With the increasing demand for legged robots, the importance of the joint drive is increasing. The dynamic performance of the inner-most torque/current control loop conditions the capabilities of the whole joint system. In this paper, a direct torque control based on a prediction model is proposed. The motor torque is estimated by considering calculation and measurement delay; error estimation and torque tracking error are observed and compensated. The control algorithm was implemented on a Field Programmable Gate Array (FPGA) board to apply the capabilities of concurrency calculation of the FPGA. The effectiveness of the proposed control algorithm was experimentally verified. Compared with the commonly used Field Oriented Control (FOC) current controller, the presented controller can not only improve the dynamic performance of the motor but also reduce the average switching times of the inverter.
Keywords: FPGA; PMSM; motor drive; torque predictive control.