Error 0x0BAE0001 after PM_DEVICE_ACTION_RESUME

[Alexandre Desgagne]

Hi,

I'm currently using nRF Connect SDK v1.9.0 and developing on a nRF52832 onto a custom board. I'm using a BMI090L from Bosch connected over i2c.

I'm trying to suspend the i2c device after shutting down the gyro using the following code:

m_bmi090l.gyro_cfg.power = BMI090L_GYRO_PM_SUSPEND;

if ( bmi090lg_set_power_mode(&m_bmi090l) != BMI090L_OK ){
    LOG_ERR("Failed to shutdown gyroscope.\n");
}
pm_device_action_run(i2c_dev, PM_DEVICE_ACTION_SUSPEND);

 Then I resume it using this:

pm_device_action_run(i2c_dev, PM_DEVICE_ACTION_RESUME);

m_bmi090l.accel_cfg.power = BMI090L_ACCEL_PM_SUSPEND;

if (bmi090la_set_power_mode(&m_bmi090l) != BMI090L_OK){
    LOG_ERR("Failed to shutdown acceleration.\n");
}

After the resume, I'm unable to interact with the gyroscope but only with the accelerometer. I think it may come from somewhere around the pm_device_action_run function since the problem disappear when commenting those lines. I didn't have this issue with 1.7.0 too.

Here's the device tree part where I configure the device:

&i2c0 {
    compatible = "nordic,nrf-twi";
    status = "okay";
    clock-frequency = <I2C_BITRATE_STANDARD>;
    sda-pin = <25>;
    scl-pin = <27>;
};

Here's the I2C and PM part of my prj.conf file:

# I2C
CONFIG_I2C=y
CONFIG_I2C_NRFX=y
CONFIG_NRFX_TWI=y
CONFIG_NRFX_TWI0=y

# Power management
CONFIG_PM=y
CONFIG_PM_DEVICE=y
CONFIG_PM_DEVICE_RUNTIME=y

I know some changes have been done about power management between 1.7.0 and 1.9.0. Do you have any idea of what could cause this issue?

Thanks,

Alexandre

[Alexandre Desgagne]

Alexandre Desgagne said:

I'm using a BMI090L

I also know that gyroscope and accelerometer both use different addresses.

[Håkon Alseth]

Hi,

 

When placing the external sensor in suspend, it usually has a power up time. This seems to be 1 ms on the accelerometer, but it is much higher on the gyro. Could you try waiting as per the electrical specifications on this sensor before fetching the sample?

 

Kind regards,

Håkon

[Alexandre Desgagne]

Hi,

I checked and there's already a wait as per the electrical specifications. I'm using the bmi090l driver: https://github.com/BoschSensortec/BMI090L-Sensor-API.

I continued my examinations and I think the issue may comes from hardware side. The gyroscope answers a "NACK" when I try to wake it up only if uninit the i2c bus using this: pm_device_action_run(i2c_dev, PM_DEVICE_ACTION_SUSPEND); When I remove this instruction, It works properly. I suspect that the bmi090l has a strange behavior when I uninit i2c.

Thanks,

Alexandre

[Håkon Alseth]

Hi,

 

Do you use external pull-resistors? If not, could you try to add 10k pull-ups and see if this has any effect on the scenario?

 

Kind regards,

Håkon

[Alexandre Desgagne]

Hi,

We currently not having external pull-ups on our board and I think we never encountered this kind of issue because we never uninit the i2c. I will add a note to add it for the next board revision.

For now, I don't have any problem about not using this function pm_device_action_run(i2c_dev, PM_DEVICE_ACTION_SUSPEND);

I also think the lack of pull-ups resistors may be the cause of this issue.

Thanks,

Alexandre