Before I conclude what I've discovered, I want to share here.
Situation:
MSP430F1611 I2C Master Write every 20 seconds. (No interrupt used, only by polling IFGs)
In between Master Write, 1611 becomes Slave as USART I2C automatically clears MST bit. -> Normal
While in LPM mode for 20 seconds as slave, if a glitch is induced to I2C Bus, 1611 sometime recognize the glitch as START Condition. -> Normal
Then 1611 sets I2CBUSY, I2CBB as 1611 is in Slave mode now -> Normal
After 20 Seconds passed, 1611 Master Write, but ALIFG is set as I2CBB is not cleared. -> Normal
So No data cannot be sent. -> Normal
So i tried to clear I2CBB by clearing I2CEN in the beginning every time 1611 starts the transfer.
Well, THIS DID NOT WORK. -> Abnormal
Somehow I2C hardware module states machine got corrupted?
But I discovered following Master Write works fine.
Then i wonder, maybe clearing I2CEN one time is not enough when I2C state machine got corrupted.
So i add code to re-set I2CEN 10 times before I start the transfer. (10 times .. no solid reason just wanted to do many)
I do this following for every transfer.
void i2c_masterInit(uint8_t selctClockSource, uint8_t preScalerValue, uint8_t clockHighValue, uint8_t clockLowValue)
{
volatile uint8_t i;
// No repeat mode implementation I2CRM = 0
// DO NOT Init when previous transfer is being processed!!!
// checking the i2c status before init is important as we'll call init every time we transfer
U0CTL |= I2C + SYNC; // Recommended init procedure
for( i = 0 ; i < 10 ; i++)
{
U0CTL &= ~I2CEN;
U0CTL |= I2CEN;
}
U0CTL &= ~I2CEN; // Disable I2C
I2CTCTL = selctClockSource; // SMCLK mostly
I2CPSC = preScalerValue; //I2CPSC should be <=4 or error might occur
I2CSCLH = clockHighValue; // High period of SCL (I2CSCLH+2) x (I2CPSC+ 1)
I2CSCLL = clockLowValue; // Low period of SCL (I2CSCLL+2) x (I2CPSC + 1)
I2CNDAT = 0; // initialize data cnt;
U0CTL |= I2CEN; // Enable I2C
U0CTL |= MST; // set to Master mode to I2C register
}
May this workaround is already in Errata sheet but i did not fully understand, but hope this can help other people.
If any TI employee can confirm this i'd appreciate.
Or i think i might have to change IC to MSP430F2618 or similar with USCI I2C.
And if I made wrong assumption, please correct me.
thank you
