The analysis and development of sensors for active magnetic bearings

Abstract

This dissertation presents the analysis and development of contactless sensors for active magnetic bearings (AMBs). The rotors of AMB systems are suspended in air without any direct contact between the stator and the rotor. An AMB system requires rotor position feedback for control. To extract the position from the system some sort of contactless sensor must be constructed. The aim of this project is to develop different types of contactless sensors for AMBs. Through this project basic knowledge of sensors is gained and expertise is established in the Engineering Faculty. These models can be used for further research investigations. This project is divided into three different sub-projects; 1) an optical sensor, 2) an inductive sensor and 3) a self-sensing sensor. Each of these sensors are analysed, developed and tested on AMB models. The optical sensor constitutes an infrared transmitter and an infrared receiver. This transmitting and receiving technique relies on the amount of light transmitted and is known as the barrier optical sensor. This sensor was evaluated on a magnetic bearing demonstrator. The optical sensor has a very high bandwidth and is ideal when low noise is a requirement. The inductive sensor principle of operation is based on the changing effect of the inductance due to the air gap. If the coil is placed near a ferromagnetic material the inductance of the coil will change according to the size of the air gap. The sensor's coils are activated with a high frequency voltage. This constant frequency voltage will be applied to a constant reactance if the air gap is constant. As the distance between the rotor and the stator of the AMB changes, the air gap changes, thus the output of the sensor coils is proportional to the distance. The inductive sensor was implemented on a homopolar AMB with a maximum rotating speed of 3000 rpm. It was determined that the inductive sensor has good linearity properties over a specific measurement range. The bandwidth of the inductance sensor was designed to be above 500 Hz, which is ten times higher than the bandwidth of the homopolar AMB. In the case of very high rotating speeds, the bandwidth of the inductive sensor may be too low for geometric position control. The self-sensing technique makes use of the current response of the power amplifier and the voltage across the power amplifier. The slope and amplitude of the high frequency current component changes according to the air gap. This changing current is demodulated to determine the position of the rotor. By using a self-sensing method the voltage and current measurements could be taken externally from the AMB. These voltages and currents are then converted to an estimation of the rotor position. This external electrical measurement has the advantage that there are no sensors present inside the AMB. The self-sensing method was evaluated on a heteropolar active magnetic bearing. It was determined that the self-sensing sensor bandwidth was high enough to overcome disturbances up to 100 Hz. The sensor was successfully implemented on a heteropolar AMB with switching power amplifiers. The sensor output contains a 100 Hz noise component due to a ripple component on the power amplifier capacitor. The linearity of the sensor is impaired by eddy current losses in the AMB. Future work will focus on the improvement of the self-sensing sensor. The eddy current losses could be minimized by using thinner laminations in the AMB magnetic circuit. This will reduce the losses and increase the linearity of the self-sensing sensor. By implementing a differential sensor where the air gaps on both sides of the rotor are estimated and subtracted a more sensitive and noise immune system could be obtained. Implementation by means of a digital signal processing (DSP) device is also an important future activity. This project produces a firm basis on sensor technology in the research group.