What I am finding is that BMS SoC readings are nearly useless. And that can be a problem, especially for LiFePO4 batteries. My cells ar Li NMC so the voltage curve does help give me a reasonable SoC estimate, but the SoC coulomb counting should be better. The problem stems from how it detects a full charge.
The only way to know for sure that an LFP cell is full is to hit the knee voltage at the top. But that is actually stressful on the cells. So we certainly don't want to do that every day as it cycles. What I would suggest is to charge each module individually at just 10 amps or less until the voltage reaches 57 volts. Then hold it there for about 30 minutes as the current falls off. If the BMS shuts down, it means there is an imbalance between cells within the module. Does the BMS let you read out each cell voltage? To get to the full top voltage, all the cells will need to be within 0.01 volts. If the cells are further out, a single cell can trigger a cell over volt fault. The only way around that is to use a much lower charge current setting so that the BMS has a chance to balance the cells all at 3.56 volts (57 / 15 cells). At that point, all of the cells are at 99% full. And hopefully the SoC reading will reset and show something between 95% and 100%.
My JK-BMS is refusing to reset to full while operating in my system now that I have 2 charge sources. When my cells are charged up to 95%, it is only showing 38% on the BMS, but the solar charge controller is reporting 100% full and going to float mode. Once I finish my DC distribution panel, I should be able to easily isolate each battery module and try and get the SoC to reset using my small adjustable charger set down to under 10 amps to do a full top balance.