Research On Key Techniques Of Arbitray Wavefrom Synthesis Baesd On FI-DAC

Arbitrary waveform synthesis technology plays a vital role in diverse fields because of its flexibility and convenience of generating various signals.However,the arbitrary waveform synthesis technology using a single digital-to-analogy converter(DAC)cannot meet the requirements of signal bandwidth due to the limitations of chip manufacturing technology.Nowdays,multi-channel parallel synthesis architecture is a promising approach to overcome these limitations that can increase the output signal bandwidth of the arbitrary waveform synthesis system.Generally,time-interleaved DAC,analog multiplexing DAC,and frequency-interleaved DAC(FI-DAC)are three commonly accepted parallel synthesis techniques.The time-interleaved DAC is often limited by the zero-order hold characteristics of DAC,making it difficult to effectively improve the output bandwidth.The analog multiplexing DAC is also difficult to implement dut to its strict requirements on system clock.The FI-DAC can avoid abovementioned problems by segmenting the wideband signal in the frequency domain,making an optimal option for increasing the bandwidth.With the aim of breaking through the limitations of DAC and improving the output bandwidth of the signal synthesis system,this dissertation focuses on modelling the arbitrary waveform synthesis system based on FI-DAC(FI-AWS),analyzing the structural defects and error effects of the system,and investigating defect improvement and error correction based on the principle of FI-DAC.The main contributions of this dissertation are as follows:(1)Research on the error analysis of FI-AWS system.Initially,the ideal output model is established according to the signal synthesis system with an ideal M-channel parallel FI-DAC.On this basis,the error source analysis is carried out,the influence of errors on the output signal and the correlations between errors are discussed,establishing a theoretical foundation for error correction research.(2)To solve the problem of fixed sampling rate in the FI-AWS system,the sampling rate conversion(SRC)filter based on the Farrow structure is applied to match the sampling rate of the input signal with the fixed sampling rate of the system.For the implementation of SRC filters in the hardware system,a multi-channel parallel sampling rate conversion structure is proposed to ensure high-speed and error-free.For the filter coefficient solving problem,an approximation method based on "imaginary and real separation" is proposed to convert the complex domain solving problem into a simple pair of linear programming problems.On this basis,the relationship between sampling rate conversion resolution and the number of filter banks and filter order is discussed.Meanwhile,a filter group parameter optimization algorithm based on performance enhancing is proposed to ensure that the required resolution SRC filter can be realized with minimal hardware resource.Experimental results show that the Farrow structurebased SRC filter can effectively solve the problem of fixed sampling rate in the FI-AWS system;the proposed filter parameter optimization algorithm can save about 40% of hardware resource while achieving the same conversion resolution.(3)Research on correction strategy based on magnitude and phase errors for subchannel errors in the FI-AWS system.To estimate the channel error,a magnitude error estimation method based on the self-power spectrum of the output signal and an integrated phase error estimation based on linear fitting are proposed.For the discretization problem of the magnitude-frequency response error correction coefficients,the mapping relationship between random frequency point and the discrete correction point is investigated.The nonlinear phase error is corrected by all-pass filter,the coeffivient solving problem is then transformed into a nonlinear optimization problem.An improved genetic algorithm-based nonlinear optimization algorithm is proposed to solve the problem of low accuracy in existing methods.Experimental results show that after subband channel error correction,the output magnitude flatness of the constructed FI-AWS system is within 2 d B;the time delay deviation and initial phase deviation of sub-band channel＿2 reach-0.012 ns and 0.087 rad,respectively;and the nonlinear phase errors of the two sub-band channels are also effectively improved.(4)Research on the two aspects of error correction methods for the overlapping errors in the FI-AWS system.The first approach is to introduce the signal components with the same frequency as the image component in the digital domain.In the final output,the introduce component can offset against the image component under the conditioning of digital pre-equalizer.In this method,a multiple-input and multiple-output(MIMO)model is developed for the correction of the aliasing error,and an expression is obtained for the frequency response of the digital pre-equaliser with respect to the sub-band channels.Meanwhile,the relationshiop between phase and magnitude differences of the signal in the aliasing region is analyzed to transform the aliasing error correction problem into a magnitude correction problem.Eventually,an FIR filter with linear phase was applied for the design of pre-equalizer.In the second approach,a guard band in the subband channels is utilized to avoid the image component in the transition band,eliminating the aliasing error.In this method,the setting principle of the guard band is discussed to deploy relevant parameters of each sub-band channel.Experimental results show that the aliasing error correction method based on MIMO can effectively improve the aliasing errors on the output signal magnitude to ensure that the output signal magnitude has flatness in the entire frequency domain.However,it is difficult to solve the problem of output spurious introduced by the aliasing errors under the influence of nonlinear phase errors.The aliasing error correction method based on guard band can radically avoid aliasing of sub-band signals,improving the quality of the output spectrum,but at the expense of the system output bandwidth.(5)An experiment platform for FI-AWS system with two parallel 900 MSPS sampling rate DACs was set to verify the proposed error correction technique.The FIDAC signal synthesis system with error correction effectively increased the output bandwidth almost 1.918-fold.Based on the results of this dissertation,the proposed sampling rate conversion technique with high-speed and high-resolution,sub-band channel error correction technique,and aliasing error correction technique can effectively improve the errors in the FI-AWS system,enabling the system to achieve ultra-wideband signal synthesis by multiplying the output bandwidth on the basis of the performance indicators of a single DAC chip.