FPGA Based Sensorless Sliding Mode Observer with Phase-Locked Loop for PMSM Drives: Design, Verification and Implementation

This paper presents the design and implementation of a sensorless permanent magnet synchronous motor (PMSM) drive on a field-programmable gate array (FPGA) using a sliding mode observer (SMO) and a phase-locked loop (PLL) for accurate and robust speed/position tracking. To eliminate mechanical position sensors while reducing cost and improving reliability in harsh environments, a chattering-mitigated SMO employing a hyperbolic tangent switching function is developed and integrated with a PLL-based position/speed extraction scheme. Practical FPGA feasibility is emphasized by adopting fixed-point arithmetic and finite-state-machine (FSM) scheduling to achieve deterministic real-time execution and efficient resource sharing, without excessive algorithmic complexity. The proposed scheme is validated through both simulation and real-time experiments on an FPGA platform. Quantitative evaluation under identical conditions shows that, in sensorless operation, the proposed SMO improves tracking accuracy to MAE = 4.5 rpm and RMSE=6.24rpm,comparedwiththe sign-based SMO (MAE=9.4rpm, RMSE = 13.9 rpm) and the saturation+arctangent SMO (MAE = 7.0 rpm, RMSE =9.7 rpm), corresponding to reductions of 52.1% (MAE) and 55.1% (RMSE) versus the sign-based baseline. In transient speed reversal, the rising time is reduced from approximately 0.2 s to 0.16 s. Rotor flux-angle estimation error is reduced from about 6?–9? to 2?–4?, with phase delay improved relative to > 25 ms observed in conventional SMO during rapid variations. The FPGA implementation completes each SMO–PLL update in approximately 8.16 µs at 12.5 MHz and utilizes 12,568 LEs, 230,656 RAM bits, and 26 embedded multipliers on MAX10. Importantly, the improved estimation and control performance is achieved with only a modest increase in hardware resources compared with a traditional SMO (an increase of approximately 4% in LEs and approximately 5% in embedded multipliers), reflecting a clear and practical accuracy–resource trade-off. These results demonstrate anefficient, low-cost, andreliable FPGA-basedsensorlessPMSM drive suitable for industrial servo, robotics, and automation applications with diverse speed profiles and transient disturbances.