Why the power spectral density(PSD) of BMI088 Gyro(x/y/z) have strange peak point?

[Silva]

I sample the BMI088 sensor static data in the marble platform in 0.5 hour,  then calc the gyro(x/y/z) PSD, and found the strange peak point on x/y/z as the image below, what is the root cause?  btw, BMI088 acc PSD is normal.

bmi088 gyro PSD peak point issue(abnormal)

more test detail below for reference:

1.  BMI088 settings:   23Hz LPF， 200Hz.

2.  BMI160 settings and gyro PSD(normal,  for reference).

74.6 Hz LPF， 200Hz

BMI160 gyro PSD is normal

[FAE_CA1]

Hi Silva,

The gyro noise level is not related to full scale range. It is only correlative with the bandwidth setting. This means that the typical noise density of 0.014dps/rt(Hz) applies to all full scale ranges.

Thanks.

[dxglaw]

Hi FAE_CA1,

Thank you for your help.

For the noise amplitude, I will keep looking into it.

For the spikes in PSD, the data files are attached. To recover PSD at low frequency, I have to log the data for a long time. I attached part of my log files. From these files, you will be able to see spikes at low frequency when 230Hz LPF is chosen.

Thanks.

[dxglaw]

Hi FAE_CA1,

Today, I tested the noise level of Bosch shuttle board of BMI055. And you are right about that the noise level is not related to gyro range.

However, the noise amplitude is still very large (+/-10dps max). Please see the screenshot below. Do you have any clue?

Thank you.

 

[Vincent]

This PSD issue with big jitter found on x/y axis is confirmed from our engineer team and it is understood. 

The root cause is the digitial noise coming from the ADC module insdie our ASIC of sensor.  

[dxglaw]

Hi Vincent,

Thank you for confirmation. This PSD issue doesn't affect much. We can choose 523Hz LPF and apply our own filter.

However, do you have any clue about the noise issue in my previous post? According to the datasheet and FAE_CA1, the noise amplitude should be within +/-3dps. What I get is more than +/-10dps. This will affect the ARW (angle random walk) and cannot be improved by a LPF.