TWI driver for register-less chips with Thingy: 52

[Noon]

Helllo,

i'm using a custom shield with the Thingy:52.

I need to control some TWI chips, but these chips (ADG728) have only one register.

I have done a specific driver for these chips, based on the drv_mpu9250, and using nrf_drv_twi_tx function.

It works well, but, as the chip contain just one register, i change the 3rd parameter of the function nrf_drv_twi_tx to NULL for reding the register contents.

TWI transfer is good, but the driver return the error code INTERNAL_ERROR. 

I want to know how i can't dig deeper into the driver layers to modify the ERROR_CODE to fit my specific "no-register" case.

Thanks,

Arnaud

[Jared]

Hi,

 Which function returns the error code, nrf_drv_twi_tx ? How does SCL and SDA look like when the error is returned? The INTERNAL_ERROR code is a generic error that can be returned in more than one scenario. Modifying the error handler to ignore that error is therefore not optimal. Could you post some code?

 regards

Jared

[Noon]

This is the function inside drv_ADG728.c (based on the drv_mpu9250), where the error code is returned (from nrf_drv_twi_tx):

int drv_ADG728_read(unsigned char slave_addr, unsigned char * data)
{
    uint32_t err_code;

    err_code = twi_open();
    APP_ERROR_CHECK(err_code);

    err_code = nrf_drv_twi_tx( m_ADG728.init.p_twi_instance,
                               slave_addr,
                               NULL,//&reg_addr,
                               0,
                               true);
    if (err_code != NRF_SUCCESS)
    {
        NRF_LOG_ERROR("drv_ADG728_read Failed! = %d\r\n", err_code);
    }

    err_code = nrf_drv_twi_rx( m_ADG728.init.p_twi_instance,
                               slave_addr,
                               data,
                               1 );
    if (err_code != NRF_SUCCESS)
    {
        NRF_LOG_ERROR("drv_ADG728_read Failed!\r\n");
    }

    err_code = twi_close();
    APP_ERROR_CHECK(err_code);

    return 0;
}

I use it in a intialisation function, in shield_manager_init():

static void shield_init(void)
{
    uint32_t            err_code;
    drv_ADG728_init_t  ADG728_init_params;
    drv_AD5245_init_t  AD5245_init_params;
  
    static const nrf_drv_twi_config_t twi_config =
    {
        .scl = TWI_SCL_EXT,
        .sda = TWI_SDA_EXT,
        .frequency          = NRF_TWI_FREQ_400K,
        .interrupt_priority = APP_IRQ_PRIORITY_LOW,  //APP_IRQ_PRIORITY_LOW 
    };
  
    ADG728_init_params.p_twi_instance = &m_twi_sensors;
    ADG728_init_params.p_twi_cfg = &twi_config;

    err_code = drv_ADG728_init(&ADG728_init_params);
    APP_ERROR_CHECK(err_code);

    AD5245_init_params.p_twi_instance = &m_twi_sensors;
    AD5245_init_params.p_twi_cfg = &twi_config;

    err_code = drv_AD5245_init(&AD5245_init_params);
    APP_ERROR_CHECK(err_code);

    err_code = shield_manager_init();
    APP_ERROR_CHECK(err_code);

}

SCL and SCA look great, and i can read the data previously written onto the ADG728 register...

It's like there is an error, but i don't see it, everithing works fine...

[Jared]

Hi,

 I'm not sure if it's necessary to do a transmit sequence before you do a read based on the read sequence from the datasheet:

It seems that you assume that it's necessary to do a transmit sequence first? Maybe there is something I've missed?

[Noon]

I think you're right and missed nothing. I'll try it out and confirm later.

Thanks

[Noon]

I confirm you where perfectly right!

Thanks