Enhanced position estimation in self-sensing magnetic bearings using separate sensing and control functions

This paper presents a novel approach to enhance position estimation in self-sensing magnetic bearings (MBs) by employing separate windings for sensing and control functions. Traditional self-sensing MB systems utilize a single winding for both actuation and sensing, which leads to increased noise and estimation inaccuracies due to signal interference and mutual inductance effects. To address these challenges, a one-degree-of-freedom MB system was developed featuring dedicated bias windings for sensing and control windings for levitation actuation. Bipolar pulse-width modulation (PWM) was applied to the bias windings to amplify current slope measurements, while unipolar PWM and PI current control were used for the control windings to minimize current ripple. Experimental validation was performed using a DSP TMS320F28379D and commercial AEC PU-09 gap sensor as a reference. Results demonstrate that the proposed system achieves a 6.8-fold increase in average current slope and a 37% reduction in current slope noise compared to conventional single-winding designs. The gap estimation noise for the separate winding configuration is only 1.66 times higher than the commercial gap sensor, with levitation performance comparable to sensor-based feedback systems. The system bandwidth reaches 20 Hz, compared to the 15 Hz bandwidth of sensor-based feedback systems. These improvements confirm that separating sensing and control functions significantly enhances self-sensing accuracy in MBs, paving the way for cost-effective and reliable MB applications without external sensors.