Investigation On Diagnosis And Tuning Method Of Miniaturized Microwave Filters

The synthesis and design methods of microwave filters have been well established.However,full-wave optimization when implementing filters is still very time-consμming.In this dissertation,the concerned methods of filter synthesis can be divided into two types.One type is the coupled resonator filters synthesized with the theory of coupling matrix while the other type is the commensurate transmission line networks synthesized with the quasi-lμmped element method.This dissertation is mainly focused on the diagnosis and tuning methods of these two types of miniaturized microwave filters.（1）For the coupled resonator filters,the new extraction method is proposed for the coupling matrix by combining the conventional analytical and optimization methods.At first,the rational polynomials are obtained with the vector fitting method from the filter response.And the coupling matrix in the form of transverse filter is obtained.Then,the rotation angles of coupling matrix are taken as the variables of optimization when the transverse matrix is regarded as the initial state.The matrix with the given coupling topology can be optimized to approach the desired filtering response.The difference between the extracted and target matrices can be used to guide the tuning of filters empirically.As a sample to validate our proposed method,the diagnosis and tuning procedure is carried out for a 6^th-order generalized Chebyshev substrate integrated waveguide filter with a pair of finite transmission zeros,where the coupling matrix is extracted and corrected iteratively.（2）For the commensurate-line filters synthesized within the Richard's transformed domain with rational polynomials,a new method is proposed based on the zero-pole extraction and the optimization with space mapping,where the sensitivity matrix of zeros and poles to the physical parameters is utilized.At first,the rational polynomials are extracted from the full-wave simulated response of a commensurate-line bandpass filter within the transformed domain.Then,the sensitivity matrix of poles,reflection and transmission zeros to the critical physical dimensions is established nμmerically.When giving the specific target functions,the coarse model is constructed as the first-order approximation.The physical dimensions are optimized and the results of modified coarse model are compared with those of the full-wave simulated fine model.Finally,the automatic design of commensurate-line filter is accomplished.As a sample to validate the proposed method,a 7^th-order generalized Chebyshev microstrip filter is designed.By correcting the locations of its zeros and poles,the optimized response can be achieved automatically and iteratively.