You might remember that part of the power control strategy is to use a UVLO (under voltage lock-out).
Well, here it is. It’s a simple Schmitt trigger with a Zener diode to get a stable voltage reference. The power starts at 10.3 V and shuts down at 5.1 V (the switching regulator spec says that it needs at least 6.5V to work, but actually, if you feed it 5V, you get almost 5V).
The input voltage has to be divided before being fed to the non-inverting input of the op amp, because one has to remember that when dealing with power monitoring circuitry, the op amp is powered by the same voltage it is supposed to monitor, thus the op amp output is always less than both the power rail and the monitored voltage, either positively and negatively.
The UVLO consumes 140 mW at 23V, which I find large. This is mostly the energy absorbed by some current limiting resistors. It would be possible to save a few mW, but I haven’t found any clever way to improve a lot on this.
This is only part of the story, since there is a need to be able to maintain the onboard computer alive for some time if lighting / solar power goes down somewhat, either because of clouds or when the angle of the panel with the sun changes, which is quite expected on a boat. Since there’s this design decision not to use battery, maintaining power for a few minutes involves supercapacitors, the other op amp (it’s an AS358, so there are two in the package), and a P-MOSFET to avoid disturbing the MPPT algorithm… Stay tuned.