The full-order observer without feedback gains and with proposed feedback gains are simulated in the second part. To emphasize the improvement of the rate of convergence, the simulations in this part are only performed in low speed range. In each simulated situation, the speed command is stepped at time 0 s from standstill to 120 rpm and reverses speed command from 120 rpm to −120 rpm at time 1 s. The load torque is suddenly changed from no load to positive full load at time 0.5 s, and then returns to no load again at time 0.7 s in positive speed range. In negative speed range, the load torque is suddenly changed from no load to negative full load at time 1.5 s, and then returns to no load again at time 1.7 s. Then, the stator flux errors are compared. In simulations, the pure integrator does not suffer from saturation and drift problem, therefore the stator flux from the pure integrator is utilized as the actual stator flux. The stator flux error is the difference between the actual stator flux and the estimated stator flux from full-order observer. Fig. 7, Fig. 8 represent the d- and q-axis stator flux errors from simulations of using full-order observer without feedback gains and with proposed feedback gains, respectively. When comparing with the d-axis stator flux error of full-order observer without feedbacks gains in Fig. 7 (a), the d-axis stator flux error of full-order observer with proposed feedback gains in Fig. 8 (a) not only has faster the rate of convergence, but also has much smaller the error amplitude. The results are still the same when consider the q-axis stator flux errors in Fig. 7 (b) and 8 (b). The maximum rate of convergence of the full-order observer without feedback gains is approximately 0.4 s as appeared in Fig. 7 (a) when speed is reversed, while the maximum rate of convergence of the full-order observer with proposed feedback gains is less than 0.1 s as shown in Fig. 8 (a) when speed is reversed. The maximum amplitude errors are approximately 3 × 10−2 Wb and 0.8 × 10−9 Wb when starting motor from standstill as shown in Fig. 7 (b) and 8 (b), respectively. Moreover, when taking load to motor, the stator flux errors of the full-order observer without feedback gains slightly increase or decrease with the DC offsets that mean more stator flux errors when taking load torque a long time, these phenomena disappear when using full-order observer with proposed feedback gains as seen from Fig. 7, Fig. 8. These results well testify for the validity of the proposed feedback gain design method.