LSM303AHTR Unexpected output during windy conditions.

[victor.manske]

The LSM303AHTR sensor is housed in an enclosure and that vibration is limited due to the stability of the housing. During windy condition the sensor constantly transmit false signals even though there was little to no movement to the Accelerometer and no metal objects passing by the Magnetometer. 

First, we detect tampering using the z-axis accelerometer.  When upright, the z-axis acceleration is -9.8m/s^2; if inverted it is 9.8m/s^2.  We’re using the +/-2g setting, so basically +/- 50% of full scale from right side up to upside down.  If the driveway monitor is not vertical, the z-axis reading is reduced as the angle from vertical increased ( z-axis = 0 if sensor is horizontal).  I think we chose somewhere around 30 degrees or more as the tamper indication.  So, if it falls on the ground or someone picks it up without keeping it vertical, a tamper signal is sent. 

Wind False triggers:  We use changes in the magnetic field to detect vehicles.  Ferrous metals moving in or out of range cause changes in the magnetic field but so does moving the magnetic sensor.  To match the range of the old driveway sensor, the algorithms that process the magnetic field data are set to be very sensitive.  This also makes them very good at detecting if the sensor is moving.  To reduce false triggers, we use the x and y axis accelerometers to detect if the sensor is moving.  If motion is detected (most likely wind), we reduce the gain of the magnetometer while the sensor is moving and for a few seconds afterward.  So, if it is windy the vehicle detection range is reduced but still useful.  The trick is finding the best compromise between reduced range and fewer false triggers.  We made these adjustments using a version of the code that provides an oscilloscope-like view of the magnetic and acceleration data.   With the sensor mounted in our driveway on windy days we set the reduced sensitivity to avoid most false triggers.  

Would anyone know why we still constantly get false triggers during windy conditions?

[RhSilicon]

"Wind is air moving from areas of high atmospheric pressure to low pressure. Violent destructive winds, as well as gentle summer breezes, result from a complex interplay of different forces.

One of these forces results from a pressure gradient, or how fast pressure changes over distance. When pressure changes rapidly over a small distance, the pressure gradient force is large. Strong winds almost always result from large pressure gradients.

The greater the difference in pressure over a specific distance, the faster the air flows. Strong winds can also flow out from thunderstorms.

Since wind is air in motion, it has momentum. This momentum is transferred to the object the wind hits. Thus, the force of the wind can push objects by moving them or even knocking them over. Winds moving over and around objects can cause pressure changes around the object, which can also cause it to move."

https://wxguys.ssec.wisc.edu/2020/04/20/strong-winds/

[RhSilicon]

Perhaps one way to help detect false positives is to compare data from a second system exposed to the same conditions.