Clarification on g-Range Variation in Drop Tests (ODR Impact) LSM6DSV320X kin Sensor tile box pro

[Ahilesh_Krishna]

Hi Team,

Greetings!

Mems studio is used and LSM6DSV320X is kept on the sensor tile box pro,

We are currently performing drop tests using the SensorTile Box pro platform mounted on a flat board casing. The tests are conducted from the same height under two configurations:

Wired mode with an ODR of ~7.6 kHz

Wireless mode with an ODR limited to ~1 kHz

While analyzing the accelerometer (high-g) data, we observe a variation in peak acceleration values of approximately ~20 g between different runs, even though the drop conditions are kept identical.

We would like to understand the following:

Is such a variation (~20 g) in peak acceleration expected under these conditions?

Could the difference in ODR (7.6 kHz vs 1 kHz) significantly impact the captured peak values for short-duration impact events?

Are there any known limitations in wireless mode (e.g., filtering, bandwidth constraints, or data handling) that could lead to underestimation of peak g values? and Free Fall interrupts is not observed eventhough free fall interrupts are enabled in wired connection

Do you recommend any best practices for accurately capturing high-impact events when comparing wired and wireless configurations?

From our initial understanding, the lower ODR in wireless mode may miss sharp impact peaks due to reduced temporal resolution, but we would appreciate your confirmation and any additional insights.

Looking forward to your guidance.

Thanks, and regards,

Ahilesh

[Smithson]

Do FFT's and compare them, that will tell you a lot.  

1000 Hz only gives you 500 Hz of FFT data you are asking a lot of 500 Hz to describe your signal.  

I would expect a difference. 

[Federica Bossi]

Hi @Ahilesh_Krishna ,

Yes, lower ODR means that it's more likely that an instantaneous event such as a high g shock is not captured at its peak. This is because with a lower ODR, the peak of the shock may occur between two samples, so the maximum amplitude of the impact can be underestimated. To achieve a higher accuracy when measuring shocks, a higher ODR is strongly suggested, as well as a FS high enough to avoid saturation.

However, the variation you observe is not necessarily caused by ODR alone. In drop tests, peak acceleration can also vary due to small differences in impact angle, mounting conditions, mechanical coupling, board/casing response and surface repeatability. What setup are you using for the drop tests and is the shock profile in each drop repeatable? Is the 20g difference constant between wired and non wired drops? How many shocks have you performed?

Now, for the free-fall interrupts, can you describe in more detail what firmware and configuration are you using? Since you mentioned MEMS Studio is used, could you please share the free-fall threshold and duration settings, how the interrupt is routed, and how you are checking the event? Also, are you using only MEMS Studio, or a standard/custom firmware as well? What exactly is your setup?

Please share the above information with us, and we will be happy to support you accordingly!

[Smithson]

Hello @Federica Bossi :

I read your note on the drop acceleration, before I travelled to London, so I have thought on it for two weeks.  In the period 1950 to 1980, Professor Hall and a few of his mates, using the term in the Australian Convention, looked in great detail at pulses such as drops, often for nuclear power plants.  Newmark and Hall wrote the standard for earthquake engineering for Nuclear Power Plants in the 1960's, date is from memory.  

Hall and I were talking in the early 2000's at University of Illinois at Urbana Champaign about this work and in particularly one book that had been widely used for pulses.  He noted that it contained an error in the math about pulses, I thought it contained a few conceptual errors in engineering math, but one does not argue with a retired professor, ok, one does, but quietly.  

Hall had done a lot of court work as have I, we talked about drop pulses and the acceleration levels and the actual impact.  

You noted that 

"This is because with a lower ODR, the peak of the shock may occur between two samples, so the maximum amplitude of the impact can be underestimated."

Statistically speaking the peak is more likely to occur between the spaces in the data stream as they are much greater time than the measurement points.  You see the same thing in FFT results, where two peaks are side by side in frequency, the real peak is most likely between them.   

If you are measuring a drop, the likely range is under 1000, more likely under 100.  There is some good stuff on this on the web from people who do this for a living.  

But we are free to make decisions.  

We do not like drops, we like nice long sunny days full of thermal energy that moves our structures nicely, if you are using an impact to measure building response, then it is an interesting decision, but it only adds unnecessary complexity to the FFT. 

Regards

John