One (or more) cells from one (or more) batteries has started to age.
In other words, their internal resistance has started to get bigger.
That mean that the voltage of those cells will rise first (when charging) or drop first (when discharging).
The addition of a supplementary charger, that will increase the charging current, made this behavior worse.
From your graphs, the high voltage was around 12:00.
In that moment the voltage of the pack was not even 51V… it was around 50.8V.
But the high cell voltage was already above 3.6V.
A simple calculus, for example if a single cell is a runaway cell and is rocketing towards 3.6V, for a 50.8V pack voltage, gives that the other cells are at an average of 3.37V.
A deltaV of about 230mV. Enormous…
That because the algorithm implemented for preventing those high voltages inside DVCC scripts is too slow for adjust things and is focusing only on the voltage.
In the case of high internal resistance runaway cells this is not enough.
Also the fact that the Pylontech batteries doesn’t adjust the CVL and CCL dynamically and are keeping a too high CCL in the final stage of charging is not helping.
And updating the Pylontech firmware will not solve the high resistance cell issues, because it can’t.
I also have this situation in one of my systems, but with proper control, see below the results.
It’s a compromise, that will translate in a slightly longer charging time, where in the final stages, when the runaway cell(s) try to misbehave, is kept in check.
1 - the first point when a runaway cell is skyrocketing and the implemented algorithm sees this
2 - the algorithm is pampering the current - see the exponential decaying current, while the voltage of the runaway cell is kept linear
3 - the second point of interest where the voltage of a runaway cell is maxed out and additional steps are taken to be held in check
4 - the end of charge, where the cells are properly balanced
5 - all this time the algorithm is giving time to the balancing mechanism inside batteries to bring the low cells up to the final voltage.
No voltage and/or current oscillations, all smooth and all below 3.6V - the high cell alarm - with a safe margin.
You can also see that the behavior of the low and high cells are quite the same as yours, but without the oscillating mess at the end.
