Optimum Topology for Power Handling in Dual Bandpass Filters

The design of microwave filters begins by deriving an equivalent filter network that approximates the desired filter response. The technique for the synthesis of single band filters is well-established. However, the new generation of advanced wireless systems demand filters capable of supporting multiple passbands. A linear optimisation technique for the synthesis of dual bandpass filters is presented in this thesis. It is applicable to the direct synthesis of symmetric and asymmetric transfer functions in the lumped or distributed domain. The technique is extended to allow synthesis of multi-bandpass filters with improved numerical accuracy. It allows arbitrary placement of transmission zeros, providing full control over the stopband rejection levels. It also permits large band-to-band separation due to the frequency-variant nature of the inter-resonator couplings. Synthesis examples are given to demonstrate the technique. Combline realisations of a 7th-degree dual bandpass filter and a 10th -degree triple bandpass filter are designed, fabricated and tested successfully to validate the theory. The power handling capacity of microwave filters is limited by the internal peak electric fields that can cause multipaction or ionisation breakdown. The peak time-averaged stored energy in the equivalent filter network is related to the internal peak electric fields. A given power transfer function can be realised using multiple filter topologies. Although the total time-averaged stored energy in the filter network is constant, its distribution varies depending upon the chosen topology. This part of the research focuses on establishing guidelines that help predict the optimum power handling topology for a given transfer function. The k-means clustering algorithm is implemented to discover patterns that aid prediction of the filter topology that demonstrates the lowest peak time-averaged stored energy for a given transfer function. Key characteristics of large data sets comprising of several single band transfer functions and corresponding topologies are determined by analysing the results of the clustering algorithm. The application of this algorithm has further been extended to discover patterns in the stored energy distribution of filter topologies realising dual bandpass transfer functions. In order to validate the obtained patterns, a comparison of the peak electric fields in the EM simulation of two filter topologies realising the same dual bandpass power transfer function is provided. The two topologies are such that: 1) a topology that constitutes the optimum topology as determined using the established patterns, and 2) a topology typically chosen by a microwave filter engineer for the transfer function at hand. Simulated results were in good agreement with the theory.