Mismatch Calibration Methods For M-Channel Time-Interleaved Analog-to-Digital Converters

Analog-to-digital converters(ADCs)are essential components in mixed-signal processing systems.They connect the analog signals in real world with the digital signals in digital signal processors.Thus the performance of ADCs has a vital impact on the performance of the entire mixed-signal processing systems.The demand for high-speed highresolution ADCs boosts with the increase of the signal bandwidth and transient dynamic range.However,the performance of a single ADC is limited by the fabrication process and materials.Time-interleaved ADC(TIADC)is a promising architecture which exploits several low-speed ADCs working in parallel to increase the sampling rate by a factor of the channel number.However,this architecture is very sensitive to the mismatches between the parallel channels.The performance of TIADC deteriorates severely even with just slight mismatches between channels.In order to improve the dynamic performance of TIADCs,compensation methods are required to suppress the error which stems from channel mismatches.Consequently,the focus of this dissertation is on the compensation algorithms for M channel TIADCs.The main contributions are further explained below:Chapter 2 studies the operation principles of TIADCs and the effect of channel mismatches.In this chapter,the input-output equalities of TIADCs are formulated and the error spectrum is discussed,and the quantitative relationship between the dynamic performance of TIADCs and the mismatch level is derived.Besides,Models for frequency response mismatch error,static nonlinear mismatch error and dynamic nonlinear mismatch error are presented.The results can be used to analyze the mismatch type and strength in a TIADC,and can be used as reference for TIADC system design and compensation algorithm design.Chapter 3studies blind calibration methods for frequency response mismatches in Mchannel TIADCs and proposes an algorithm for frequency response mismatches based on channel swapping.In order to estimate the frequency response mismatch parameters,slight oversampling and channel swapping technique are adopted to extract the error signal,then mismatch parameters are estimated using LMS algorithm driven by the error signal.Then the frequency response mismatch error is reconstructed using the parameters and the error model.Simulation results show that the power of the frequency response mismatch error is suppressed greatly and the dynamic performance of TIADC is enhanced.Compared to existing methods,our method can applied to TIADCs with more channels,rather than restrict to two channels and four channels.Chapter 4 studies the foreground calibration methods for frequency response mismatches in M-channel TIADCs,and proposes a method based on training sequences.By exploiting the frequency response mismatch error model,the input output relationship in time-domain is expressed as linear equations.Then the parameters are estimated using training sequences and least square algorithm.In this method,the estimated frequency response parameters contain the mismatch part and the average part.Thus equalization is adopted before error reconstruction to enhance reconstruction accuracy.The calibration performance is verified by simulation results under different number of channels,different input signals and different training sequences.Chapter 5 studies blind calibration methods for nonlinear mismatches in M-channel TIADCs,and proposes calibration methods for static nonlinear mismatches and dynamic nonlinear mismatches respectively.The static nonlinear mismatches calibration method exploits oversampling to extract mismatch error from TIADC output,and then estimates mismatch parameters using LMS algorithm.The static nonlinear mismatch error is reconstructed using TIADC output and the error model.The dynamic nonlinear mismatch calibration method is built based on the simplified Volterra model for ADC nonlinearities,and the mismatch parameters are estimated using LMS algorithm.The mismatch error is suppressed by subtracting the reconstructed error from TIADC output.Simulation results show that the power of nonlinear mismatch error is reduced by the calibration method.Chapter 6 studies the foreground calibration methods for nonlinear mismatch in Mchannel TIADCs,and proposes methods for static nonlinear mismatch and dynamic nonlinear mismatch respectively.The method for static nonlinear mismatch expresses the input output relationship as linear equations,then estimate the static nonlinear parameters by solving the linear equations using least square method.Then the static nonlinear mismatch error is reconstructed and subtracted from TIADC output.The foreground calibration method for dynamic nonlinear mismatches exploits the simplified Volterra model for ADC nonlinearities to express TIADC input output as linear equations,then estimate the parameters by solving the linear equations using least square method.The dynamic nonlinear mismatch error is reconstructed exploiting the error model,the estimated parameters and TIADC output signal.Simulation examples validate the calibration performance of the proposed method under different simulation settings.