Miniaturized Microstrip Filter Design Using Active Learning Method

Abstract

Relating coupling and external quality factor of a filter to the physical parameters of the structure which is the final step of any filter design is usually complicated due to geometrical complexities of the filter, or in the case of microstrip resonators due to the lack of the exact solution for the field distribution. Therefore, common approach is using time consuming full wave simulations. In this paper active learning method (ALM) which is a fuzzy-based modeling technique developed by a procedure algorithmically mimics the information-handling process of the human brain, is proposed to overcome this drawback. Modeling steps of an unknown function using ALM will be described using an illustrative example. Afterwards, the modeling approach will be implemented to model coupling factor between two coupled spiral resonators (SRs) for two different coupling structures and external quality factor of the same resonator. Accuracy of the extracted surfaces is validated using two different criteria. Using the extracted surfaces; a four pole chebychev bandpass filter was designed and fabricated. Good agreement between the measured response and simulation validated the accuracy of the extracted surfaces again. Comparing the fabricated SR filter with a square open loop resonator (SOLR) one demonstrates more than 70% of filter area reduction.