BQ78Z100 reset on continuous I2C register refresh while charging/discharging

Hi,

we have severe problems while refreshing the registers of the BQ78Z100 via I2C. We observed that the BQ frequently stops charging/discharging by deactivating both of its FETs. The effect first seems to be randomly happening, but we managed to produce an environment which can reproduce the error:

System Configuration

• BQ78Z100EVM  (alternative our own Battery Module based on the BQ78Z100EVM)
• MCU connected to the BQ I2C interface (I2C running at 100 kHz)
• 2x 450 mAh LiPo cells connected to the BQ78Z100EVM
• 8.5 V power supply connected to PACK+ / PACK - on EVM board for charging
• Test SW on the MCU which runs only a single test applicatoion:

const uint32_t TestDly = 250;
	
void getGaugeStatus () {

	int16_t voltage = 0;
	int16_t current=0;
	int16_t remaining=0;
	int16_t full=0;
	int16_t soc=0;

	I2CRegRead (BQ_I2C_SLAVEADDRESS, VOLT, 2, (uint8_t *) &voltage);
	delayUs(TestDly);
	I2CRegRead (BQ_I2C_SLAVEADDRESS, INSTCURR, 2, (uint8_t *) &current);
	delayUs(TestDly);
	I2CRegRead (BQ_I2C_SLAVEADDRESS, RM, 2, (uint8_t *) &remaining);
	delayUs(TestDly);
	I2CRegRead (BQ_I2C_SLAVEADDRESS, FCC, 2, (uint8_t *) &full);
	delayUs(TestDly);
	I2CRegRead (BQ_I2C_SLAVEADDRESS, SOC, 2, (uint8_t *) &soc);
	delayUs(TestDly);

	return;
}

void chargeBatteryTest () {

	deactivateDsgFET ();
	deactivateChgFET ();
	for (int i=0; i< 100; i++) {
		getGaugeStatus ();
	}
	activateDsgFET ();
	activateChgFET ();
	for (int i=0; i< 100; i++) {
		getGaugeStatus ();
	}

	while (1) {
		getGaugeStatus ();
		delayUs(TestDly);
		uint32_t safetyStatus = readStatusReg(SAFETY_STATUS);
		delayUs(TestDly);
		uint32_t safetyAlert = readStatusReg(SAFETY_ALERT);
		delayUs(TestDly);
		uint32_t pfStatus = readStatusReg(PF_STATUS);
		delayUs(TestDly);
		uint32_t pfAlert = readStatusReg(PF_ALERT);
		delayUs(TestDly);

		while( (safetyStatus != 0) && (safetyAlert != 0) && (pfStatus != 0) && (pfAlert != 0) ) {
			getGaugeStatus();
		}
	}
}

Test Procedure

1. Power supply off
2. Start MCU
   
   • read BQ status registers
   • activate charge/discharge FETs
   • continuously read the BQ status registers via I2C
3. Wait ~ 30 seconds ==> OK
   
   • no errors occured
   • read status registers works fine
   • battery voltage present @ PACK+/PACK- contacts due to active FETs --> OK
4. Activate Power supply providing 8.5 V 200 mA charge supply
5. Wait 1 - 10 seconds ==> ERROR

Error Behavior

• After the above testing procedure the BQ does not respond on the I2C bus anymore
  • no action on SDA lane (neither ACK nor data transfer): SDA lane stays high due to pull-up whenever BQ has SDA control
  • no attempt to hold SCL lane low to throttle bus speed
  • non responding BQ may appear at any stage
    
    • within transfering a read byte from BQ to MCU
    • after a successful completion of a read/write bus transfer
    • while writing the register address to the BQ
• BQ deactivates the charge & discharge FETs
• After a certain time (not further evaluated at this point) the BQ continous to respond normally to any I2C requests
• The charge & discharge FETs stay disabled
• The FETs can be re-activated via I2C again

Error Analysis

• Increasing the delay (TestDly) between 2 consecutive I2C register reads reduces the probability of this error
  • When I run the Test with the Battery Management Studio & the EV2300 adapter I could not reproduce the error (but I noticed a delay of ~14ms between 2 reads)
  • However I have not found any hint in any datasheet about any delay requirements except the "bus free time between stop and start" of 1.3 us (which we do meet of course)
• The error was never observed when no charging current was provided
• I have captured the activity on the I2C bus during the error:
  
  • normal i2c read
  • 
  • I2C Bus behavior during error
  • 

• I personally have the following theory:
  
  • There is a watchdog present on the BQ
  • 
  • Is it possible that due to continous I2C read request the keep the BQ so busy that it violates the watchdog reset interval which in turn causes a full reset causing all FETs to get deactivated?

My collegue Philipp Schmidbauer already has asked for a service request at your stand at the Electronica in Munich 2 weeks ago, but never received any answer.

I would highly appreciate any help on this topic as it is a absolut kill criteria for our application, i.e. you can't build a UPS which accidently resets itself and thus turning the full system off. That's not what I would commonly call a "uninterruptable power supply"!