Distributed Models for Filter Synthesis

Most of the circuit prototypes used in practical applications were proposed many decades ago, when the present computational resources were not available. Simplicity was a key factor in their conception. Although these traditional prototypes model the real structures, they usually neglect second-order effects, the frequency dependence of the filter parts and, in some cases, they do not have an exact correspondence with the physical structure. All of these inaccuracies can affect the equivalence between the prototype and the real filter, thus degrading the initial structure extracted from the synthesized prototype. Optimizations carried out from poor initial dimensions are computationally burdensome and more liable to provide a filter with unsatisfactory response. The synthesis methods reviewed in this article are based on prototypes designed to faithfully represent the real components. These prototypes include elements to model the relevant effects of the structure in the particular designs to be carried out. Moreover, accurate EM simulators are used to find an almost exact equivalence between the prototype and the filter parts. Although more elaborate synthesis techniques are usually required, these enhanced prototypes and synthesis procedures can be used to extract outstanding initial structures. The design is essentially solved after the dimensional synthesis of the component since, at most, only a slight and fast final refinement is needed (using an EM optimizer or the tuning elements included in narrowband filters). The examples shown in this article reveal the capabilities of these methods to overcome the limitations of classic circuit synthesis.