BMI270 gyro noise

[Anatol]

Hello. I experience a very high level of noise from BMI270 gyro. My settings are:

0x7C PWR_CONF: 0x03; > low power mode;

0x42 GYR_CONF : 0x27 > ODR= 50 Hz;

0x43 GYR_RANGE: 0x04 > +/- 125 dps, 262 LSB/dps.

While still, I get rms noise about 50 LSB. It gives 0.2 dps - correct? Seems high but my application can tolerate it.

But when I move the bench with IMU fastened on it slowly over the table (without noticeable rotation) I get noise peaks about 1000...1500 LSB = 4...5 dps, it makes the data useless. My first idea was it is high frequency vibrations from friction between the table and my improvised bench, but when I mounted IMU on a dampfer (0.5 kg weight on 2cm elastic foam) - nothing changed.

And, near the same amount of noise when I tap by fingers on the table near the standstill IMU. But in this case there is definitely no vibration about 4 dps amplitude.

I tried OSR4 mode - seems even worse.

I'm not very experienced in those gyro/ accelerometer matters - do I do/understand something wrong?

The gyro data is needed mainly to calculate the tilt over the short time, to distinct acceleration from the gravity. 

I'm completely clueless what can cause such a noise, any ideas will be very appreciated.

[Anatol]

Some update. I tried MPU6050 at the similar conditions, results approx the same. Is it normal that gyro noise is about 3 dps even by slightest hint of vibration?

[BSTRobin]

Hello Anatol,

To test the noise, the sensor needs to be in a stable state. Noise cannot be evaluated in a moving state.
According to your sensor configuration, the noise I measured is as follows:
x: 135 mdps
y: 110 mdps
z: 146 mdps

0x7C PWR_CONF: 0x03; > low power mode;
0x42 GYR_CONF : 0x27 > ODR= 50 Hz;
0x43 GYR_RANGE: 0x04 > +/- 125 dps, 262 LSB/dps.

[Anatol]

Hello, BSTRobin   thanks for reply and for data - as in datasheet there is no data for that mode. It corresponds good to my results for the motionless state.

I'm aware "noise" is not a right definition to my issue, just didn't know how to name it. I made some experimenting, I see now my problem is really vibrations from friction while moving, my improvised damper was not good enough to suppress it (has a resonance frequency 11 Hz while the spectrum of noise is 12...24  Hz. Spectrum to measure is like 1...2 Hz). I think I have to find a mechanical solution (a better damper), low pass filters would not help much as calculating the angular position from speed I make integration already for 330 mseconds, still errors are too big.

My goal is to have about 0.2 deg precision after moving during couple of seconds at about 20 dps speed, hope it is realistic?

[BSTRobin]

Hello Anatol,

If your working environment has strong vibration, you can consider using BMI088 with high vibration robustness.

For BMI270, you can run CRT and configure the sensor to high performance mode, and then look at your test results.