Sensorless Control for PMSM Using FPESO and Improved PLL in the Estimated Coordinate

Sensorless control systems based on position error are widely used across both zero-low speed and medium-high speed ranges owing to their inherent closed-loop characteristics that provide strong anti-interference. Such estimation systems are generally implemented in the estimated rotational reference frame (γδ frame). During acceleration and deceleration under a speed ramp, the widely used quadrature phase-locked loop (QPLL) exhibits notable transient errors, which reduce control accuracy and load capacity. To address this limitation, this study proposes a Type III improved phase-locked loop (IPLL), inspired by the analysis of the QPLL’s transfer function. The proposed IPLL can reduce estimation errors to zero, offers easy parameter tuning, and requires only minor modifications to the QPLL structure. To obtain the position error used as the PLL input, the motor’s extended back electromotive force is estimated using an extended state observer (ESO), followed by an arctangent calculation. This study comprehensively examines the effects of the parameter-adaptive ESO (AESO) and the fixed-parameter ESO (FPESO) on estimation errors. Results show that the AESO can reduce steady-state errors in the two-phase stationary frame (αβ frame) but exerts no effect on transient errors. Owing to stability criteria, the AESO is applicable only to motors with low rated speeds. This lack of effect on position error in the γδ frame has been verified via simulations and experiments. Finially, effectiveness of the proposed IPLL reveals that it reduces the position error from 22.5° and 15.6° to 3° and 2° under no-load and rated-load conditions, respectively. These improvements enhance the torque control capability, dynamic response, positioning accuracy, reliability, and robustness of the sensorless system while also reducing energy loss.