What Is Overvoltage?
Got a call from a pack integrator in Ohio last Tuesday. They had a 14S LFP pack that came back from a solar install with two cells reading 3.91V. LFP. Should never see anything above 3.65V in normal use. The cells looked fine from outside but when we cracked one open the cathode foil had turned brown at the edges. Classic overcharge damage.
Turns out they were using a lead-acid charger. Customer swapped it in because the original died. Lead-acid 48V charger puts out 58.4V float. On a 14S LFP pack that works out to 4.17V per cell. Not a problem for lead-acid. Big problem for LFP.
This kind of thing happens more than people admit.
Overvoltage means pushing a cell past its maximum rated charge voltage. The number depends on chemistry. NMC and NCA top out at 4.20V. Some high-energy NMC variants are rated to 4.35V but those are specialty cells and you need to know what you are doing with them. LFP chemistry has a 3.65V ceiling. LTO is around 2.85V. These numbers come from the cell vendor datasheet. Ignore them and you will have problems.
Internal cell degradation
What happens inside the cell at overvoltage is not complicated. The cathode material wants to give up oxygen when you force too much lithium out of it. That oxygen reacts with the electrolyte. Meanwhile lithium metal starts plating out on the anode surface because the graphite cannot absorb ions fast enough. The plating is bad for two reasons. It is irreversible capacity loss and it creates dendrites that can eventually short the cell internally.
A lot of people think there is margin built into the 4.20V spec. There is not.
Cell manufacturers set that limit at the point where degradation becomes unacceptable. Going to 4.25V once is probably fine. Going there every cycle will kill the cell in a few hundred cycles instead of a few thousand. Going above 4.30V and you might not get a few hundred cycles. I have seen cells swell up at 4.35V after a single charge. Depends on the cell.
The Role of the BMS
The BMS is supposed to catch this. Every cell in the pack gets its own sense wire. The AFE chip reads all the cell voltages and compares against a threshold. If any cell goes over, charging stops. Pretty simple.
Except the BMS can fail. I have seen BMS boards with cold solder joints on the sense wire connectors. One cell stops reporting and the firmware defaults to zero instead of flagging a fault. I have seen AFE chips that drift out of cal over temperature. TI's BQ76940 is generally solid but the older BQ76925 had issues with the internal reference shifting. Cheaper Chinese AFE chips can be all over the place.
Balancing matters more than people think. A pack with ten cells in series will have some spread in capacity. One cell hits full charge before the others. If balancing is too slow the high cell sits at 4.20V while current keeps flowing into the pack. The voltage on that cell creeps up. With passive balancing you are limited by how much heat you can dump through the bleed resistors. Most designs run 50mA to 100mA balance current. If your cells are mismatched by more than a few percent that might not be enough.
Active balancing moves charge from high cells to low cells instead of burning it off. More expensive. More complex. Makes sense for large packs where the wasted energy adds up or for applications where you cannot tolerate any capacity spread.
Charger design is the other half of the equation. A switching converter with sloppy feedback regulation will overshoot at light load. I have measured chargers that put out 42.5V when the pack draws less than 100mA near end of charge. That extra half volt distributed across ten cells is 50mV each. Not a disaster but it adds up with the other tolerances.
Temperature compensation in the charger matters too. Lithium cells should charge to a lower voltage when hot. Some chargers adjust the CV setpoint based on a thermistor. Most cheap ones do not. A pack sitting in the sun at 45C getting charged to the normal 4.20V per cell is effectively being overcharged.
Two protection layers are better than one. The BMS watches cell voltages. A secondary protection IC can watch pack voltage and cut a FET if something goes wrong. For the Ohio pack that started this whole discussion, neither existed. They had a dumb BMS that only did balancing. No protection. Customer assumed the charger would handle it. Bad assumption.
Design Checklist
If you are designing packs the checklist is pretty short.
- Use a BMS with real cell-level OVP.
- Set the threshold with some margin, maybe 4.18V for NMC.
- Make sure balancing can keep up with your cell spread.
- Qualify the charger across its operating envelope not just at room temp on the bench.
- Add a secondary protection path if the application justifies it.
The Ohio pack is getting rebuilt with a proper BMS and a charger spec'd for LFP. Expensive lesson. Could have been worse. Nobody got hurt and nothing caught fire. That is not always how these stories end.
