Condition monitoring using LSM6DSO software setup suggestions.

[MBond]

I'm using the LSM6DSO sensor in a condition monitoring situation. I've compared it against an industrialised analogue sensor in a repeatable testing situation.

The LSM6DSO registered a pk reading 0.69 in the X axis and the industrial sensor had a reading of 0.5 in the X axis. This is a pretty significant difference in results, I was wondering if anyone had any general initialisation steps that might make the results a little better.

For reference the sensor is currently setup using FIFO mode high performance enabled with just the high pass filter enabled at ODR_DIV_200. The gyroscope has been disabled.

[niccolò]

Hi @MBond​ ,

I don't really know from where this difference may come from, would you mind better explaining how the test was set up, to help me in helping you?

Are the other axis aligned with your reference sensor?

Can you confirm that your values are [g]? (so the difference between the reference sensor and the LSM6DSO is 0.19g, or 190mg)

Is the peak value a one time only occurrence or does it reiterate with a certain frequency?

thank you in advance.

[MBond]

Hi @niccolo.ruffini​ 

The test was performed using a shaker table that was shaking at approximately 14hz with a backwards and forwards motion of 1mm. Due to the positioning of the MEM inside our sensor housing for the LSM6DSO the LSM6DSO's Z axis aligned with the X axis of the industrial sensors the Y axes are aligned with each other and the X axis of the LSM6DSO is aligned with the Z of the industrial sensor. My initial post I said that 0.69G was detected in the X axis however in reality the reading was detected in the Z axis.

I can confirm that our readings definitely in G, and the peak value is consistent throughout the waveform.

[niccolò]

Hi @MBond​ ,

I am curious about your reference sensor, what device is it?

anyway, I have 2 tests that you can perform to check if the situation improves:

• set the BDU bit (reg CTRL3_C (12h)) to be sure that there are no strange changes while reading the data
• try to disable the HP filter to check if the problem is there

Then, is the reference using the same parameters as the LSM6DSO?

I'm sorry if I'm asking you so many questions, but I never faced your type of problem, so I'm trying to find where the problem could be.

[MBond]

No need to apologise, I appreciate the help and I'm sure there is still details that I've left out that might be helpful to you so ask away if you are after more information.

I am not sure of the exact model of the sensor that we have been comparing against but it is an industrialised Piezo electric sensor and accompanying software manufactured by SKF. Id expect the noise level on the SKF sensor to be significantly less the LSM6DSO but I wouldn't expect it to account for such a large difference in readings.

I am enabling the BDU bit as part of the initialisation steps of the sensor. I am using the C libraries provided on the ST github so just passing an enable to the function block_data_update_set. I will do another test with none of the filters in the filtering chain enabled and I will report back the results.