Estimation of Non-Idealities in Sigma-Delta Modulators for Test and Correction Using Unscented Kalman Filters

In this paper, an enhanced unscented Kalman filter is used for the estimation of non-idealities in sigma-delta modulators. As sigma-delta modulator based analog-to-digital converters are known to be prone to performance degradation from many circuit non-idealities, testing and selection of the manufactured chips as well as post-correction procedures are required to ensure a reliable performance. In contrast to most state-of-the art techniques, the proposed estimator is capable of simultaneously tracking a large set of non-ideal parameters with high accuracy and in a very short offline time within the range of microseconds. Three possible implementations are proposed and it is shown how the technique is used to perform an in situ non-ideality estimation and subsequent calibration of a manufactured chip, bringing the modulator back into its designed performance range. It is further demonstrated that the estimator can reliably detect the non-idealities over a wide range when intentionally degrading the performance of the chip by manually sweeping its trimming parameters. The presented hardware measurement results prove that the proposed technique is practically applicable to the test and/or calibration of sigma-delta data converters.