Continuous Position Estimation Technique for Optimized Torque Ripple using Low-Resolution Sensor

With electrification of transport system, high torque and power density motors such as permanent magnet synchronous and variable reluctance motors are being explored and used for traction application. These high-performance machines require continues angle variation information for robust control action. This continuous angle variation data is measured using sensors such as optical or magnetic encoders, which are very expensive. Moderately expensive devices such as GMR or TMR based sensors, linear hall sensors can also be used. These sensors provide output with certain accuracy and may require suitable motor design or complex signal conditioning to improve accuracy. Bi-polar hall-based sensors are being increasingly looked for as cheap alternates to such sensors. These sensors provide position data at a low resolution and provide inaccurate data due to improper physical mounting. This results in degraded motor performance. The low-resolution data can be converted to high-resolution using simple interpolation-based algorithms. These algorithms often use speed or acceleration values calculated from previous state changes for interpolation and resulting into high resolution position data. The error in angle is corrected at each instant where a hall sensor state change is detected. This angle correction leads to sudden jumps in position data which may lead to significant torque ripple at the drive end. This paper addresses two above mentioned problems associated with the bi-polar hall sensors. Back-emf based correction algorithm is suggested to correct the inaccuracies arising due to physical mounting of hall sensors. To prevent jumps in position data, a method is proposed to the optimize the trade-off between torque ripple and data inaccuracy. The paper also proposes an overall position estimation scheme to help identify the optimal points to switch between hall-based, speed interpolation-based and acceleration interpolation-based methods for effective drive control.