Improving the accuracy of low-cost resolver-based encoders using harmonic analysis

We present a software-based error compensation method for improving the accuracy of low-cost, resolver-based 16-bit encoders. The error profiles of 10 such encoders have been obtained by calibrating them on a high precision Rotary Table. Repeated calibration of these encoders, carried out over a period of few months, suggest that error profiles although unique for a particular resolver–decoder combination, are predominantly of systematic in nature and hence can be, in principle, corrected for by using an appropriate compensation procedure. The error profiles of the encoders, when subjected to detailed harmonic analysis, reveal that in addition to the dc term there is also strong presence of periodic components with contributions from both lower and higher order harmonics. After incorporating the contribution from most prominent 10 harmonics, it is found that the measured error profiles can be approximated quite accurately by using a Fourier series. The software-based error compensation procedure thus involves obtaining an appropriate Fourier series representation for each of the resolver–decoder combination separately and then use the same for calculating the corrected angle by subtracting the predicted error from the measured value of the encoder angle. We have implemented this error compensation procedure for all the 10 encoders and the final results indicate that the accuracy of a low-cost, resolver-based 16-bit encoder can indeed be improved from quoted accuracy of ∼ ± 6 to ∼ ± 0.6 arc min . Apart from presenting the methodology followed for calibrating the encoder and the details of the software-based error compensation procedure, a brief discussion about nonideal characteristics of a resolver such as Amplitude imbalance, Quadrature error, Inductive harmonics and Excitation signal distortion is also presented in the paper.