BQ77PL900 destruction connecting battery pack

I don't understand this damage. There are 2 main reasons for TI's concern on the cell inputs:

First, early test boards used capacitors from VCx to ground. As cells were connected, one input would be pushed up by the cell connection while others were held at ground by the capacitors. This exceeded abs max ratings on the inputs and damage was observed. The data sheet diagram on page 6 and the EVM have the filter capacitors between the pins so the capacitors push other inputs up as connections are made to avoid exceeding the ratings.

Second, if the battery is connected to the monitor circuit with N +1 wires rather than N+3 wires, that is the top and bottom cell monitor connections are made on the board rather than at the battery, when power and ground are connected, VC1 is pulled up and VC11 is pulled down.  Initially the capacitors will distribute the voltage approximately equally across the inputs, but voltages can then drift.  This situation is recognized and a programming method suggested in the datasheet.

The main concern in connecting the cells is to avoid exceeding absolute max limits. Ground reference (battery-) is recommended first since all device signals are refereced to "GND".  The bottom up conneciton scheme is believed to minimize stress on the inputs. Certainly this is difficult when signals are grouped on connectors.  When you start connecting, there is no reference between the board assembly and the cell assembly, be sure there is not a potential difference and ESD when they are mated.

Pin 9 is the BAT pin, or the first power pin you would be connecting. The part should not be awake at that time. I've not seen damage on that pin.  The resistive cell divider and power supply should be very safe. I typically power EVMs in this way with 20 to 40V step function and have not had a failure on likely 100's of cycles. You might look for unexpected current paths or check if prior testing of the board assemby may have stressed or damaged the part before mating to the cells.

I'll contact you separately to investigate further. Perhaps other community members will share connection situations they know to avoid or that were particularly successful.

Friday I have solved the problem.