A few of a batch of chargers, based on the BQ2954, and using a circuit very similar to the one in the data sheet, are
intermittently failing to terminate the charge on a 2-cell Li-ion battery.
End of charging for this circuit (which works great most of the time) is indicated by the red LED going off and the green
LED coming on. The LED normally switches from red to green when the charging current level has dropped to Imax/15=Ifull. Current continues flowing until the level drops to Imax/20=Imin at which point the current to the battery actually terminates. The LED still remains green until the battery is removed from the charger, at which point the LED goes off.
The battery is considered 'charged' anytime after the LED turns green.
The problem I have is that some of the chargers occasionally do not switch to green and continue to stay red, delivering current to the battery even though the current level has fallen below the normal Ifull and Imin current levels.
At this point, the battery can be actually disconnected from the charger, and the RED charging LED doesn't go off! The
output voltage of the charger is still ~8.4 to 8.5 Volts, even though no battery is present (Under normal operation, output voltage of this 2-cell charger would rise to about 15.4V when battery is removed).
Even though the defective chargers work perfectly some of the time, changing the BQ2954 IC usually seems to fix the problem. (This is not a good production solution as chargers are usually tested only once before shipping to customers; they may fail on the 2nd or 3rd they are used).
I have also found, that on two of the 'defective' chargers, increasing the regulator timebase capacitor at TPWM pin 9 from 470pF to about 800pF seemed to make the problem go away (This lowered the frequency from about ~190KHz to ~110KHz: the data sheet example circuit uses a 470pF for C9.
Although I have not been able to get these two units to fail yet, after testing for a number of cycles, I'm not confident enough to suppose this would take care of the production problems -- I don't want them failing in the field.
The problem almost appears as if the internal 'charge control state machine' is sometimes jumping into an illegal state when a battery is plugged into the charger.
Any suggestions or help would be much appreciated. ....Brian