Novel modeling and layout optimization technique for highly compact planar bandpass filters on LCP

This paper presents a novel and highly compact design of planar microstrip narrowband handpass filters with insertion Io s s of 1.5-3dB and a fractional bandwidth of 4.12% in 5-6GHz center frequency ranges. It only takes an area of -6x1.5mm´on low cost Liquid Crystal Polymer substrate. In order to reduce theoretical complexities in actual design, performance data (center frequency, bandwidth and insertion loss) of the filters have been mapped against interdependent layout parameters using neural networks. The worst-case prediction error has been determined to he 5.93%. Sensitivity analysis has been performed to determine the impact of each layout parameters. Layout optimization is performed on the hasis o f response surface plots of performance parameters. The proposed modeling and optimization algorithm can he used for a design tool of complex microwave structures.