Channel Mismatches Adaptive Calibration Algorithm For Time-Interleaved ADCs

Next-generation aerospace and defense and test and measurement system bandwidths are moving from 10s to 100s of megahertz to multi-gigahertz of instantaneous bandwidths.Trends in phased-array radar,5G wireless test systems,electronic warfare and digital oscilloscopes are pushing bandwidths higher and dramatically increasing the urgent need for high-speed and high-resolution analog-to-digital converters(ADCs)in a system,which makes time-interleaved ADCs(TIADCs)the focus of the world's competing studies.The TIADC is a promising solution to improve the sampling rate of the high-resolution ADCs.Ideally,a TIADC is composed of M completely identical ADC chips(or cores)working alternately at the equal sampling intervals.As a result,the overall equivalent sampling rate is M-time multiplied and the resolution maintained.However,an actual TIADC will suffer from the mismatches among ADC channels due to the fabrication dispersion and the ageing,voltage and temperature variations.These mismatches,e.g.offset,gain and time skew,will produce spurious components on the output signal,resulting in the performance degradation of a TIADC system.The channel mismatches calibration is crucial to the application of the TIADC system.Focusing on the low complexity fast blind adaptive correction algorithms for the linear mismatch errors in a TIADC system,theoretical and technical researches have been carried out in this dissertation on the error analysis and modeling,frequency-domain and time-domain based calibration algorithm,high precision time compensation filter design,circuit design and algorithm implementation.The key works and innovations are as follows:Firstly,aiming at the offset,gain,timing and bandwidth mismatches in a TIADC system,their characteristics and the influence on the output signal are analyzed.A frequency domain modeling method is discussed and the linear mismatches calibration method based on frequency domain adaptive filtering is presented.Secondly,a digital blind algorithm based on cyclic auto-correlation is proposed to correct the timing errors in a dual-channel TIADC system.The basic idea is to construct a systematic errors measurement function using the cyclic auto-correlation.The problem can easily be solved benefiting from its strict convexity.At the same time,the algorithm uses the analytical/synthetic filters based method to realize the correction network.By reasonably setting parameters and the Taylor series approximation,the algorithm of correcting network structure and mismatch parameter estimation is simplified.Thirdly,a first order statistics based all-digital background method is proposed for the linear mismatches calibrations in a TIADC system with arbitrary number of channels.This method can correct the offset,gain and timing errors within five step iterations for wide stationary input.This method has the advantages of low complexity and fast convergence.It is easy to be modularized and can be used for fast online blind calibration of the TIADC mismatches.Fourthly,a weighted least square(WLS)optimal design of the FIR filter for high precision time compensation of a TIADC system is proposed.On the premise of the ensured precision,this method effectively reduces the computation compared to the discrete spectral parameter method.The derivation of the WLS optimal closed solution of the filter coefficients is detailed in this dissertation.The fractional delay digital filter designed by the given algorithm can be used for high precision dynamic time compensation.Finally,the software and hardware design of a 400 MHz 12 bit dual-channel TIADC system is presented to verify the effectiveness of the algorithm and system design.