【作者】 王瑞；

【导师】 徐军；

【作者基本信息】 电子科技大学 ， 无线电物理， 2012， 博士

【摘要】 微波谐振腔滤波器和双工器是现代通信系统（如无线基站和通信卫星）的重要器件。对一个设计师来说，从给定的指标开始到最后获得一个典型的滤波器或双工器的物理实现，这往往是个漫长的路程。这个漫长的过程通常包括传输零点和滤波器阶数的确定、耦合矩阵的综合，为获得物理尺寸在EM设计阶段的诊断与调谐、加工与测试、和/或在产品阶段的诊断和调谐。本文研究的重点在谐振腔滤波器和双工器综合与诊断上。详细的研究给出如下：首先，提出了利用遗传算法（GA）的优化方法确定谐振腔带通滤波器（包括自均衡滤波器）的阶数和最优的传输零点位置。此方法是基于根据给定的幅度和/或群延时指标构造一个目标误差函数，然后利用遗传算法通过最小化这个目标误差函数得到滤波器的阶数和传输零点（TZs）。该方法解决了具有广义切比雪夫响应的谐振腔滤波器耦合矩阵综合的首要问题。此外，给出了基于单一的SolvOpt优化方法综合谐振腔滤波器的耦合矩阵，为寻找代价函数的全局最小值，给出了SolvOpt初始值的设置规则。第二，对不包含源与负载耦合的谐振腔滤波器的诊断，提出了两个诊断方法，包括三参数优化方法和五参数优化方法。该方法可以用于从窄带交叉耦合谐振腔带通滤波器的测试或电磁仿真的S参数中提取耦合矩阵和无载Q（也称为滤波器的诊断）。在提出的方法中，柯西法用于确定在归一化低通频域中滤波器S参数的特征多项式，测试（或电磁仿真）的S参数的相移效应首次通过优化移除，谐振腔的无载Q（假设所有的谐振腔具有相同的无载Q）也同时通过优化得到。第三，为准确地从有耗的交叉耦合谐振腔带通滤波器测试（或仿真）的S参数中提取耦合矩阵和每个谐振腔的无载品质因数（无载Q），提出了一个两阶段优化方法。该方法是基于提出的五参数优化方法正确扩展到能处理存在源与负载耦合和每个谐振腔有不同的无载Q的情况。第四，提出了一种通过使用导纳参数（也就是从所周知的Y参数）的新方法提取包含源与负载耦合的有耗交叉耦合谐振腔带通滤波器的耦合矩阵。Y参数是通过测试的S参数对应的特征多项式计算，然后谐振腔的无载Q以及耦合矩阵可以从这个Y参数中提取。该方法允许对一个包含源与负载耦合的滤波器进行诊断，而不需要处理Y参数的去极点问题和测试噪声问题。最后，提出了两种综合耦合谐振腔双工器的方法，该双工器由TX和RX滤波器构成（连接TX和RX滤波器的两种结节类型被研究）。一种方法是分别单独综合TX和RX滤波器的耦合矩阵（MTX和MRX），然后双工器的耦合矩阵通过MTX和MRX获得。另一种方法是利用提出的线性频率变换和混合优化方法，直接得到整个双工器的“N+3”耦合矩阵。提出的滤波器的诊断方法也可用于指导双工器每个信道的调谐。综合出的滤波器或双工器的耦合矩阵可用于分析这个滤波器或双工器的幅度和群时延响应，以及获得它们初始的物理尺寸。提出的诊断方法可用于指导交叉耦合的谐振腔滤波器或双工器在EM设计阶段和/或产品阶段的调谐，进而加快该滤波器或双工器的设计和物理实现，从而可以降低产品成本和节约时间。更多还原

【Abstract】 Microwave resonators filters and diplexers are important components in moderncommunication systems (such as wireless base stations and communication satellites).Starting from given specifications, it is often a long journey for a designer to obtain thefinal physical realization of a typical filter or diplexer. This long journey usuallyinclude the determination of transmission zeros and filter order, coupling matrixsynthesis, diagnosis and tuning in the stage of EM (electromagnetic) design forphysical dimensions, fabrication and measurement, and/or diagnosis and tuning in thestage of the production. This paper focuses on the study of the synthesis and diagnosisof resonators filters and diplexers. The detailed studies are given as follow:First, a method to determine the filter order and the optimum locations oftransmission zeros of resonators bandpass filters (including self-equalized filters) ispresented using genetic algorithm (GA). This method is based on constructing anobjective error function according to given amplitude and/or group delay specifications,and then the filter order and transmission zeros (TZs) can be obtained by minimizingthe proposed error function using GA. The method has solved the first problem ofcoupling matrix synthesis of general Chebsyshev coupled-resonator filters. In addition,a single optimization algorithm based on single SolvOpt that synthesizes couplingmatrices for coupled-resonator filters is presented. The rules for setting initial values ofSolvOpt are proposed to find global minimum of the cost function.Second, for the diagnosis of the resonators filters without source-load coupling,two diagnosis methods, including a three-parameter optimization method and afive-parameter optimization method, are presented. The methods can be applied forextracting the coupling matrix and the unloaded Q (also called filter diagnosis) fromthe measured or electromagnetic simulated S-parameters of a narrow bandcross-coupled resonator bandpass filter. In the methods, the Cauchy method is appliedfor determining the characteristic polynomials of the S-parameters of a filter in thenormalized low-pass frequency domain, the phase-shift effects of the measured (orelectromagnetic simulated) S-parameters are removed for the first time by optimization,and the unloaded Q (assuming all the resonators with the same unloaded Q) is also obtained simultaneously by optimization.Third, a two-stage optimization method is proposed for accurately extracting thecoupling matrix (CM) and the unloaded quality factor (unloaded Q) of each resonatorfrom the measured (or simulated) S-parameters of lossy cross-coupled resonatorbandpass filters. The method is based on the proposed five-parameter optimizationmethod with proper extension to handle the presence of source-load coupling andnon-uniform unloaded Q for each resonator.Fourth, a novel approach using admittance parameters (also know as Y-parameters)is proposed to extract the coupling matrix of a lossy cross-coupled resonator bandpassfilter with source–load (S-L) coupling. The Y-parameters are calculated bycharacteristic polynomials corresponding to the measured S-parameters, and then theunloaded Q of resonators and the coupling matrix can be extracted from theY-parameters. The method allows one to diagnose a filter with S-L coupling withoutnecessity of dealing with the degenerate poles problem of the Y-parameters and themeasurement noise.Last, two methods for synthesizing coupled resonator diplexers composed of TXand RX filters (two types of junctions connecting the TX and RX filters areconsidered). One method is that the coupling matrices of the TX and RX filters (MTXand MRX) are synthesized independently, and then the coupling matrix of the diplexerscan be obtained from MTXand MRX. Another method is that the”N+3” couplingmatrix of overall diplexer can be obtained based on the proposed linear frequencytransformation and hybrid optimization methods. The proposed filter diagnosismethods can also be applied to guide the tuning of each channel in the diplexer.The synthesized coupling matrix of a filter or diplexer can be used for analyzingthe amplitude and group delay responses of this filter or diplexer and obtaining theirinitial physical dimensions. The proposed diagnosis methods can guide the tuning inthe stages of the EM design and/or production of a cross-coupled resonator filter ordiplexer, and thus accelerate the design and the physical realization of this filter ordiplexer, which can reduce production costs and scheduling.更多还原