I'm using the ADXL362 for activity and inactivity detection. Activity detection appears to trigger accurately at the correct threshold. However, the inactivity threshold appears to be inaccurate by ~25%.
A little more information about the configuration...
Mode: Linked Mode with activity and inactivity detection and referenced activity selected
Activity threshold = inactivity threshold = 50mg, 500mg (two separate tests)
Inactivity Time = 1 second
Noise setting = "ultra low"
My test setup is on a shaker table where I have very good control of the frequency and amplitude of vibration seen on the device. Inactivity occurs correctly when amplitude drops below about 20-25% of the set activity threshold after 1 second. It may go inactive after 10 or so seconds if within 10-20%, and does not go inactive if within 10% of the threshold.
Please let me know if you have any thoughts, or can confirm that others have seen similar behavior. Or if I can provide some more information about my setup.