A position estimation method for active magnetic bearings based on the star point voltage

Self-sensing position estimation methods for active magnetic bearings allow for the reduction of system cost and complexity. Many known methods rely on current measurements or their derivatives to estimate the rotor position. However, measuring a voltage often results in a better signal-to-noise ratio compared to current measurements. Despite that, star-connected active magnetic bearings and the related star point voltage, are rarely addressed in research. This is notable, as the star point voltage provides valuable information for self-sensing in other types of electric machines. This research aims to utilise the star point voltage of an active magnetic bearing to estimate the rotor position. The outlined study shows an analytical derivation of the star point voltage for a four-phase active magnetic bearing followed by a description of a differential measurement procedure for the star point voltage. It is demonstrated that the resulting measurements only depend on the coils inductance and the known DC supply voltage. To extract the position information from the obtained signal, the plant is characterised at standstill. The resulting measurements were used for polynomial fits, implemented on the microcontroller. Finally, the proposed method is experimentally validated inside the control loop. Different experiments show mean absolute estimation errors ranging from 0,8 um to 10 um.