This study is on the development of an Extended Kalman filter (EKF) used in conjunction with a simple PD+ controller for the position control of a direct drive system (DDS). The DDS consists of, specially, a single link arm driven by a permanent magnet synchronous motor (PMSM) and controlled by a DSP32 based motion control unit. The EKF is used to estimate the angular position and velocity, beside the total inertia of the system, which is used to calculate the load information for the PD+ controller. The estimation algorithm is based on the Bayes method known for its high accuracy. Precision being the main objective of the study, the control input error usually ignored in past literature is also take into account in the filter design. The control method thus developed demonstrates an improved performance over the classical PID in terms of accuracy and robustness through simulations and experiments that were carried out.