Reconfigurable Quantization of Oversampled Signals Under Discrete-Time Filtering

摘要

In modern wireless applications such as carrier aggregation, oversampled ADCs are required to be reconfigurable, to accommodate varying spectral signal energy distributions and bandwidths of interest. While progress has been made using several circuit and system level techniques, a fundamental approach that considers the signal's spectral properties is lacking. In this paper, we re-evaluate the oversampling problem: we use a time-domain analysis to show that given a signal spectrum, quantization noise is decreased under certain discrete-time filtering operations. Using this alternate approach, a collection of ADC architectures are constructed, special cases of which, reduce to well-known data conversion techniques in literature. Analysis and simulations confirm comparable performance of the proposed ADCs with other oversampled ADCs such as modulators. When additional information about signal energy distribution is available, it is shown that the proposed ADC architectures are adaptable on a system level, thus surpassing the performance of modulation. Detailed theoretical explanations along with behavioral simulations are shown.