New Low-cost High-resolution Rotor Position Detector Research

Rotor position sensor is very important in the area of motor control. Especially, high-resolution sensor is one of the key components in control system like that of BLAC motor. Though many technologies of sensorless motor control have been developed, they may not work well under some conditions, such as starting, low-speed or high-speed operations. Traditional high-resolution rotor position sensor like optical encoder, resolver are expensive and not sufficiently reliable. In this thesis, a low-cost high-resolution rotor position sensor is developed. It consists of two linear Hall ICs, a 2-pole parallel-magnetized magnet ring, and an electronic processing unit.In the developed sensor, the 2-pole parallel-magnetized magnetic ring is attached on the rotor. The two linear Hall ICs are fixed on the stator with a90-degree phase difference. The Hall ICs generate a sin(θ) and a cos(θ) signals. By further analog and digital process to the signals, the angle signal 6 can be obtained.Firstly, a calculation formula for the air-gap magnetic flux density excited by the 2-pole parallel-magnetized magnetic ring given in the literature is validated with FEM. Further, some more formulae for air-gap magnetic flux density excited by different excitation frameworks are derived, and also validated with FEM. Using the formulae, magnet ring of the proposed sensor is designed.Extended Kalman Filter (EKF) is employed to process the output signals of the Hall ICs. Moreover, a low-pass filter is used before the EKF in order to normalize the signals, including regulating the amplitude of the AC component and the DC offset in the signals.Simulation using software of Matlab is done during the process ofdesigning the EKF arithmetic. Firstly, the signal model is built. Steady-state and dynamic simulations under different circumstances are also implemented. Then, principles and methods of choosing the parameters for the EKF arithmetic are investigated. And, influence of the phase shift between the two signals is also discussed. Finally, simulation for programme of waveform detecting and waveform regulating is performed.A piece of Programmable System on Chip (Psoc) produced by Cypress MicroSystems is used to process the signals digitally. It contains sufficient internal hardware, hence, the whole electronic processing unit is simple. However, the Psoc is relatively slow. To speed the processing, the EKF arithmetic is simplified. Then, some characteristics of EKF have been investigated through experiments.The EKF works well during steady function and is robust to the variation of the signal parameters. However, its performance is deteriorated during dynamic progress, for example, when the motor speed changes quickly.In this thesis, a method to switch between the mode of EKF based position estimation and the mode of directly calculating (using atan or asin/acos function) via judging the speed characteristic (speed is one of the state-variables of EKF) is proposed. Experiments show that it works well, improving the performance of the rotor position sensing during dynamic progress.Because of the uncertain variety of the two signals caused by exterior circumstance, waveform detecting and regulating programme is inserted before estimating or calculating the angle. It works well. Moreover, solution to the problem of phase-delay caused by the low-pass filter is discussed in the thesis. Also, the unique criterion of the output position signal in both rotary direction is verified experimentally.An incremental optical encoder is utilized to generate a reference rotor position, which is compared with the output signal of the developed position sensor.