Charging multiple Li-ion battery modules in series

[Simon]

I will be running 3 independent Li-ion battery modules in series. Each has it's own BMS. For discharging I am fairly happy. Theoretically as identical batteries they should always be at the same state. But what about series charging? Each BMS is independent, each does not know about the others I would be expecting the voltage to just equal out across the 3. Is this a good idea? I expect all is well until one of the modules decides it's charged and cuts out thus cutting all 3 off.

So does it make much more sense to switch the batteries into parallel for charging? Each would be managed separately by it's own BMS so if one wants to cut out the other 2 can continue to charge. I can also give each module it's own charger. This is advantageous as my charging source is at a lower voltage so it seems silly to be boosting the voltage by nearly 4 times with the inefficiencies that brings.

[Siwastaja]

Changing to parallel connection from whatever configuration is pain in the ass, you need to separately charge/discharge the packs first to close enough voltages before you can parallel them, otherwise massive current -> BMS safety cutoff, hopefully. Just don't do that.

Each having their own charger would be simplest and guarantee the packs staying top-balanced against each other, but you need isolated chargers to be able to do that from single power source. That's what I'd go with, maybe, if engineering the thing to use a single suitable-sized pack is out of question.

Also remember to verify that the power switches the BMS uses are properly rated to interrupt the 3 times higher voltage with good margin! This is always a concern when connecting protected cells or packs in series, because the original designer have limited the max voltage somewhere, and there rarely is motivation to go much beyond the pack's own voltage because it would only cost extra.

[NiHaoMike]

Also remember to verify that the power switches the BMS uses are properly rated to interrupt the 3 times higher voltage with good margin! This is always a concern when connecting protected cells or packs in series, because the original designer have limited the max voltage somewhere, and there rarely is motivation to go much beyond the pack's own voltage because it would only cost extra.

Put a zener diode across the output of each pack.

[digsys]

I have been designing / building / repairing LiIon packs for Solar race cars for over 20 yrs, and so have many other race teams around the world. We have a minimum of 45, and up to 55 (in our case) blocks of cells in series,and some much more, with 5-7+ cells per block. I also design the BMS. All team packs are charged using C/Current mode.
That has been all pretty standard for the last 25+ years, the next part is where I differ slightly to other teams -
I NEVER put ANY active silicon across any pack, irrespective of how "safe" it is supposed to be. Part of that is the strict rules on transporting the packs on planes. The BMS job is to monitor all block voltages / run mean / median / averages across the pack under various loads and report any tolerance exceptions. Also to switch / terminate the C/Current value depending on SOC per block. Other teams will run a uP controlled switched protected PFET across each block, to halt voltage increase on any block that reaches preset charge points, and allow the others to "catch up".
In my experience, IF you need to "correct" block voltages more than once per 20 - 50+ times per heavy cycles, or 50 - 100+ per light duty cycles, you have either poorly matched cells or are working them out of spec. Once the BMS issues an alarm / or shuts down the pack, we use the BMS data to balance out any errant cells. As I said, some teams have up to 100 blocks in series, and would have no hope of managing a pack if they had to constantly worry about leveling blocks.
Of all the BMSs I've come across, I haven't seen any that "switch off" the block, usually always just "switch on" some bypass current. As always, YMMV