Another possibility is a change in the firmware way to estimate the SOC.
Here is some of my experience with my FXS 2016.
My old firmware was mainly calculating the SOC from the estimated capacity of the pack in Ah, and the Coulomb counting in Ah. That made the SOC monotonously decreasing, and accurate. Nice !
Indeed, all the logs with this firmware contained lines like
- "Switching to voltage SOC mode"
- "Switching to coulomb couning SOC mode" (there was a typo in the logs: it was written couning instead of counting)
And I am pretty sure also that the new firmware that has been installed is mainly calculating the SOC from the pack voltage, that is greatly oscillating depending on the way you use the bike.
And since this new firmware was installed, there is no line in the logs saying that the SOC mode is using Coulomb counting.
That is why the first thing is to answer if the change in SOC behaviour is correlated with a firmware change or not.
Of course, if the firmware has not been changed, the problem probably comes from the battery.
But it is still not certain. I have also experienced a fundamental change in my bike behavior due to the fact that the MBB had lost all its settings. The most visible consequence was a jump in the odometer (it increased by 3000 kms in one night), and completely wrong odometer count, wrong speeds (all that came from a lost transmission ratio), and also some less visible parameters were lost (pack capacity, VIN number,...). Of course, in that case, the SOC calculation became quite confused also...
Zero reprogrammed the bike and after that it worked fine.
The conclusion is that you should be attentive to anything else that could have changed in the bike (in my case: firmware, and MBB losing its settings).