I’ve seen this a few times. Thought I’d put up a log and see if anybody has any ideas. This is usually a catastrophic failure. I lucked out and the airplane recovered.
As expected Roll and Desired Roll separate. The airplane is uncontrollable. Note Roll and Desired Roll meet back at line 127. This where the airplane flips back up-right and gracefully lands.
This is typical behavior for a vibration event, right? But the throttle is off. The airplane is gliding.
Here is RC3 out. I killed power at line 114, well before the vibration event at line 123. The airplane is gliding. Brake is set on the ESC so the prop is not even spinning. But still the airplane goes into a typical vibration death spiral.
Here’s an idea: Could the “vibration spike” actually be acceleration due to gravity? We’d expect a major sign change if the plane rolls onto its back (so that gravity is accelerating the plane towards its roof) and that’s what I observe.
I don’t have time to look at the log myself right now, but I’m intrigued, so I’ll look at it later, unless someone else can spot the issue first?
(Great documenting makes the problem easy to understand! Thanks for that)
This looks like a stall to me. The undesired roll happens at a point in the flight where the airspeed is low, and happens just as up elevator is applied which will tend to initiate a stall.
I agree with @hunt0r, the “vibration” is probably the acceleration from the aircraft flipping and tumbling. I don’t think vibration is an issue here.
Stall speed would be pretty high, though. Is this a heavily loaded plane?
I’m not an expert but from your logs it appears that the problems start when in FBWA mode you ask some rapid changes to roll attitude, and your plane seems to be quite “lazy” to react.
In Automode looking at DesRoll and Roll , changes in DesRoll are very small and the plane is able to follow while in FBWA changes are more wide and rapid and the plane has troubles to follow.
Just before the event , you were making a turn, the red line of DesRoll has a big change and the green line tried to follow badly, immediately after another big change in DesRol in the opposite direction and the plane stalled even if the speed was always good.
Vibrations appears after the stall.
I think you have wrong PID and suggest to run an Autotune again .
Thanks for the replies. Yes, there is indeed a stall component to it. I went back out and duplicated the stall over and over. The issue appears to be - once the airplane stalls, and spins, the Pixhawk gets negative Gs and it can’t cope. Here are a few more logs:
The same results happen in STABILIZE as well. Once Z values go, and stay positive the airplane is uncontrollable. The rolls were fine while in STABILIZE.
