RF Filter Design Of Measurement Receiver Based On Blind Signal Processing

For large-scale electrical equipment or fixed equipment that requires on-site testing,such as high-speed EMU,aircraft,LED display screen,etc.,it is impossible to complete the radiation emission test in anechoic chamber.However,if the test is performed in the field,the electromagnetic signal emitted by the equipment under test(EUT)and the background noise of the environment are mixed together and are received by the measuring antenna,resulting in invalid measurement results.To solve this problem,a dual-channel measurement receiver can be used to acquire the mixed signals,and then the blind signal processing technique can be used to extract the EUT radiation electromagnetic signals from the mixed signals.In this thesis,the filter and low noise amplifier(LNA)of the receiver RF front-end module used in the field measurement of radiation emission are designed and the separation of the received mixed signal is realized by the blind signal separation algorithm.Radio frequency(RF)front-end filters for radiation field measurement not only require good amplitude characteristics,but also have good phase characteristics to reduce signal distortion.In this thesis,we derive the expression of filter transfer function theoretically and study the influence of transmission zeros and different approximation functions on amplitude characteristics and phase characteristics.The results show that the amplitude characteristics and phase characteristics of the filter designed by the traditional approximation function are mutually restricted.However,the introduction of transmission zeros at different positions in the filter optimizes both amplitude and phase characteristics.Based on the theoretical research,a band-pass filter with pass band of 30MHz?1GHz and a low-pass filter with a cut-off frequency of 10MHz are designed.The parasitic characteristics of the filter at high frequencies are studied.The test results show that the filter performance meet the requirements.On the basis of chip selection,a LNA was designed by weighing the noise figure and gain.The measured results show that the gain and gain flatness of the developed low noise amplifier meet the performance requirements.For RF front-end circuits of the dual channel measurement receiver,device differences,PCB manufacturing,welding process will result in the mismatch between the two channels.Finally,the thesis studies and compares the separation effect of the algorithm based on signal to noise ratio and the algorithm based on joint approximative diagonalization of eigenmatrix on the mixed signal under dual-channel mismatch conditions.For blind separation algorithm based on signal-noise ratio,dual-channel inconsistency has a serious impact on the separation factor of the mixed signal.This thesis propose a method to balance the dual-channel difference in the frequency domain by using FIR filters.The FIR filter is used to equalize the frequency domain dual-channel mismatch.After the channel equalization,the separation coefficient for the mixed signal can be greater than 0.98.However,this method requires known channel characteristics.The blind separation algorithm based on joint approximative diagonalization of eigenmatrix regards the delay caused by channel inconsistency as a part of the mixed matrix and can accomplish the blind signal separation under dual-channel difference with the separation coefficient above 0.95.This algorithm is only suitable for the separation of narrowband signals,but its implementation is simpler than the algorithm based on signal to noise ratio.