MPU9250 TWI Data Retrieve Crashing

[aplatts]

Hi

I'm have and NRF52840 with a MPU9250 connected via TWI. The connection to the MPU9250 is successful, I can find its address on the bus (0x68).

I've implemented a very simple couple of uint32_t structs to based upon other examples available:

Read from a TWI register:

uint32_t nrf_drv_mpu_read_registers(uint8_t reg, uint8_t * p_data, uint32_t length)
{
    uint32_t err_code;
    uint32_t timeout = MPU_TWI_TIMEOUT;

    err_code = nrf_drv_twi_tx(&m_twi, MPU9250_TWI_ADDR, &reg, 1, false);
    if(err_code != NRF_SUCCESS) return err_code;

    while((!twi_tx_done) && --timeout);
    if(!timeout) return NRF_ERROR_TIMEOUT;
    twi_tx_done = false;

    err_code = nrf_drv_twi_rx(&m_twi, MPU9250_TWI_ADDR, p_data, length);
    if(err_code != NRF_SUCCESS) return err_code;

    timeout = MPU_TWI_TIMEOUT;
    while((!twi_rx_done) && --timeout);
    if(!timeout) return NRF_ERROR_TIMEOUT;
    twi_rx_done = false;

    return err_code;
}

 

and read accelerometers values:

uint32_t app_mpu_read_accel(accel_values_t * accel_values)
{
    uint32_t err_code;
    uint8_t raw_values[6];
    err_code = nrf_drv_mpu_read_registers(MPU_REG_ACCEL_XOUT_H, raw_values, 6);
    if(err_code != NRF_SUCCESS) return err_code;

    // Reorganize read sensor values and put them into value struct
    uint8_t *data;
    data = (uint8_t*)accel_values;
    for(uint8_t i = 0; i<6; i++)
    {
        *data = raw_values[5-i];
        data++;
    }
    return NRF_SUCCESS;
}

I have the following definitions for the MPU:

#define MPU9250_TWI_ADDR 0x68

#define MPU_REG_ACCEL_XOUT_H       0x3B // ACCEL_XOUT[15:8]
#define MPU_REG_ACCEL_XOUT_L       0x3C // ACCEL_XOUT[7:0]
#define MPU_REG_ACCEL_YOUT_H       0x3D // ACCEL_YOUT[15:8]
#define MPU_REG_ACCEL_YOUT_L       0x3E // ACCEL_YOUT[7:0]
#define MPU_REG_ACCEL_ZOUT_H       0x3F // ACCEL_ZOUT[15:8]
#define MPU_REG_ACCEL_ZOUT_L       0x40 // ACCEL_ZOUT[7:0]

/**@brief Structure to hold acceleromter values. 
 * Sequence of z, y, and x is important to correspond with 
 * the sequence of which z, y, and x data are read from the sensor.
 * All values are unsigned 16 bit integers
*/
typedef struct
{
    int16_t z;
    int16_t y;
    int16_t x;
}accel_values_t;

I'm running a call once and writing to NRF_LOG to test within the main loop, as follows:

    // Read accelerometer sensor values
    uint32_t err_code;
    accel_values_t acc_values;
    err_code = app_mpu_read_accel(&acc_values);
    APP_ERROR_CHECK(err_code);
    NRF_LOG_INFO("Accel %d\r\nX: %06d\r\nY: %06d\r\nZ: %06d", acc_values.x, acc_values.y, acc_values.z);
    NRF_LOG_FLUSH();

My NRF_LOG Output is:

<info> app: TWI started.

<info> app: TWI Scanner Started.

<info> app: SparkFun MPU-9250 detected at 0x68

<info> app: TWI Scanner Finished.

<info> app: MPU Sensor Connected

<error> app: Fatal error

<warning> app: System reset

Is anyone able to advise. Thanks in advance.

[Vidar Berg]

Hi,

Are you using Segger Embedded Studio? In that case, please select the "Debug" target from the drop-down menu on the top left side before you start debugging. This will make the crash log more verbose. Instead of just "Fatal error" it will include the file name and line number of where the error occurred.

[aplatts]

I’m using VSC, can I get a similar output with NRF_LOG?

[Vidar Berg]

Yes, if you add the "DEBUG" flag to your list of preprocessor symbols ( ie add -DDEBUG)

[aplatts]

Thanks, newbie ish. How do I do that?

[Vidar Berg]

Sorry. Are you using Makefiles from the SDK to compile? In that case, add this line "CFLAGS += -DDEBUG" together with the other C flags.

[aplatts]

Great, set that an now I get a much better view of the issue (but still don't understand):

New -DDEBUG Output:

<info> app: Read accelerometer sensor values

<error> app: ERROR 13 [NRF_ERROR_TIMEOUT] at ../../../main.c:741

PC at: 0x0002B413

In my code line 741 is line 31 in the code snippet below:

int main(void)
{
    // Initialize.
    log_init();
    leds_init();
    timers_init();
    buttons_init();
    power_management_init();
    twi_init();
    ble_stack_init();
    gap_params_init();
    gatt_init();
    services_init();
    advertising_init();
    conn_params_init();

    // Start execution.
    twiScanning();
    advertising_start();

    NRF_LOG_INFO("StingRay Started.");

    NRF_LOG_FLUSH();

    if (MPU9250) {
        // Read accelerometer sensor values
        NRF_LOG_INFO("Read accelerometer sensor values");
        uint32_t err_code;
        accel_values_t acc_values;
        err_code = app_mpu_read_accel(&acc_values);
        APP_ERROR_CHECK(err_code);
        NRF_LOG_INFO("Accel %d\r\nX: %06d\r\nY: %06d\r\nZ: %06d", acc_values.x, acc_values.y, acc_values.z);
        NRF_LOG_FLUSH();
    }

    // Enter main loop.
    for (;;)
    {
        idle_state_handle();
    }
}

[Vidar Berg]

Thanks, so it looks like you are getting the NRF_ERROR_TIMEOUT from nrf_drv_mpu_read_registers(). Maybe you can add some more logging in your TWI callback and see if you get the NRF_DRV_TWI_EVT_ADDRESS_NACK or NRF_DRV_TWI_EVT_DATA_NACK event instead of the expected NRF_DRV_TWI_EVT_DONE. 

[aplatts]

Hi Vidar

You are correct, I had ported a library which was using other legacy library's which was causing the issue. It has been a big learning curve, thanks for your excellent support.