| Analog-to-digital converters(ADCs)are the bridges which connect the analog world and digital world,and are the cornerstone of the modern electronic systems.ADCs are widely used in radars,oscilloscopes,communication systems,signals intelligence,etc.With the continuous increment of instantaneous bandwidth of the signals,ADCs with higher sampling rate and higher resolution are urgently needed.However,under the existing technology,the sampling rate and the resolution of a single ADC are mutually restricted.In order to improve the sampling rate while maintaining the original resolution,the researchers proposed to use multiple relatively low-speed ADCs to form a timeinterleaved(TI)system.However,TIADCs are highly susceptible to channel mismatches which seriously influence the system's signal-to-noise-and-distortion ratio(SINAD)and spurious-free dynamic range(SFDR).This dissertation conducts researches on TIADCs' nonlinearity mismatches and bandwidth mismatches,proposes the corresponding blind calibration methods,and gives the application to the systems with intensive channels.The main contributions are further explained below:Chapter 2 studies the nonlinearity mismatches and proposes a low-cost blind calibration method.Firstly,the proposed method employs a slight ratio of oversampling to generate an input-free frequency band,and by using high-pass filters one can acquire the nonlinearity mismatch errors in this band to drive the least-mean-square(LMS)algorithm and then adaptively estimate the mismatch coefficients.Then,by conducting Hadamard transform,differentiation,and multiplication on TIADC's output signal,one can get the pseudo distortions.Afterwards,by weighting the pseudo distortions with the estimated mismatch coefficients and summing them up,one can reconstruct the mismatch errors.Finally,the calibration is realized after the reconstructed error is subtracted from the TIADC's output signal.The simulations testify the method's performance on the conditions of different input signals,different mismatch magnitudes,different nonlinearity orders,different channel numbers,different differentiator orders and different high-pass filter order,and the results show that the proposed method can effectively suppress the nonlinearity mismatch error and then improve the TIADC's dynamic performance.The comparison is made on resource consumption between the proposed method and the existing method,and it shows that the proposed method cost 15%?29%?38% fewer multipliers than the existing method in 4-channels,8-channels,and 8-channels TIADCs.Chapter 3 studies the nonlinearity mismatches and proposes a blind calibration method which adapts to input signals with arbitrary bandwidth.Firstly,the pseudo distortions are generated with TIADC's output signal.Then,the correlations between the TIADC's output signal and the pseudo distortions are used to drive the LMS algorithm and then adaptively estimate the mismatch coefficients.Afterwards,by weighting the pseudo distortions with the estimated mismatch coefficients and summing them up,one can reconstruct the mismatch errors.Finally,the calibration is realized after the reconstructed error is subtracted from the TIADC's output signal.The simulations testify the method's performance on the conditions of different input signals,different mismatch magnitudes,different nonlinearity orders,different channel numbers,different differentiator orders and different samples' numbers for correlation,and the results show that the proposed method can effectively suppress the nonlinearity mismatch error and then improve the TIADC's dynamic performance.The comparison is made between the proposed method and the existing method,and it shows that the proposed method breaks through the bottleneck that the existing method is only applicable to broadband signal,and has good applicability in the case of narrowband signal at the cost of a little slower convergence speed.Chapter 4 studies the bandwidth mismatches and proposes a low-cost blind calibration method.Firstly,the correlations between the adjacent channels' output signals are used to drive the LMS algorithm and then adaptively estimate the mismatch coefficients.Then,by conducting differentiation,time-delay and downsampling on TIADC's output signal,one can get the pseudo distortions of different channels.Afterwards,by weighting the pseudo distortions with the estimated mismatch coefficients and summing them up,one can reconstruct the mismatch errors of different channels.Finally,the calibration is realized after the reconstructed errors are subtracted from different channels' output signals.The simulations testify the method's performance on the conditions of different input signals,different mismatch magnitudes,different channel numbers,different differentiator orders and different samples' numbers for correlation,and the results show that the proposed method can effectively suppress the bandwidth mismatch error and then improve the TIADC's dynamic performance.The comparison is made between the proposed method and the existing method,and it shows that the proposed method cost 70 and 146 fewer general multipliers than the existing method for 4-channel and 8-channel TIADCs.Chapter 5 studies the application of the proposed methods above to intensive-channel TIADCs,and gives the strategy of efficient implementation.The calibration is conducted in two stages: In the first stage,the gain mismatch error is calibrated,and in the second stage,the timing and bandwidth mismatch errors are calibrated.In every stage,Hadamard transform is first utilized to generate the pseudo distortions,and then the correlations between the pseudo distortions and the TIADC's output signal are used to drive the LMS algorithm and then adaptively the mismatch coefficients are estimated.Afterwards,by weighting the pseudo distortions with the estimated mismatch coefficients and summing them up,one can reconstruct the mismatch errors of different channels.Finally,the calibration is realized after the reconstructed errors are subtracted from different channels' output signals.The simulations testify the method's performance on the conditions of different input signals.Aiming at the demand of real-time calibration,the efficient implementation strategy is proposed by using signed-LMS algorithm,simplified correlation,and full parallel differentiator. |
PDF Full Text Request