Efficient Polynomial Inversion for the Linearization of Pipeline ADCs

We present an efficient solution to the digital enhancement of the characteristic of pipeline analog-to-digital converters (ADCs) given estimates for non-linearity behaviors. A third order polynomial model is assumed for non-linearity and several inversion mechanisms are analyzed in terms of complexity and power consumption. Two efficient implementations based on a predictor-corrector scheme and on Newton-Raphson iterations are presented. Missing codes appearing after linearization are emphasized and the benefits of increasing resolution are shown. Limitations of linearization techniques are finally reported in terms of achievable ENOB and SFDR