Meanwhile, it is conventionally known that I_Δ and ?_Δ are not changed depending on an angular error ({tilde over (θ)}_r). Thus, the current signal (i_sig) is known as being in the form of a sine wave with respect to the angular error. In addition, an angle estimator (or, an observer) is known as operating such that the input current signal (i_sig) is converged to a zero up-crossing point at which a negative number changes into a positive number to estimate a rotor position. Thus, it is known that the angular error is converged to ?_Δ in a normal state.

However, unlike the conventional technique known in the art, during an actual sensorless operation, a current driving current changes according to an angular error, and also a motor inductance changes according to the current driving point and the rotor position.

FIG. 3 is a diagram for illustrating that a current driving point is changed according to an angular error ({tilde over (θ)}_r). Assuming that the motor operates with a maximum torque per ampere (MTPA), the current driving point is determined by a torque command. In FIG. 3, a dotted line 31 shows a MTPA curve on a (dr, qr) current plane of a synchronous coordinate system. In this case, the current driving point of any torque command is determined as a point on the dotted line 31.

However, since the current is controlled on the basis of the estimated coordinate system (d{circumflex over (r)}, q{circumflex over (r)}) during the sensorless operation, if an angular error occurs, the driving point is moved by the angular error ({tilde over (θ)}_r). Thus, the driving point of the torque on the MTPA curve 31 intended by a user is changed to a driving point on another curve 32.