strange problem with lsm6dsv320x HG accelerometer

[richardp]

Hi

We are testing out the lsm6dsv320x, which will be used on our new board.

The low G accelerometer is working fine, however the high G looks to be reporting wrong values.

If I use a simple calculation to computer the angles, the high G  is off by about 3-4 degrees

This is the formula i use to calculate the angles

 AccXangle = (math.atan2(ACCy,ACCz)*RAD_TO_DEG)
 AccYangle = (math.atan2(ACCz,ACCx)+M_PI)*RAD_TO_DEG

Below is an example of the results of both formulas when the sensor is horizontal. The last two columns are for the high G sensor.  you can see that it is different from the low G sensor. My expectation is that they should be reporting the same result.

ACCX_angle	ACCY_angle		ACCX_HG_angle	ACCY_HG_angle
-0.06	-0.20		-4.44	-4.620
0.03	-0.17		-3.67	-4.120
0.03	-0.23		-3.84	-3.610
-0.03	-0.17		-3.61	-4.240
-0.03	-0.17		-2.91	-5.130
0.03	-0.20		-3.50	-4.240
0.00	-0.20		-4.47	-3.080
0.00	-0.20		-3.05	-5.680
0.03	-0.20		-4.15	-4.380
-0.03	-0.20		-3.71	-3.250
-0.03	-0.20		-3.96	-4.070
0.00	-0.20		-2.97	-4.880
0.00	-0.23		-2.94	-3.790
-0.03	-0.17		-2.74	-3.480
0.00	-0.20		-3.70	-3.760
0.00	-0.20		-3.99	-3.690
-0.03	-0.23		-4.56	-5.360
0.03	-0.20		-2.46	-3.690
0.03	-0.23		-3.52	-4.790
0.00	-0.20		-3.15	-4.010
0.00	-0.17		-2.87	-2.350
-0.03	-0.20		-3.75	-3.750

 

This is how we initialise  the sensor

 writeByte(LSM6DSV320X_ADDRESS,LSM6DSV320X_CTRL3, 0b01000000) # Block data update
 writeByte(LSM6DSV320X_ADDRESS,LSM6DSV320X_CTRL1, 0b00000101) # High performance mode, 60Hz
 writeByte(LSM6DSV320X_ADDRESS,LSM6DSV320X_CTRL1_XL_HG, 0b10011000) #Enable high G. 480Hz ODR 32G
 writeByte(LSM6DSV320X_ADDRESS,LSM6DSV320X_CTRL8, 0b00000011) # 16G
 writeByte(LSM6DSV320X_ADDRESS,LSM6DSV320X_CTRL2,0b00011010) #1.92KHz, high-accuracy ODR mode
 writeByte(LSM6DSV320X_ADDRESS,LSM6DSV320X_CTRL6, 0b00000100) # 2000dps

 

[Federica Bossi]

Hi @richardp ,

The low-G and high-G accelerometers inside the LSM6DSV320X are separate sensing elements optimized for different ranges and sensitivities.

They have different offsets, scale factors, and noise characteristics that cause slight discrepancies in angle calculations.

I suggest you to use low G accelerometer for angle calculation.