twi_scanner only detects I2C devices in debug mode

[doryhealth]

Hello all

I copied the peripheral>twi_scanner example and adapted it to the nRF52833 chip on our custom board.

I have 4 chips (0x0b, 0x48, 0x52, 0x68) that use I2C on the custom board.
If I simply connect / erase / download the program, "no device" is found
If I debug (F5) the program and step through, the devices are found.

What am I missing?

The way I will go about it is the following:

Attempt 1) Compare debug and release configurations, adapt if needed
Attempt 2) Build the twi_scanner code step by step, see if any command is "board specific" and cannot be adapted from nRF52840/nRF52833 to our own board

Any other ideas?

Thank you, in advance, for your time and help.
Tell me if there is anything I can share (preprocessor directives, code, details about the custom board...)

Have a great Friday

Jerome

[Kenneth]

Hi,

Maybe look into the electrical configurations? E.g. check that boards and sensor have common GND and VDD, make sure there are external pull-up resistors on the CLK and DATA pins connected to VDD. Try to add a small delay in the start of main() to ensure that power have stabilized and that the sensors have time to power up. Double check that you are running the twi scanner without modifications, in case you have done some changes to sdk_config or other that may by accident have changed things.

Kenneth

[doryhealth]

I added a delay (100ms) after each "nrf_drv_twi_rx" call, to no avail
Let me state the problem in a more specific way: no devices are detected on debug (F5), either, unless I step through each instruction (F10). I'll try some more.

[doryhealth]

(Also, thank you Kenneth!)

[Kenneth]

I can't really think of anything, do you by any chance have access to a logic analyzer so you may look at the twi transfers for comparison?

Kenneth

[doryhealth]

Hi Kenneth

Thank you for your time and help

So essentially, the sensors are discovered but something else is wrong.

I used my picoscope 2000 and here's what I get (with UART/putty and without), both while debugging (F5) or uploading to the target.

#include <stdio.h>
#include "boards.h"
#include "app_util_platform.h"
#include "app_error.h"
#include "nrf_drv_twi.h"

#include "nrf_delay.h"

#include "nrf_log.h"
#include "nrf_log_ctrl.h"
#include "nrf_log_default_backends.h"

/* TWI instance ID. */
#if TWI0_ENABLED
#define TWI_INSTANCE_ID     0
#elif TWI1_ENABLED
#define TWI_INSTANCE_ID     1
#endif

 /* Number of possible TWI addresses. */
 #define TWI_ADDRESSES      127

/* TWI instance. */
static const nrf_drv_twi_t m_twi = NRF_DRV_TWI_INSTANCE(TWI_INSTANCE_ID);


/**
 * @brief TWI initialization.
 */
void twi_init (void)
{
    ret_code_t err_code;

    const nrf_drv_twi_config_t twi_config = {
       .scl                = DH_SCL,
       .sda                = DH_SDA,
       .frequency          = NRF_DRV_TWI_FREQ_100K,
       .interrupt_priority = APP_IRQ_PRIORITY_HIGH,
       .clear_bus_init     = true
    };

    err_code = nrf_drv_twi_init(&m_twi, &twi_config, NULL, NULL);
    APP_ERROR_CHECK(err_code);

    nrf_drv_twi_enable(&m_twi);
}


/**
 * @brief Function for main application entry.
 */
int main(void)
{
    ret_code_t err_code;
    uint8_t address;
    uint8_t sample_data;
    uint8_t old_sample_data;
    bool detected_device = false;

    APP_ERROR_CHECK(NRF_LOG_INIT(NULL));
    NRF_LOG_DEFAULT_BACKENDS_INIT();

    NRF_LOG_INFO("TWI scanner started.");
    twi_init();

    for (address = 1; address <= TWI_ADDRESSES; address++)
    {
        err_code = nrf_drv_twi_rx(&m_twi, address, &sample_data, sizeof(sample_data));
        nrf_delay_ms(100);
        NRF_LOG_INFO("Error code: %d", err_code);
        if (err_code == NRF_SUCCESS)
        {
            detected_device = true;
            NRF_LOG_INFO("TWI device detected at address 0x%x.", address);
        }
        old_sample_data = sample_data;
    }

    if (!detected_device)
    {
    NRF_LOG_INFO("No device was found.");
    }

    while (true)
    {
      /* Empty loop. */
    }
}

/** @} */

[Kenneth]

Are you sure that the nrf_delay_ms() works as expected when not debugging? Can it be that the compiler may optimize it away?

A read sequence typically is:

[START] device_address + read bit, device_register [re-START] device_address, sample_data [STOP]

Though in your case it seems you are consistenly receiving NRF_ERROR_DRV_TWI_ERR_ANACK, which means "NACK received after sending the address in polling mode".

So this is indeed strange.