Balance can be achieved in the upper portion of the charge curve. 3.65V is ideal, but 3.55V can get you close enough. The goal is to get cell SoC imbalance at peak charge to a very very low %.
Most BMS begin balancing well before 3.55-3.65. Once cells hit 3.40V under modest currents, they are close to full. Typically, a BMS begins pulling <100mA off the high voltage cells to balance as the cells fill.
Not even close.
NCA, NCM, LMO are near constant slope linear between the legs. LFP is FLAT.
View attachment 87637
Your experience with Cobalt chemistry is clouding your perception. LFP voltage in the upper/lower legs is "mushy" like lead-acid. When you terminate Cobalt at 4.20V, there is very little voltage settling. LFP behaves much more like lead-acid. Charge at 2.4V, settle at 2.15V in a day or two. LFP isn't that bad, but it is similar. LFP resting voltage will fall from 3.65V to 3.50V within a day or two. Smaller cells can drop dramatically faster. I have some CALB cells that will drop from 3.65V to 3.35V within just a few hours. After a year of sitting, they dropped to 3.30V, yet they retained 97% of their charge, i.e., were at 97% SoC even at 3.30V.
The combination of charge termination current and peak voltage matter. Most specs call for charge to 3.65V @ 0.5C terminating at 0.05C. If you terminate at even lower current, you can get fully charged at 3.55V. I have personally charged cells to >98% full at 3.45V with very long absorption periods and low charge termination current.
The practice of balancing only during charge and starting at or around 3.40V when cells are out by more then 20mV can maintain the balance of healthy cells even if they aren't charged all the way to 3.65V.