In the other words, we can say that we are performing PID control on the angular velocity, not the absolute angle. So our P term will be the reading from gyro, which is the angular velocity. I term is the integration of the gyro reading, which is the angle. and D term is the derrivation of the gyro output which is the angular acceleration. So the set point for our PID control would be angular velocity = 0. I think that's why my problem 1 exist. It's because if I move my stick forward, then the DF will tilt forward. then I center the stick and the DF will remain tilting forward because it will try to match the set point and refuse to level back....Danvel, is my explanation correct??


p/s : adding the D term which is the angular acceleration will help much on the algorithm. To minimize the noise, I sample the gyro reading 100 times and I use the average value.