A Software Based Robust Resolver-to-Digital Conversion Method in Designed in Frequency Domain

Abstract

Resolvers are extensively used in applications that demand instantaneous, accurate and high-resolution information angular position or speed. A resolver's analog outputs have been modulated by angular position and an R/D converter is always adopted to recover the angular position in digital form. With the rapid growth of micro-processor technology, more and more attention has been focused on software based R/D conversion methods because of their merits, such as saving in cost, weight and size. In practical applications, noise is often unavoidable and might cause severely negative effect on results of software based R/D conversion schemes. To tackle the problem, a new R/D conversion method designed in frequency domain is presented in the paper. Firstly, a new complex signal is constructed by taking the cosine output as the real part and the sine output as the imaginary part. After transforming the complex signal into frequency domain, only the components at the carrier frequency are utilized to calculate the angular position. Thus, the negative effect of noise, whose energy is located out of the band of the carrier frequency, can be removed. In the simulation experiments, dc drift, white noise and pulse disturbance are utilized to evaluate the performance of R/D conversion algorithms, and the experimental results show the frequency domain based method is favorably robust to all the three types of noise.