Self-sensing hysteresis-type bearingless motor

This thesis presents the design, implementation, control, and experimental evaluation of a self-sensing hysteresis-type bearingless motor (SSBM). Bearingless motors are well-suited to high-speed applications due to their frictionless operation, but the magnetic levitation of the rotor is unstable without airgap sensors. Eliminating these sensors decreases system cost and volume while increasing system robustness. This work presents the design for a hysteresis-type bearingless motor that operates without the use of airgap sensors.

Bearingless motors use a single stator to generate torque and suspension forces to control the position of the rotor. Some measure of the airgap length is needed to enable stable magnetic suspension, so this thesis proposes the injection of a high-frequency carrier signal to the stator windings to amplify the change in coil inductance with rotor position. The coil response is demodulated against the carrier signal to provide an estimate for airgap length.

This design uses a stator with 12 independently controlled windings to generate a 4-pole magnetic field for torque and a 2-pole magnetic field to control the rotor suspension. Analytical and finite-element simulation demonstrate that if the current in the windings is controlled, demodulating the winding voltage against the carrier signal gives a good estimate for rotor displacement. When the winding voltage is considered the input, the current through the coils can be used for estimation, but the result is highly dependent on the suspension field.

The prototype developed to test this operating principle has been constructed and tested with voltage as the winding input. Inductive sensors are used to provide a "ground truth" signal to evaluate the estimation result and to provide feedback for the rotor suspension while the estimation is being developed. Initial results show that even under voltage control the SSBM prototype is able to estimate the rotor displacement, but only when the suspension control is active. Current control for the system has been developed and will be implemented as the immediate next steps.