Hi all,
Should anyone face the same problem, I found the root cause: brownout threshold. I burned a new bootloader (Optiboot 8.0) without such trigger and the node has been working with used batteries (~2.7V) since May.
Hope this helps.
@JahFyahh - I still haven't managed to et round to it. I've been really busy since this project and its still hectic here. I'm sure i read a few reviews somewhere, here or google. Not sure.
@NeverDie said in The Harvester: ultimate power supply for the Raybeacon DK:
Edit: This guy found another couple of possibilities as to what the E50D chip might be: https://www.electroschematics.com/pfm-module-circuit-surgery/
I have the same board as this guy. I removed the LED and resistor because they are on the input side, so they are almost useless anyway. For a board such as this, with a promised 0.9v startup voltage on the input side and a promised 5v on the output side, they really should have been attached to the output, not the input, to show roughly when it is that the board is producing useable output.
At a 20mv "keep alive" input voltage, the input measures 0.5ma using a uCurrent Gold. That means the keep alive energy is 10 microWatts. Offhand, for the application we've been discussing on this thread, I suspect that number is too high to be practically useful, because at only 20mv input we'd be collecting far less than 0.5ma current from a tiny panel. At least now we know.
