We were facing drift in the horizontal axes when the quadrotor was rotated in the yaw axis. The EKF2_MAG_TYPE parameter was in automatic mode and when the flight logs were analyzed, it was observed that the drift is usually occured when the MAG_TYPE was automatically changed to 3-axis Magnetometer. When the mag type was forced to simple magnetic heading during the flight, sometimes the yaw performance was improved greatly while some other times we have faced with dangerous situation. The common characteristic of the dangerous situation was that the EKF rejects the height sensor input for sometime and then EKF GPS Fusion timeout and rng hgt timeout occurs. and this occurs continuously.
We know that forcing the EKF2_MAG_TYPE parameter to magnetic heading during the whole flight is a mistake however we still could not find the reason why using simple magnetic heading causes the healthy height and velocity sensor inputs to be rejected by the EKF.
Note: The problem was experienced on pixhawk 2.1 cube with 1.9.0 release.
Any insight on what caused this problem when the EKF2_MAG_TYPE was chosen as MAGNETIC HEADING would be appreciated.
Thanks in advance,