* increase filter bandwith to make it sync faster * use transport_frame instead of audible_frame for calculating the loop error (thanks robin!) * reduce the deadzone to 0.1% (thanks robin)