The algorithm described in this paper was tested again for
each of the sensor types. This time the algorithm was run for
20 steps so as to compare the final torque ripple estimates
for each. Table 2 shows the values of the estimated torque
for each of the three sensor inputs used, after convergence.
If the data obtained when the torque sensor is used to
provide feedback to the algorithm is considered as the ideal
response, the results from the other two sensor types can be
compared to these. The error in the magnitude for the
estimated torque using the speed measurement as the sensor
is 0.77% for the 20th harmonic and 1.28% for the 24th.
When a microphone is used for feedback, the error is 0.83%
for the 20th and 1.95% for the 24th.