BM2 battery cell undervoltage (CUV) protection in action

One of Pumpkin's customers is currently TVAC testing Pumpkin's ECS1 sequencer as part of an integrated test campaign. The ECS1 includes Pumpkin's manned spaceflight-rated Pumpkin's EPSM1 power system and BM2 battery, and Pumpkin's MBM2 OBC running a version of Pumpkin's GUTS flight software (FSW). GUTS records telemetry from all of the Pumpkin modules at 1Hz, and stores it locally for instantaneous and later retrieval; by default, up to 30 days of telemetry are stored on the MBM2. In the ECS1, several hundred telemetry points are being recorded each second, all in engineering formats (mV, mA, dK, etc.).

During a multi-hour cold test / eclipse simulation involving external heaters under ECS1 control, the power draw of the heaters exceeded the system's available and stored energy, and depleted the BM2 battery to the point where it shut down safely. Upon later retrieval, the stored GUTS telemetry reveals what happened, and what followed. Let's understand the data:

From the BM2's cell voltage telemetry, we see that just before 02:00 one of the four cells in the 4S2P BM2 battery hit the cell undervoltage (CUV) threshold of 2.5V, and the BM2 immediately disconnected itself from the system. Note that the other three cells were (still) well above the CUV threshold, and the pack voltage was above 12V. Herein lies the importance of the safety features of an integrated Battery Management System (BMS); every cell in the BM2 is individually monitored to stay within specified safety limits, as is the entire pack. A system that had only been monitoring pack voltage instead of all four cell voltages would have missed this critical situation. It's also apparent that despite careful impedance matching of 8-cell 18650 sets that are combined into the four cells of a 4S2P BM2, the cell voltages diverge fairly dramatically as the battery approaches zero SoC.

No BM2 data was recorded between 02:00 and ~10:50, since the BM2 was offline due to its CUV condition. Then, the external heaters were disabled,  system power became available for battery charging, and BM2 recharging (and telemetry) began immediately, along with some BM2 heater action due to ambient conditions. The four cell voltages converged tightly with one another; that's partly because they had around nine hours of rest in their depleted state.  The staircase pattern is due to the BM2's internal cell heaters cycling on and off as the cell temperatures rise and fall between the heater's setpoint and setpoint+hysteresis temperatures.  Individual cell temperature telemetry during this time period can be seen below.