I am working on a very big octocopter project.
The propeller size is 40.5 inch
We did the first flight today. I was about to initiate autotune but I noticed that motor noise was increasing and decreasing periodically. Later I looked at the log, half of the motors were speeding up and speeding down at approx 0.25Hz frequency which matched with slight yaw variations i was experiencing while flying.
I’ve changed Yaw rat P and Ang P multiple times in the flight to see if it makes any difference, which it didn’t
Link for the log : Google Drive Link
The Yaw problem: motor outputs indicate a definite physical yaw problem like twisted motor mounts or arms. Possibly even the whole frame twists when operating. Tuning with parameters wont really help, only mask the problem, so we wont go there yet.
Counter-clockwise motors are working harder to counteract the problem with clockwise motors.
Fixing this may also help with the FFT being a bit confused - you may be able to continue using the FFT if the yaw bias is fixed.
There is also a manual process for yaw tuning that could be appropriate for this beast - see how yaw performs after the physical fixes.
I would enable the Harmonic Notch Filter as follows
INS_HNTCH_ENABLE,1 ← set this then refresh params to see the rest
The FFT picked up on the giant peak at 90Hz but missed what is probably the primary source of vibrations as 40Hz.
I would probably disable the FFT unless you fine tune the FFT and change the HNOTCH to use it
You’ve got some dangerous looking oscillations right before landing on each flight. Is this when landing gear makes contact?
If not, then apart from better PID tuning, it could be because of MOT_THST_EXPO.
The general rule is
- set too high you can see instability at low throttle
- set too low you can see instability at high throttle
So you could potentially try lower values of MOT_THST_EXPO like 0.6 but it might not be advisable to experiment too much with such large expensive/dangerous props.
It might be worth your time and effort to set up one motor and prop on a thrust stand and actually calculate the MOT_THST_EXPO properly.
Maybe a picture of the airframe would be nice,
Also always interesting to see others build!