mpu9250

. im calibrating the bias for the acc the giro and the mag . every initialization i get results just a little bit different. when i run it through the mahoney function i get a slow drift on the pitch and the yaw.
in order to try and compensate the drift i try to manually add/subtract small values from the calculated sensors acc and gyro. that doesnt help.
i can see that deltat in mahoney algorithm has an effect on the frequency of the drift.
has anyone encounter this phenomenon and can suggest a solution ?

Wow.

...R