This chart is updated hourly. It displays all the data since the prototype was first installed. Any quasi-linear sections are due to gaps in the records.
So, what happened in the time around 10/15/2020? Well, two things...the first is that I violated my own rule and implemented a software change without testing it on my prototype system which managed to break the sleep state (bad engineer, no cookie!), the second thing was that I was in the process of a major change to my home network which took a few days to settle out, plus it was raining and I didn't feel like getting wet (no one pays me to do this...). On the good side, in the process I took a harder look at current consumption in the sleep state, and now it is well under 1mA while sleeping. All of the energy discharge is directly coming from the time to wake up, update the display and go back to sleep. This also pointed out that I should (haven't yet) stop the updates at night since no one can see the display anyway.
Additional data captured but not displayed includes the SoC (state-of-charge), RSSI (received signal strength indicator), and free memory (which is really boring since it doesn't change..which is good)
New note for 4/1/2021 - That was a network misconfiguration (aka user error ;) All fixed now...I think.