Harvesters don’t need their own power supply and ideally the system can come back to life from totally empty (meaning it doesn’t need to be deployed with any charge level)

I was doing my thing back before the packaged harvester management chips were available and I made a PIC based microcontroller system that could monitor the charge level in a capacitor fed from a piezoelectric vibration energy harvester and decide when to wake up and transmit a data reading based on the capacitors charge level.

The trick was getting the PIC’s sleep current draw low enough that the capacitor was charging even at milli g level vibration levels - from memory you can get a PIC’s sleep current well down in the tens of nano amp range.

PICs have a built in voltage reference that is stable (a zener or schotkey diode from memory of 0.6V?) and a way to read what the Vsupply (basically the capacitor) is compared to it. So the PIC would wake up every 30 seconds, check Vsupply to the built in reference and, when high enough fully wake up, take and transmit a reading and go back to sleep.

From memory I got it running at 50mg transmitting every few minutes.

I got a nice journal publication in IEEE for that work.