Search

An integral equation technique for the analysis of multiport H‐plane microwave circuits composed of an unlimited number of arbitrarily shaped metallic and/or dielectric elements is proposed. The structure under study is divided into different regions after the application of the surface equivalence principle. Following this approach, the access waveguide ports are modelled by means of parallel plate Green's functions, whereas the central region is characterised by 2D rectangular cavity Green's functions. The use of this kind of Green's functions represents a novelty for solving H‐plane integral equation problems. For the first time, the Ewald method has been employed in order to accelerate the computation of the 2D rectangular cavity Green's functions and their derivatives, needed in the integral equation. In addition, convergence studies show the numerical convenience of using the Ewald method to achieve a fast and accurate evaluation of these Green's functions. Results show that the analysis can be very efficient if the structure is segmented in a proper way based on fitting the cavity resonator within the geometry of the structure under analysis. In fact, all the walls of the structure that are coincident with the auxiliary resonator need not to be discretised during the numerical solution. Finally, several simulation examples of practical inductive microwave circuits are presented and discussed, showing a good agreement and higher computational efficiency when compared to results provided by commercial full‐wave software tools and measurements. [ABSTRACT FROM AUTHOR]

In this paper, neural networks are used to efficiently calculate the multilayered media boxed Green's functions needed in integral equation (IE) formulations. The analysis of complex multilayered shielded circuits is computationally time consuming, due to the need to calculate the multilayered-media boxed Green's functions. Using neural networks as radial basis function networks, the boxed Green's functions can be calculated quickly, thus greatly reducing the computational time associated with the analysis of practical circuits. Once the neural network is trained with a known set of pairs of inputs and outputs, new outputs are quickly calculated, thus increasing the efficiency of the IE method. © 2004 Wiley Periodicals, Inc. Microwave Opt Technol Lett 44: 61–66, 2005; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/mop.20548 [ABSTRACT FROM AUTHOR]

In this paper, the design of a novel planar narrowband microstrip filter is presented. The structure is designed and manufactured in microstrip technology, and it is formed by set of pairs of resonators of different lengths. The length of one resonator in each pair is adjusted so that a suitable odd resonance is tuned to the center frequency of the filter. The length of the other resonator is adjusted so that the next even resonance is tuned again to the center frequency of the filter. The different path of the signal in each resonator produces a cancellation of energy at a given frequency, therefore implementing a transmission zero in the insertion-loss response of the filter. First, this paper presents a design example of a single set of resonators (basic unit cell) working at the fifth and sixth resonances. Then, two basic unit cells working at the first and second resonances are first optimized, and then they are cascaded to show how higher-order filter responses can be implemented. The measured results confirm the theoretical predictions, and validate the new structures for high-selectivity applications. © 2004 Wiley Periodicals, Inc. Microwave Opt Technol Lett 43: 450–455, 2004; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/mop.20499 [ABSTRACT FROM AUTHOR]