Fast and stable response is essential for dynamic robots due to their rapid change in state of motion and environments. To achieve such high speed, such robotic system requires high control bandwidth. Moreover, the bandwidth of current controller must be even larger than that of other controllers since current controller places in the innermost of the general cascade control loop. However, the measurement of current inevitably includes noise, limiting the bandwidth of controller. In this study, Kalman Filter based current controller for high control bandwidth is proposed. For implementation of Kalman Filter in discrete time domain, PAA is used for discrete-time system identification. High control bandwidth is achieved by filtering noise without time delay through Kalman Filter. High control bandwidth is verified through deriving frequency response of the implemented closed-loop control system with sine-by-sine system identification method.