I. Introduction
For the field-oriented control it is necessary to know the precise position of a rotor at every instance of time. Sensorless drives are preferable due to smaller cost, size advantages and reliability. There are three main sensorless methods of rotor position estimation: methods based on the back-EMF estimation, methods which detect magnet anisotropy of the motor and rotary transformer methods applicable to electrically excited synchronous machines. For the first type of observers, the accuracy of the position estimation depends on the value of the back-EMF, which is very small or absent at low and zero speeds. This limits the operating range of back-EMF methods [1]–[3].