Some thoughts and observations after a first run
The ESC
First and foremost, I poorly dimensioned the ESC since the afroESC 30A is rated LiPo 4s (14.4V), while the panel VOC is about 22.5V and the maximum power voltage under good lighting condition is about 18.6V. I was hoping the voltage would drop rapidly when drawing power, but this is not so the case, so I’m gonna switch to the spider 30A (rated 6s). Rightly or wrongly, I’m not expecting any “burn” problem because, unlike a battery, a solar panel is a current limiting (theoretically a constant current) device, so the current won’t go beyond 11A with 2 panels, whatever happened.
Measuring power
I’ve been using the Attopilot 90A, because it’s rated 50V, while my understanding is that the 45A version is only rated 13.6V. The 90A outputs 36mV/A, that is one unit read on the ADC (10 bits for 3.3V) is about 100mA, not precise enough. Furthermore, the signal on the “I” pin (current reading) appears very unstable when the motor is rotating (this of course is the motor, not the Attopilot), and it seems to me the response is non linear with low amps (< 300mA), so this doesn’t really fit my needs. I’m considering switching to an ACS712 based breakout board for the current measurement “to see”. In that case, I’ll have to do the voltage measurement, but this is just a voltage divider, I can do this by myself.
At first, current reading wasn’t working the way I expected. After some measurements, I got a 20mV voltage between the ground pin of the Attopilot logic circuit and the power ground. WTF ??? Most likely the current from the Attopilot “I” pin flowed back to the switching regulator ground to a certain point. Might be because the wires between the Attopilot and the switching regulator input are quite long thick AWG 14 (they should sustain at least 12A…), making place for some parasitic capacitance (otherwise the two points would be short-circuited, leaving no way for a voltage difference). Anyway, I added a wire between the 2 “grounds” and everything went into order. Classic multiple power paths problem.