ADXL362 Noise on Thingy 91

[hubert_11]

Hi,

I would like to use the Thingy 91 to count rotations making use of the accelerometer. So I wrote this function that seems to work to get the acceleration values from the ADXL362:

int ext_sensors_accel_lp_get(double *accel)
{
	int err;
	struct sensor_value data[ACCELEROMETER_CHANNELS];

	err = sensor_sample_fetch_chan(accel_sensor_lp.dev, SENSOR_CHAN_ALL);
	if (err) {
		LOG_ERR("sensor_sample_fetch_chan, error: %d", err);
		return -1; 
	}

	err = sensor_channel_get(accel_sensor_lp.dev, SENSOR_CHAN_ACCEL_XYZ, data);
	if (err) {
		LOG_ERR("sensor_channel_get, error: %d", err);
		return -1; 
	}

	accel[0] = sensor_value_to_double(&data[0]);
	accel[1] = sensor_value_to_double(&data[1]);
	accel[2] = sensor_value_to_double(&data[2]);

	//k_spinlock_key_t key = k_spin_lock(&(accel_sensor_lp.lock));
	//*accel = sensor_value_to_double(&data);
	//k_spin_unlock(&(accel_sensor_lp.lock), key);

	return 0;
}

However, the values this function returns vary a lot, more than I expect from reading the datasheet:

 0.068646    -0.558979  -8.620045
 0.039226    -0.019613  -8.953471
-0.284392    0.323619   -8.786758
 0.304006     0.225552  -9.855683
 0.235359    -0.941438  -6.982334
 0.774725    -1.029698  -10.012589
-0.470719   -0.274586   -8.933858
 0.608012    -0.980665  -10.208722

(The one higher valued axis does change like you would expect when rotating the Thingy 91)

I also tried changing the measurement rate to 12.5Hz:

# ADXL362 - Low-Power accelerometer used for activity and inactivity detection.
CONFIG_ADXL362=y
CONFIG_ADXL362_TRIGGER_GLOBAL_THREAD=y
CONFIG_ADXL362_INTERRUPT_MODE=1
CONFIG_ADXL362_ABS_REF_MODE=1
CONFIG_ADXL362_ACCEL_RANGE_2G=y
#CONFIG_ADXL362_ACCEL_ODR_400=y
CONFIG_ADXL362_ACCEL_ODR_12_5=y

But this did not improve the results. I'm I missing something or just reading the values in the wrong manner perhaps?

[Sigurd]

Hi,

However, the values this function returns vary a lot, more than I expect from reading the datasheet:

What value in the datasheet are you comparing it to?

[hubert_11]

I currently don't have the Thingy next to me, so just using the numbers already posted here, but converted to g from m/s^2:

0.007 -0.057 -0.878
0.004 -0.002 -0.913
-0.029 0.033 -0.895
0.031 0.023 -1.005
0.024 -0.096 -0.711
0.079 -0.105 -1.022
-0.048 -0.028 -0.912
0.062 -0.100 -1.042

If I then calculate the std for each axis:

Column X - mu: 0.2363, std: 0.2502
Column Y - mu: -0.4039, std: 0.4279
Column Z - mu: -8.8655, std: 0.8518

Than for example the Z axis has a much higher std than the "SELF TEST DELTA (mg)" chart shows in the datasheet, right?

[Sigurd]

Hi,

hubert_11 said:

Than for example the Z axis has a much higher std than the "SELF TEST DELTA (mg)" chart shows in the datasheet, right?

Yes, it seems a bit higher, while the X axis is lower. The chart is for supply voltage 2.0V, while on the Thingy:91 it's using 1.8V, so that might explain it. There are also some numbers in Table 22 in the datasheet. Maybe you could check with the manufacturer of the sensor, Analog Devices, https://www.analog.com/en/support/technical-support.html

[hubert_11]

Will do, but just to be sure, there is no inherent fault in the way that I'm requesting the sensor data?

[Sigurd]

hubert_11 said:

but just to be sure, there is no inherent fault in the way that I'm requesting the sensor data?

No, I don't see anything wrong with the code you have posted.