Self-sensing approach and experiments of active magnetic bearings based on synchronous-sampling direct current estimator

Abstract: Frictionless magnetic bearings can contribute to modern agricultural equipment, such as pump, fan and blast, thereby to achieve high efficiency, energy-saving and long-term reliable operation. But magnetic bearings are confined to the configuration and price of position sensors. In the recent years, the self-sensing and sensor less mode have been developed to extend the application of the magnetic bearings. Most previous studies are focused on the estimation algorithms, such as the state observer approach and parameter estimation approach. Moreover, the extra phase-shift is introduced by the estimator which consists of the analogue filter or FIR or more complex mathematic algorithm. The self-sensing path turns complex and longer due to the fundmental ripple needs to be extracted, amplified and filtered. Therefore, the key research target of this paper is the operation of position estimation. An absolute mathermatical approach was proposed to extract the rotor positon using the zero ripple characteristic based on the cosine-function of coil current. A self-sensing algorithm based on SS-DCE (Synchronous-Sampling-based Discrete Current Estimator) can be used to remove the extra phase that introduced by the analog or digital filters of the sensing path. Synchronous-sampling technique was used to sample the coil working current in the estimator due to PWM switching harmonics without the discrete current. Moreover, the phase shift can be minimized in SS-DCE scheme because the filters of amplitude demodulation have been eliminated, and the configure of AMBs are simplified. The numerical results of SS-DCE were also compare with that of the digital filtering amplitude demodulation approach(DFAdM). The test results of the position estimators showed that the optimal performances can be obtained, when the switching frequency and synchronous sampling frequency are 2 kHz, control frequency is 50 Hz, sampling frequency of DFAdM is 100 kHz, and some electrical conditions as the bias current 3.0 A, nominal inductance 13.2 mH and power supply 50 V. The SS-DCE was also verified by the simulation in 1-DOF AMBs. The static performance evalution was performed on the proposed estimator, indicating that the precision of SS-DCE was about 2.72%, and the absolute error was 5.43 μm in the position range of 50-250 μm. The precision of DFAdM estimator was 4.85% and the absolute error was 9.71 μm. Frequency response analysis was used to evaluate the dynamics performance in open loop state, indicating 158° phase margin at the eigen frequency 19.26Hz. Moreover, a rig of self-sensing AMBs was setup with TMS320F2812, LEM HX-05, PM10CJS060 and micro-positioning platform. The precision of SS-DCE was about 2.53% referring to eddy current sensor HZ-891. SS-DCE-based rotor showed good stability, rapid convergence and acceptable overshot under pulse disturbances that applied through a direct knock on the truss by a rubber hammer. The proposed approach can provide a direct guidance for the active control of magnetic levitation, magnetic damper and other engineering fields. The influence of PWM duty-cycle and low SNR on magnetic bearings can be performed in the near future research works, thereby to enhance the application of magnetic bearings for modern agriculture.