Do you want the time to go into emonCMS (or similar?) I think it might be easier to count revolutions of the feed screw, because that will give you a direct measurement of the quantity - always assuming the pellets don’t run out.
You can do that either with a commercial proximity detector, with a magnet on the shaft going past a reed switch, or with a reflective optical sensor. Without knowing exactly what the feeder looks like, it’s hard to say which would be the best. Optical might not be a good idea if the pellets create a lot of dust, and a reed switch might have problems with vibration.
If the feeder is “industrial”, then my inclination would be to use a proper industrial inductive proximity sensor to detect a metal flag on the shaft - or maybe bolt heads on a coupling? Power might present a problem, because those are likely to require 12 V or 24 V d.c. They are not cheap, but reliable.
I’d forgotten about on-time. My thinking was: the quantity is proportional to time if the speed is constant. OK, it should be. But also, if the calorific value of your pellets is constant, then the amount of heat you get is directly proportional to the pulse count. And emonCMS can directly relate pulses to kWh.
I can’t believe there is nowhere to find any documentation on that setting. I have to say that while the hardware looks nice and promising, the software lacks of everything.
The learning curve is steep, the interface is nowhere from user friendly, the documentation is basically non-existent appart explaining some ready made receipts for some popular use case.
There are no demo versions, no dummy sensors to help understand how the cms works (when one gets it working by installing it himself).
It’s not possible to get an idea of what’s possible and what’s not. The software doesn’t look so open to other data that the sensors sold on the website. For anything else you must develop some code to make it fit to the cms.
I’m really hesitant to put any money in this project.
Not having a device that switches on and off at fairly short intervals, I used my PV systm’s time importing and exporting as the values to track the on-time function. Plotting them on a bar chart yielded: