Ok, I've decided to come clean on the dirty little secret about this project. :)

It's about the MCP73123 chip on the base #LiFePO4wered/USB module. I may have already alluded to this problem elsewhere, but it's time to fully document it. It also helps explain why it's a bigger deal than it should be to do a build.

Even though this project is built on top of a #LiFePO4wered/USB, the current stock I have that I'm selling on Tindie does actually not work as a base for #LiFePO4wered/Pi. The reason is that my #LiFePO4wered/USB was built using the MCP73123-22SI version that is commonly available in distribution. When I set out with #LiFePO4wered/Pi, I intended to use my #LiFePO4wered/USB stock and just add the top PCB. But my first prototypes that I made that way didn't work.

The reason is that the MCP73123-22SI version has a 6 hour "elapse timer". It will stop charging if the cell hasn't fully charged yet after 6 hours. The point of this feature is to stop charging in case the LiFePO4 cell is bad and doesn't take charge. Of course, the problem is that if the #LiFePO4wered/Pi is used as a UPS, the system will continuously draw current from the USB, not so much to charge the cell as to provide power to the Raspberry Pi. What would happen is that after 6 hours the charger would quit, then the battery would be used up and the Pi would shut down. Not much of a UPS that way. :-/

The solution is to use a different variant of the charging chip, the MCP73123-22JI. It has the elapse timer disabled, and works beautifully with the #LiFePO4wered/Pi. I created all my prototypes by removing the original chip from a #LiFePO4wered/USB and replacing it with a MCP73123-22JI instead. This is fine for some proto's, but not for production of course. So instead of being able to use my existing stock, I'll actually have to do a new #LiFePO4wered/USB build as well, in addition to the #LiFePO4wered/Pi add-on boards... Unfortunate, but that's the way it is.