Gimbal Lock smooth transition not working properly in 2.70
  • In my gimbal with motor order Cam-Roll-Pitch-Yaw newest 2.70 firmware Gimbal Lock smooth transition not working properly - have a large vibration in action near gimbal lock position Roll and Yaw axis (Cam see vertical). Checkbox "Gimbal Lock smooth transition" is activated.

    It can be fixed?
  • Hello,

    The parameter name "Smooth transition" may be a bit confusing, because it relates only to the IMU calculations, not stabilization. When you enable it, IMU does not go crazy in gimbal lock position. Another problem is that stabilization suffers when two motors become close to parallel - their interference significantly increased. In this position only adjusting PID values may help.

    A good solution may be installing 2nd frame IMU in the position "Below outer motor" (YAW in your case). It starts working in the gimbal lock position and improves the system response. W/out 2nd IMU, encoders are used as a closed-loop signal source, which makes it less effective than the normal gyro-based closed loop.
  • I have 2nd IMU below YAW
  • today I make a downgrade to 2.56 firmware and all is working done and properly, all motor and axis i correct workin in gimbal lock mode without artefacts. maybe Alex can fix this problem in next firmware?
  • 2.71 version have this bug too
  • Hi Alex, Is there a downside to having the '"Gimbal Lock" smooth transition' button checked all the time? It seems to quiet down my errand Y-axis "MOT_Y_OUT" output, so I'd like to just leave it checked for all my robots but I'm wondering if it causes any type of error accumulation elsewhere as my systems stay on for days at a time.
  • Hi! This parameter doesn't have any downside and is reccomended to be enabled. It starts working when PITCH angle approaches +-60-70 degrees, where the attitude expressed in Euler angles, becomes unstable.
    If you have vibrations at a certain position (what I have guessed after checking your video showing RMS error plot), it may relate to PID tuning. The solution would be adjusting PIDs in this problematic position. Using the GUI's "optimize" tool you can do it automatically, by capturing system response in two or more positions and optimizing PID parameters for them all at once (i,e, load several responses for one axis and run tool for them, with the "Do for all loaded data files" option checked).