 enters overvoltage protection at 6.5V. So once charged, and not loading the panel, you could end up in a situation where the battery wouldn't be charged again because the panel's voltage was too high. It would be flaky at the least. So I decided to add a buck regulator to keep the voltage at 5V and below.
It still didn't work well. The system would end up in a hiccup mode. The charging chip would decide the voltage was high enough, turn on the charging current, which would make the voltage sag, and the chip would stop charging. Over and over.
The problem is that the LiFePO4wered/USB uses a fixed charging current, set with a resistor. So I decided to add a little transistor current source circuit connected to the PROG pin of the charging chip to see if I could make the charge current dependent on the input voltage, so the charger wouldn't instantly dunk the panel voltage when starting to charge. It seems to work great!
I will need to do some more testing with this, but it could make it possible to add a solar option to the LiFePO4wered/USB in the future.