I keep Blowing up my i2C (ISO1540) isolator when plugging and unplugging my battery

Question

I am using a ISO1540 to read the SMBus of some BB2590 batteries. The relevant part of the circuit is shown in the images below

The issue that I am having is that sometimes when connecting or disconnecting the batteries, U14 gets damaged (no signs of damage, no short circuit or anything visible) and I stop getting telemetry of the Cell 1 of the battery until the IC is replaced with a new one.

The connector of the battery is rigid and that allows all the cells to connect 'at the same time'.

I suspect that the issue might be related to the fact that U14's Ground is the Positive side of Cell 0 of the battery and maybe during the connection/disconnection process there is a voltage spike or similar, but not 100% sure.

Anyone has any pointers on how I could protect the ISO1540 to avoid damaging it with the normal use of the battery?.

EDIT: J10 is a JST, J5 and J6 is an XT30, they are wired to a connector similar to this one: tacticaleng.com/bb-2590-smbus-male-connector

With a PCB ike this one: tacticaleng.com/pcb-2590

So when the battery is plugged in, it is plugged in to the connector in the links. J5, J6 and J10 are "hard wired" to the PCB

[

Answer 1

The earths for the two I2C busses are pins 2 and 5 on the round connector, and may mate before or after the 'pogo-pins' of the 2 SMB busses make contact. You have to assume that things may end up 30V apart, briefly, if the ends of the series pair connect before the test circuit connects to the middle. If the chip ends up doing the splits seeing a voltage between SCL and SDA because earth breaks first (or makes last) then it may go into SCR latch up and fail once power is properly applied. Series resistances of ~ hundred ohms to the pogo pins and if need be diodes up and down from clock and data and data to ground and supply (or a ~ 4V zener between the clk/data pin and ground pins next to the chip to clamp down to -0.6 V and up to Vzener ) will limit the stress. The reg will die either when there is an over-voltage on the input, or a voltage on the output (perhaps a load side capacitor holding up) and a low impedance on the input side, so a large current flows the wrong way. Again current limiting resistances (between Batt CO and the capacitor on pin 2 of the reg )turn a dead short fail into a survivable overload. That resistor should not drop more than 5-10V on the largest current the interface would draw working normally.

Answer 2

So taking into account all the comments, we went and re designed the i2C protection.

Assuming that we were having issues with the ISO1540 IC due to ESD in the I2C bus lines. We increased the protection.

The current solution is to add a filter and surge protection to the I2C Bus. For that we are using in series 33 ohm resistors, in series Wideband ferrite beads and TVS diodes. We are using the proposed method explained in Wurth's application Note 121 (https://www.we-online.com/components/media/o734709v410%20ANP121a%20%20Filter%20and%20surge%20protection%20for%20I2C%20Bus%20EN.pdf)

The added protection to the I2C bus looks like this:

The new design has been tested (there was a 50 cycles of connection and disconnection of the battery performed) and we have not seen the problem occur again.

Took a while to answer due to the time it took to make the new PCB prototypes with the added protection and other features. Time will tell if the PCB stands the test of time and usage.

Many thanks for the help, gracias totales!