Noise-robust software based R/D conversion methods using over-sample technique

A resolver is a type of rotary electrical transformer used for measuring degrees of rotation. Resolver’s outputs have been modulated by angular position and Resolver-to-Digital (R/D) conversion is required to retrieve the angular position in digital form. Software based R/D conversion methods have the merits of saving in cost, weight and size, however, they also suffer the shortcoming of low noise robustness. In this paper, two software based R/D conversion methods are proposed to improve the robustness against noise by using over-sample technique: one is designed in the time domain; the other is designed in the frequency domain. The time domain method calculates average amplitudes of resolver’s outputs and retrieves angular positions by arctangent operations; the frequency domain method only employs spectral components at carrier frequency to calculate angular positions. Experiments conducted on simulated and real resover’s outputs have demonstrated the validity of the proposed approaches.

Figure optionsDownload as PowerPoint slideHighlights
► We report two software R/D conversion methods based on over-sample technique.
► Robustness of software R/D conversion is improved by calculating one angular value with multiple data pairs instead of one data pair.
► From a filter point of view, the time domain method removes noise by a mean filter.
► The frequency domain method removes noise by a band-pass filter with the center frequency being set to the carrier frequency.