The balancer zorlig linked is basically the balancer section of my JK (Heltec) BMS with the "Super Capacitor" active balancing. I am using this unit
Smarter Shopping, Better Living! Aliexpress.com
www.aliexpress.com
I do have to say, it works great and will pull the cells into balance in just a couple days, if there is no charge current or load on the pack. More about that later. The one I have can pull up to 2 amps from the highest voltage cell, and then push 2 amps into the lowest voltage cell. The one limitation is that it is only ever pulling or pushing current to just one cell. On a 4S pack, that is not a big limitation, but on a 14S or 16S 48 volt system, it will take longer to pull them all in. I still had to do a bit of manual balance before I assembled my pack because I was connecting the cells into a 2P configuration. I had to be sure that both strings were near perfectly matched or it would have pulled crazy currents between the cells. But once I had the pack all built up, I intentionally pulled down 4 cells with a 12 volt inverter, and then turned on the active balancing and watched it pull them back in to balance. NMC cells do act a bit different than LFP cells. The voltage on these does change faster with state of charge than it will on LFPs. But pulling the cells to the same voltage is still the right thing to do. With a 0.4 volt difference on my 360 amp hour pack, it took about 40 hours to get those 4 low cells back up to within 0.005 volts of the high cells. As my pack charges through the knee (at 60% charge on NMC cells) I do sometimes see the voltages spread to 0.005 (5 mv) but they converge again once past the knee, so I set the balancer to only move power when it sees a 6 mv (0.006 volt) difference. At that setting, I rarely ever see the balancer do anything. On LFP cells, I would expect the balancer to sit and do nothing when the pack is between 30% and 80% charged as the discharge curve on LFP cells is so flat in that area. When it gets above 85% or so, the top state of charge cell will start to climb in voltage, and the balance will start pulling 2 amps from that cell. But it can only do that for a bit less than half the time. So figure an average of 1 amp being drawn from the first cell entering the knee. It will then push 2 amps into the lowest voltage cell. If there are several that are very close, it will spread it out and push into a few cells, but still only one at a time. If you have 3 at the same voltage, each would get the 2 amps, for just 1/6th of the time. If one cell is lagging behind, it will get the full 2 amps for 1/2 the time. So if your pack is far out of balance, it can still take a long time to pull it in.
Now you still need to understand the other limit here. If you are charging at just 5 amps, and you have one cell that is already into the knee, the balancer will only be able to pull the 1 amp average from it. So it is still getting 4 amps of effective charge current. So it can still start to run, and the balancer can't stop it. The BMS function will kick in and shut off charging at the set over voltage set point though. And while in over volt shut off, the balancer can still work to pull that cell down. In on grid backup mode, I still charge / discharge my pack at over 25 amps, and could hit 140 amps max if I fully load the system in a power failure. So that tiny 2 amps of balance current can't do much. If you do plan to run the system so it goes into the top knee on every charge cycle, you may have issues keeping the cells all in line. As LFP cells go into the knee, the voltage can start to change very quickly, especially at higher charge currents. While I am on grid (99.9% of the time), my battery bank does get a 10 hour rest, but it is at 50-60% charge after it runs my house for the peak power time each evening, until the sun comes up the next day. During that 10 hours, the balancer can pull and push power around and balance the cells while there is not charge or discharge current. This might not help much on LFP as they have virtually no voltage difference from 35% to 70% state of charge. You would want the resting/balancing time to be at near full charge to do a "top balance". My setup finishes charging from the sun around 3 pm, and starts exporting power at 4 pm, so just one hour in that state each day. With well matched cells, this is working perfectly, but with bulk Chinese LFP cells, I am not as sure this balancer would be able to keep up with drift. If (when) I go to a large LFP bank, I will not charge to over 3.35 volts each day, and I will monitor it close for a while to see if the voltages start to stray. I really do like the way the active balancer works, and will use one like this again, but with how LFP can run away at the top, I may also add a second resistive balancer that can kick in a large 10 amp draw resistor when a cell is going too high. That is throwing away some power, but that is way better than having a run away cell causing the BMS to shut down the system. and 10 amps at 3.5 volts is just 35 watts for each cell going too high while charging from solar.
The better matched the cells are, the less work the balancer will need to do. Some systems with good cells have no problem with having a BMS that only monitors the cells, or has even just a tiny 30 ma passive balance during the top of charging.