POSITION SENSORLESS CONTROL METHODS: PERMANENT MAGNET BRUSHLESS DC MOTOR

POSITION SENSORLESS CONTROL METHODS: PERMANENT MAGNET BRUSHLESS DC MOTOR

The basic idea of position sensorless control methods is to eliminate the position sensors (usually three Hall sensors). To accomplish this task, additional circuitry and computational efforts are required to estimate the
commutation instances of the PERMANENT MAGNET BRUSHLESS DC MOTOR from the voltage and current signals which can easily be sensed.
Therefore, sensorless techniques demand high performance processors with large memory and program codes for computation and estimation, as compared to sensor-based drive systems. PERMANENT MAGNET BRUSHLESS DC MOTOR can be modeled by the same equivalent circuit for each phase winding, where the source voltage ‘v’ supplies current ‘i’ to the phase circuit consisting of series-connected resistance ‘R’, inductance ‘L’, and back EMF ‘e’. The back EMF is a result of the movement of the PM rotor, thereby, dependent on rotor position and proportional to rotor velocity. The machine voltage and current waveforms reflect the rotor-position dependence of the inductance and back EMF. Therefore, the voltage and current waveforms can be analyzed to extract the back EMF or inductance (or a combination of the two), from which the rotor position can be estimated in the position sensorless schemes. The
position sensorless approach has many advantages, e.g. minimum installation cost, minimum space requirement, no environmental restrictions (e.g. high pressure and temperature environment in HVAC compressors), EMI free position information, reduced controller cost etc. These sensorless techniques may be broadly categorized as:

1. Back electromotive force (BEMF) sensing,
2. Inductance variation sensing,
3. Flux-linkage variation sensing.

BACK ELECTROMOTIVE FORCE (BEMF) SENSING CONTROL METHODS: PERMANENT MAGNET BRUSHLESS DC MOTOR

In PM brushless DC machines, the magnitude of the back EMF is a function of the instantaneous rotor position and has trapezoidal variation with 120º flat span. However, in practice, it is difficult to measure the back EMF, because of the rapidly changing currents in machine windings and induced voltages due to phase switching. The back EMF is not sufficient enough at starting until the
rotor attains some speed. Therefore, it is a usual practice to make the initial acceleration under open-loop control using a ramped frequency signal so that the back-EMF is measurable for the controller to lock in.
One of the popular starting methods is “align and go”, in which the rotor is aligned to the specified position by energizing any two phases of the stator and then the rotor is accelerated to the desired speed according to the given commutation sequences. The “align and go” method suffers demagnetization of permanent magnets due to large instantaneous peak currents at starting. The zero-crossing points of the back EMF in each phase may be an attractive feature to use for sensing, because these points are independent of speed and occur at rotor positions where the phase winding is not excited. However,
these points do not correspond to the commutation instants. Therefore, the signals must be phase shifted by 90° electrical before they can be used for commutation.The detection of the third harmonic component in back EMF, direct current control algorithm and phase locked loops have been proposed to overcome the phase-shifting problem. However, the direct current
control algorithm suffers filtering problem of sensed voltage signals which limits the operation range above 200rpm. The third-harmonic approach assumes equal inductance in all three phases, which is only valid for surface-mounted magnet motors; however, in the case of rotors with saliency, errors in position estimation arise due to rapidly changing phase currents. To measure the back EMF across the terminals of a star-connected machine, it is necessary to have the machine’s star neutral terminal. The back EMF method has been applied in special-purpose low-cost applications for fans and pump while ignoring these problems.

INDUCTANCE VARIATION SENSING CONTROL METHODS: PERMANENT MAGNET BRUSHLESS DC MOTOR

The fundamental concept behind the inductance variation is the rate of current change in the motor which depends on the inductance of the winding. The inductance variation can be sensed after injection of a current pulse in
the armature windings. This scheme is particularly useful at zero speed when there is no back EMF. This method is suitable for the IPM (Interior Permanent Magnet) BLDC motor with high performance material such as the NdFeB magnet. In order to get various inductance profiles, a large current pulse is
required. Thus, these methods are not suitable for a SPM-type BLDC motor with ferrite magnets. Therefore, the application of inductance variation sensing methods may be useful to address the problem of starting, including identification of the rotor position before full excitation of the machine. Initial rotor position identification is particularly important in applications such
as traction, where any reverse motion is not acceptable. It is also reported the detection of initial rotor position of a salient pole PM motor by high-frequency injection methods using voltage pulses.

Despite implementation difficulties, several methods of position sensing from inductance variation have been applied for sensorless operation. Low frequency excitation pulse results in large current amplitudes which facilitate
easy detection, but can cause audible noise from the motor. Whereas high frequency avoids audible noise, but reduces current amplitudes. Therefore, choice of an appropriate modulation frequency and modification in the machine rotor can further improve rotor position sensing using this method.

FLUX-LINKAGE VARIATION SENSING CONTROL METHODS: PERMANENT MAGNET BRUSHLESS DC MOTOR.

This is based on the phase voltage equation of the motor. Since the phase flux linkages are a function of current and rotor position, therefore, phase flux linkage can be estimated continuously by integrating the voltage after subtracting the resistive voltage drop from the phase voltage. The open-loop integration is prone to errors caused by drift, which can be reduced if the pure integrator is replaced by a low pass filter or an alternative integrator structure. In most electrical machines, it is not practical to measure the phase voltages directly, because of isolation related issues; therefore, applied phase voltage is estimated from DC supply voltage of the solid-state
converter.