TEM modelling of comb-line filter bandwidth expansion

Abstract

Combline filters frequently show w-ider bandwidth then initially synthesized and analyzed. The main cause, as presented ig the paper, is distortion of the TEM electromagnetic fields at the resonators open ends. These distortions depend on the three dimensional structure of thn region, requiring heavy computation for 3-D field analysis. A modification to the classic equivalent network of combline filters , which represents this effect is described and formulas to calculate its elements are given, which enables good approximation of the filter's actual performance by using inexpensive network analysis tools.Part 1:Introduction Combline and interdigital filters, are synthesized and analyzed assuming that the filter structure is a h t e length section out of an infinite. length multi coupled TEM line array with appropriate end conditions (i.e.ground shorts and open ends reactive loading) . Therefor, the mathematical approach of classic. synthesis and analysis of these filters, is done assuming that the Maxwellian capacitance matrk is uniform through the entire length of the resonators. The classic equivalent netuorks of these filters were derived based on the validity of t h s assumption [ 1-41. Practically there are distortions in the electromagnetic field lines pattern at the open ends of the resonators. In this region the electric field pattern is not perpendicular to the resonators [ fig. 11, creating a TM propagation mode. he approach presented herein approximates the 3-D field problem as a deviation from the original TEM values. Therefor the values of the Maxwellian capacitance matrix vary along the coupled lines towards the open end region of the resonators. Combline filters are highly sensitive to this effect, resulting mainly in bandwidth expansion, sometimes up to twice the synthesized (or analyzed) bandwidth. The classic equivalent networks do not include representation of this phenomenon, thus analysis of these networks will not show this effect.