Testing mostly complete
Here is the Rev 2 Solar Pi Platter with the Pi Zero V1.3 Camera taking pictures of my neighbor's house across the valley while testing the solar panel we plan to ship to backers.
Those of you following along may remember that when I started testing the Rev 2 board I found two silkscreen labels that had been reversed (I guess I wasn't paying attention when I was moving the labels around for best visibility). Subsequent testing revealed that a change I made in the circuitry included in this rev for the USB UPS functionality exposed an issue with the leakage current in a schottky diode (this revision of the Solar Pi Platter will power your Pi and peripherals directly from a USB wall wart, only switching to the battery if USB power fails). Schottky diodes are amazing diodes that can provide very low forward voltage drops (the one I am using provides as low as 0.2 volts drop which is great). Unfortunately they suffer from a problem of increased leakage current (that's current flowing in the reverse direction when they are being used to block a voltage). And the leakage current is very temperature sensitive, increasing dramatically as the temperature increases. To save parts I included the voltage sense the UPS functionality uses to detect when the USB power is failing in the same circuitry I use to configure the charging circuit for correct operation when connected to a low-impedance USB power supply as opposed to a high-impedance power supply like a solar panel or inductive charger. The leakage current can confuse the charger configuration circuit. I looked at a bunch of fixes (including just making the high-impedance charging path less efficient) and ultimately decided to add another resistor to shunt the leakage current to ground.
A long story to explain that I am definitely going to rev the board again - in fact I finished that tonight. If the swapped silkscreen labels had been the only problem I could have just changed the documentation - although that's a bit sketchy, it would have been fine. I am going to take a risk and order the first batch of production boards without testing again. I feel it's a reasonable risk because I was able to add the resistor without touching anything else on the board and I have tested the fix on my lab bench. All other hardware functionality on the board seems to be fully functional (the firmware as well although we'll continue to test that since we can make changes to it for some time to come).
Next up is to change the manufacturing documentation to reflect the addition of the new resistor and send the design package out for quotes to some domestic (US) and some Chinese assembly houses. Then we'll order parts and I'll start on the design of a test fixture to verify the assembled boards.