Freefall in SITL

Hay, my first post! I’m trying to simulate a freefall scenario in SITL (using just mavproxy). What I do is basically climb to some altitude and then disarm the throttle. I pick up the raw acceleration values from RAW_IMU.zacc. I pick up the servos values from SERVO_OUTPUT_RAW. The thing is, while initially the z acceleration basically goes to zero (as it should, since all servos are set to 1000), zacc immediately starts dropping back to baseline value of -1000. Which is weird, considering that all the engines are cut. What am I missing here?

I just would like to add: I get that it reaches terminal velocity at some point, but it’s only something like 14 m/s, which seems a bit too low to be honest. How fast do these things actually fall, anyway? The real vehicle is an 8 kilo hexacopter.

I think you would expect to see something like what you have shown. In the hover the motors exactly counter the force of gravity, so the accelerometer experiences -1G (the difference between the vehicle acceleration and gravity). The instant the motors stop providing a force counter to gravity, the accelerometer will experience an acceleration of zero, as the vehicle will initially accelerate at 1G (1G + -1G = 0). The wind resistance will begin to build immediately, which will slow the vehicle acceleration progressively more the faster it goes. This will result in the accelerometer measured value decreasing from 0, back to -1G again when the vehicle reaches terminal velocity (acceleration=0).

Exactly how long this takes will depend on the terminal velocity of the vehicle, which could be lower than you expect, if the effect of windmilling propellers taken into account. I don’t know how accurate SITL models the aerodynamic drag on the vehicle; I imagine it will get complicated very quickly, depending on orientation and prop behaviour.

It would be interesting to attempt to stabilise a (falling) multi-rotor by using windmilling props to generate enough power for the flight controller, and ESC regenerative braking to control the attitude (it would probably be academic when it hit the ground).