Setup:
1. Target hardware used for testing has a BQ34Z100-G1 gas gauge and a BQ24725A charger.
2. The battery used is a 3S2P 4400 mAh Li-Ion pack built with LG ICR18650 S3 2200 mAh cells.
3. A West Mountain Radio (WMR) battery analyzer is connected in parallel with the system load to allow me to discharge the battery at a known rate.
4. The hardware's I2C is connected to an EV2300 to a laptop, and the hardware's I2C is disabled so there aren't multiple masters.
5. TI's bqStudio is set to log all registers every second.
6. Calibration was performed prior to the test.
Procedure:
1. At 10:28 AM, I set the WMR to start discharging the pack at 1A. I let this run until the battery voltage reached 8.4V.
2. At a little under 9V, the low voltage protection circuit on our hardware shut the rest of our system down, so only the 1A load was being pulled (normally, the system draws about 125 mA when the display is on and idle). This shows up at 2:05 PM (row 12916). The bqStudio couldn't talk to the gas gauge at this point since the I2C pull-ups are on a separate board that was shut down as a result of the low voltage circuit.
3. At 2:50 PM, I had to turn the rest of the system on so that charging would start (the micro has to momentarily talk to the charger, even though the watchdog timer is disabled). So, there's a section of time where it took over I2C communication and the bqStudio couldn't talk to the gas gauge. This shows up starting at row 14373 in the data. I had to restart the system once shortly after between 2:51:19 PM and 2:52:18 PM. After that, the system was left alone and the I2C was once again controlled by the bqStudio app. Charging was set to 3.3A (0.75C).
Observations:
1. The initial full charge capacity was 3653 mAh instead of closer to 4400 mAh.
2. Sporadically during the discharge, there are all sorts of what look like random values for remaining capacity, temperature, state of health, average time to empty, state of charge, etc. What could cause something like this? I highlighted some of these cells in red in the Excel file. There are also a few graphs in the second worksheet that show this.
3. After the discharge, the full charge capacity improved substantially to 4242 mAh but then dropped a few minutes later to 4159 mAh. The drop down wasn't gradual; it happened within one second at row 14471. The FCC stayed at 4159 mAh all the way until the state of charge hit 97%. When it hit 100%, both the remaining capacity and FCC dropped substantially to around 3800 mAh.
4. The formula from SLUA450 for full charge capacity, assuming it is still applicable to the BQ34Z100, is FCC = Q_start + PassedCharge + RM. Q_start and PassedCharge basically stayed constant when this drop in reported capacity occurred. So, what could cause remaining capacity to change so quickly? The charge current was still tapering during this period, and the remaining capacity doesn't ever go past 3796 mAh while this occurs.
Attached is a zip that contains the following:
1. Excel file of log data from bqStudio along with some annotations and graphs
2. GG file with data flash values
3. SREC file
Thanks!
Matt(Please visit the site to view this file)