Unstable wobble on auto mode

[IHA1 (Bükre)]

Hello,
This our 24 kg quadcopter without payload. I have made all calibrations except autotune. İt’s quite good tuned in stablize and althold modes. The problem is when we try take off or loiter modes it makes this wobbly movement. M9N GPS status are 28 sat. and status 4. I thought it’s about compass calibration according to datalog and made iron calibration and it got better but not enough. You will see in the video after the take off we changed to althold and it’s hanging in the air directly.

Main Hardware:
Orangecube+
M9N GPS
SIYI MK15
44.4V 22000mAH battery
T-Motor BLDC’s

https://drive.google.com/drive/folders/1jiV1EoyFOk9PPaZGzeKqmSVtkhs_IleO?usp=sharing

[amilcarlucas]

On such a big copter, did you tune it professionally? Did you bring a consultant to do it? How exactly did you tune it?

[IHA1 (Bükre)]

Thanks for the reply.
I applied this guide you posted step by step.
IMU temp calibration
IMU And GPS orientation
ESC Calibration
Initial Tune according to drone
Simple accel calibration
Compass iron calibration done after first guided take off
Hardware report

NOTCH_FILTER_SETUP done.
16_notch_filter_setup.param (4.9 KB)

After first flight test in stablize and althold modes it looked good. No hot motors.
Notch filter Calibration

updated param file for notch
17_notch_flter_results.param (338 Bytes)
updated throttle control file. Hover throttle appeared 0.4
18_throttle_controller.param (971 Bytes)

Last step for me was magfit. Before doing the script we wanted to take off in guided but it appeared we have problems in guided mod. I’m not sure what parameters to check when the problem is in only guided mode.

[amilcarlucas]

I’ll have a look on Tuesday at the office. I’m doing a workshop on tuning drones for professionals on the 16 of April. Want to join?

Methodically configure and tune ArduCopter for professionals - In presence Workshop at IAV GmbH

Join us for a hands-on workshop tailored for both UAV DACH members and non-members who are eager to gain practical knowledge in utilizing the ArduCopter framework as a multicopter control system. Explore the intricacies of unmanned aerial systems and enhance your skills in this interactive session. Don’t miss this opportunity to delve into the world of ArduCopter technology with us! Description ArduCopter offers a robust framework for managing multicopter drones, utilizing modern STM32 hardware…

I’ll also be presenting the ArduPilot methodic configurator GUI software:

New ArduPilot methodic configurator GUI

[App_screenshot1] We created a python GUI application to edit the intermediate parameter files and upload them to the flight controller. All GUI elements have meaningful tooltips. The latest documentation is embedded, It validate min and max ranges, but allows users to override them. It displays calibrations in yellow. If a reset is required because of a parameter change it automatically does it. It reads back the parameters after writing and automatically validates if the write was successfu…

[Jai.GAY]

@IHA1

The problem is when we try take off or loiter modes it makes this wobbly movement. M9N GPS status are 28 sat.

For big drone, these two steps helped me. But, before that, Autotune first and must sure is a low noise and low vibration drone frame.

[IHA1 (Bükre)]

Hello again. I can’t get a visa soon but thanks for the invite.

I have made a few changes and corrected some mistakes. There is definitly a problem with the compasses. I disabled orange cubes compass and fixed GPS Compass 45 degress (yaw270) cause magfit tool shows 45 degree even the gps looking forward and it got better.

Here is the filter graph

İn datalog rate.out values below 0.1
Vibrations below 15.
I’m not sure if ATT roll-des.roll values are okey.

Here is the last auto flights video, parameters and datalog.
I lowered waypoint speed for safety.
It’s still not good enough.
quadcoper - Google Drive

We are doing magfit calibration with script today. I will update here the results.

[IHA1 (Bükre)]

We suppose found the main problem. Inaccurate gps position causing the quad to going that direction and powerful motors reasoning overcorrecting itself.
I lowerd PSC_POSXY PID parameters to reduce that overreaction and it’s stable right now.
The problem to solve right now is 1m accurisy so travelling 1m radius around even we have 30 sattelites, 0.5 HDOP GPS and 16 sat 0.9 HDOP RTK FIX.