Algorithmic ADC Offset Compensation by Nonwhite Data Chopping: System Model and Basic Theoretical Results

This paper is devoted to show the impact of nonwhite chopping on the offset compensation in time-interleaved analog-to-digital converters. We develop a theoretical framework allowing the selection of optimal chopping sequences. We show that, on the one hand, the adoption of these (generally nonwhite) sequences allows to achieve faster offset compensation (thus increasing the signal-to-noise ratio) and, on the other hand, a better spectral shaping (thus increasing the spurious-free dynamic range). As a byproduct of our analysis, we prove that the average offset estimation which is used in many ADC implementations is asymptotically the best available linear estimation of offset that, in turn, is the best estimation when the signal to be converted can be assumed to be a Gaussian process.