Two high performance position estimation schemes based on sliding-mode observer for sensorless SPMSM drives

To improve the control performance of surface permanent magnet synchronous motor (SPMSM) without speed sensor, two sensorless schemes are proposed based on the sliding-mode observer (SMO). As the estimated back electromotive force (back-EMF) are contaminated by high frequency switching noise, inverter nonlinearities and flux spatial harmonics etc., direct extraction of the position information from the estimated back-EMF may result in rich error ripples. This would greatly deteriorate the control performance. In this paper, two simple methods, the programmable cascaded low-pass filter (PCLF) and complex coefficient filter (CCF) are proposed to suppress the error harmonics in the estimated position without complex online adaption algorithm and the phase delay. Furthermore, the discretization error caused by digital implementation is carefully considered and compensated to ensure satisfactory performance over a wide speed range. Both methods are simulated and experimentally tested on a two-level inverter fed 2.4 kW SPMSM platform. It is shown that both methods can achieve good steady state and dynamic performances.