2 I2C Writes are sent instead of 1

[morpho]

I am using TWI0 for MCP9808 and I notice that a single Write operation causes two Writes (one without a stop condition while the 2nd one with a stop condition)

void handler(nrf_drv_twi_evt_t const * p_event, void * p_context)	   
{
    Sensor* sensor = static_cast<Sensor*>(p_context);
    
    switch (p_event->type)
    {
      case NRF_DRV_TWI_EVT_DONE:
        if (p_event->xfer_desc.type == NRF_DRV_TWI_XFER_TX)
        {
          m_xfer_done = true;	
        }
	    break;

        default:
	  break;
    }
}

void Sensor::Write(uint8_t reg, uint8_t size)
{
    ret_code_t error_code = nrf_drv_twi_tx(&m_twi, SENSOR_ADDR, &reg, size, false);
    APP_ERROR_CHECK(error_code);
    while(!m_xfer_done);	// set by the ISR
}

void Sensor::Start()
{
    i2cConfig = 
    {
       .scl                       = TWI_SCLK,
       .sda                       = TWI_SDA,
       .frequency                 = static_cast<nrf_drv_twi_frequency_t> (TWI_FREQ),	  
       .interrupt_priority        = TWI_IRQ_PRIORITY,
       .clear_bus_init            = false
    };
    
   ret_code_t err_code = nrf_drv_twi_init(&m_twi, &mI2cConfig, handler, this);
   APP_ERROR_CHECK(err_code);
   nrf_drv_twi_enable(&m_twi);	
   
   Write(0x5, 1); // send the register address that we'd be interfacing with
}

// main.cpp
Sensor sensor;

int main()
{
   sensor.Start();
   while(1);
}

[Vidar Berg]

Yes, a valid I2C transaction must start with an address byte to address the slave you want to communicate with on the bus.

[morpho]

Sorry but that didn't address the concern here. Why could I be seeing two writes instead of one?
It should just be <0x30 + ACK +0x05 + ACK> followed by a stop condition, no?

[Vidar Berg]

I was confused because I was only focusing on the last two images you uploaded. Both show a single write transaction with the address byte and the 1-byte payload

From the first image, it appears there is a reset between the transactions, which causes sensor.Start(); to be executed twice. I recommend placing a breakpoint in main() to verify this.

[morpho]

When I run in debug mode, I see a single write but when I Run the application, I see twice the writes. :/

[Vidar Berg]

Then, it would be helpful to know how you are testing this. Do you observe the double transaction after a power cycle? You may also consider adding a delay at the beginning of the main function, such as nrf_delay_ms(1000), to make it more clear if the program is re-entering the main.