[Request] Base load corrected voltage on inverter DC input power

[schlimmchen]

Is your feature request related to a problem? Please describe.

The load corrected voltage is currently calculated based on the AC output power of the inverter. However, the inverter's AC output power has little to do with either the battery voltage nor the battery current draw.

The only thing we are interested in is the voltage drop of the battery due to its load. This allows us to delay stopping the discharge beyond the stop threshold. Once discharging stops, we can expect the battery voltage to bounce back up to the stop threshold.

The voltage drop is proportional to the current draw, as the (internal) resistance that causes the voltage drop is constant (we ignore the battery's temperature for now).

Hence, what we actually need to know to calculate a load corrected voltage is the battery's current draw. The inverter's AC output power does not correlate well with the battery current draw, as the battery voltage can be significantly different. We could ignore that as we use the load corrected voltage only when the battery is near the stop voltage threshold, so it is okay if the load correction factor is tuned based on that one battery voltage.

We then still have the inverter's efficiency to deal with, which depends on its output power and hence can be different each time when approaching the stop threshold, in which case the battery current draw will also be different.

Describe the solution you'd like

For that reason, we should calculate the load corrected voltage based on the inverter's DC input power (not AC output power).

Describe alternatives you've considered

No response

Additional context

If we want to take this a step further, we should base the load corrected voltage offset/diff solely on the DC current at the inverter's inputs. This would allow us to estimate the battery's idle voltage using the same load correction factor at different battery voltages. We then need to convert the existing load correction factor. As long as we only need the load corrected voltage in the context of one battery voltage (the stop threshold), we might as well keep using the DC input power.

[schlimmchen]

For that reason, we should calculate the load corrected voltage based on the inverter's DC input power (not AC output power).

Depending on the situation, this also does not make sense. The battery could actually be in an unexpected state, e.g., the battery could actually be idle as the power the inverters consume is actually provided by a solar charger, or mostly by a solar charger, so the corrected voltage is too high. I mentioned this already in #676.

There is also #874.

The idea in #1111, which gets rid of the load-correction factor altogether, seems to be the best option to me, especially since setups without solar charge controller data and without a battery interface provide very limited means to mitigate the issues.