Research On Performance Analysis And Calibration Method Of Sampling Model Based On Parallel Architecture

Signal acquisition and processing technology has a wide range of applications in many fields of the electronic information industry,and one of the core components of acquisition technology is analog-to-digital converters(ADCs),the performance of which has a direct impact on the overall performance of the system.Due to the rapid advancement of digital technology,the signal acquisition system is continuously expanding its capacity to handle broader signal bandwidths and demanding higher levels of accuracy.The performance of a single-chip ADC is no longer enough to meet the growing demand for high-speed and high-precision signal acquisition.Therefore,the parallel sampling technique has been purposed.Nevertheless,the performance of the parallel acquisition system is significantly compromised by the noise and inconsistency error that occur among the channels due to the system’s multiple channels.As a result,it is required to analyze the elements that influence the performance of the parallel acquisition system,and then investigate the calibration algorithms to improve the acquisition system’s accuracy and performance.This dissertation focuses on studying the performance of parallel acquisition systems by selecting Time-Interleaved ADC(TI-ADC),Quantization-Interleaved ADC(QI-ADC),and Hybrid Interleaved ADC based on TI-ADC and QI-ADC as the research objects.The performance of the parallel acquisition system and the high-precision processing algorithms are investigated as follows:(1)On the basis of an analytical model of the TI-ADC system’s mismatch error,an algorithm for estimating the gain versus time skew mismatch of the TI-ADC system is proposed.Specifically,a set of sinusoidal signals of different frequencies is used to estimate the gain and time skew mismatch with respect to the input frequency.For the gain mismatch,the information is extracted by calculating the autocorrelation function for each channel output.The time skew mismatch is estimated by utilizing the particle swarm optimization(PSO)algorithm.The algorithm determines the value of the time error by finding the optimal particle.The estimation method can accurately estimate the value of the mismatch as it varies with the input signal,which provides a foundation for subsequent calibration work.(2)A TI-ADC background calibration algorithm based on the frequency shifting technique is suggested to address the three categories of typical mismatch issues in narrowband signal sampling.Analyzing the relationship between the TI-ADC mismatch error and the input signal leads to the calibration method of utilizing the frequency-shifted signals to reconstruct the channel mismatch.By employing appropriate weighting factors,the frequency-shifted signal may be multiplied to reconstruct the mismatch error signal,as both signals have the same frequency component.Specifically,the Hadamard transform is adopted to generate frequency-shifted signals,which can simplify the signal processing procedure.The estimation of the weight coefficients is achieved using the Least-mean Square(LMS)adaptive algorithm.The advantage of the proposed calibration method is that it does not require any filters,thus considerably decrease the complexity of the calibration circuit.(3)A foreground calibration technique is proposed for the frequency response mismatch(FRM)calibration of TI-ADC systems.Based on the theoretical analysis in(2),TI-ADC systems is modelled by the Hadamard transform,and then a bank of timeinvariant filters is designed to extract the mismatches.Specifically,this scheme uses the Hadamard transform instead of the complex exponential modulator,which can achieve comparable performance to conventional calibration methods and further simplifies the modulation process.(4)The influence of the QI-ADC system on acquisition system resolution is explored,as well as the effect of the channel offset error on QI-ADC system performance.A hybrid interleaved acquisition system based on TI-ADC and QI-ADC is introduced,combining the characteristics of both systems.Due to the incompatibility of the sampling rate and resolution of the acquisition system,the optimal acquisition performance is analyzed in conjunction with the characteristics of the hybrid interleaving system.This study examines the optimal configuration problem of the acquisition system,taking into account the combined effects of input noise and quantization noise.It derives the conditions that result in the hybrid acquisition system achieving optimal acquisition performance.It also provides a theoretical foundation for the specific design of the acquisition system.(5)Based on the conclusions of the above analysis of the optimal configuration of the hybrid interleaved system,the feasibility of noise decomposition for acquisition systems is explored.A noise decomposition method for the acquisition system is then introduced,which involves employing a Convolutional Neural Network(CNN)to identify and categorize each noise component present in the system noise.Additionally,a Long Short-Term Memory(LSTM)neural network is utilized to estimate the power of each noise component.The efficacy of this approach is confirmed through training and testing.