Comparative Evaluation of Methods for Continuous Rotor Position Estimation using Low Resolution Hall Sensors

Permanent magnet motors used in mild hybrid vehicles are usually driven by trapezoidal control with low resolution position sensors to crank IC engines. Field oriented control (FOC) can be used for enhancing the dynamic performance, which demands high resolution position data obtained from encoder, resolver or similar sensors. Additional cost for these sensors can be avoided by estimating continuous rotor position using low-cost, low-resolution hall sensors. The accuracy of the estimation is affected by the inaccuracies in the hall sensor signals in the form of constant position error in single direction, zero position jitter and periodicity error. In this paper, comparative evaluation is discussed for three major types of rotor position estimation methods for different cases of hall sensor errors. The first method uses linear interpolation based on least squares method. The second method uses transformation of hall rotary vector into orthogonal axes and synchronous frequency tracking filter. The third method uses combination of back emf observer and hall effect sensor output with a phase locked loop (PLL) like structure. All these methods are simulated and analyzed using a MATLAB based PMSM model. The accuracy of these methods during steady state and transient operation considering different types of errors in hall sensor signals are discussed in this paper. Finally, hardware implementation of least square estimation method is presented.