Calibration for Realization Errors of Two-Channel Hybrid Filter Bank Analog-to-Digital Converters Based on Frequency-Dependent Model

摘要

Hybrid filter bank (HFB) analog-to-digital converters (ADCs) are used to achieve higher sampling rates for medium-to-high resolution applications in many communication systems. However, the analog filter realization errors and channel mismatches severely degrade the reconstruction performance of HFB ADCs. A general error analysis, which takes analog imperfections and channel mismatches into consideration based on a single-channel filtering (SCF) HFB architecture is presented in this paper. A frequency-dependent model (FDM) for sub-ADCs output is then derived to integrate all the realization errors to an equivalent channel-transfer function (ECTF). Based on the ECTF, the perfect reconstruction condition is derived as the optimization object for the synthesis filter banks. Finally, a two-channel SCF HFB ADC prototype with 12-bit resolution and 200 MHz sampling rate is implemented. After mismatch correction with 64-tap finite impulse response (FIR) filters, 75.75 dB of spurious-free dynamic range (SFDR) is achieved over the whole bandwidth. The experimental results show that the average SFDR is enhanced by about 10-25 dB compared with the traditional calibration method.