An advanced unscented Kalman filter algorithm for parameter estimation in continuous-time sigma-delta modulators

In this paper it is shown, how the convergence of a previously presented version of the unscented Kalman filter can be highly improved. The Kalman filter is thereby used for estimations of nonidealities in continuous-time sigma-delta modulators. The performance of sigma-delta modulators is known to be strongly affected by several nonidealities, located inside the circuit. The state of the art shows many techniques to estimate these parameters for calibration of the modulator, so that its original behavior can be restored. Also an unscented Kalman filter was shown, being able to estimate many of the nonidealities in a joint fashion. In the following, a major change of the Kalman filter is presented. When applying the novel algorithm to the estimation task, the convergence is improved such that the time for an estimation run, leading to equal accuracy, can be decreased by up to 40%.