Layout-level synthesis of RF bandpass filter on organic substrates for wi-fi applications

A fast and accurate layout-level synthesis technique for RF passive circuits such as handpass filters composed of quasi-lumped embedded inductors and capacitors on multilayer laminate type organic suhstrate is presented. The proposed approach is based on a combination of segmented lumped circuit modeling and artificial neural networks (ANNs). A complex multilayer passive circuit has been segmented into uncoupled sections which are then modeled using lumped circuit models that ensure broadband correlation with EM based data. The complete circuit model which includes all parasitics is then optimized and mapped using polynomial relations to geometrical parameters such as width, length and spacing of the components. To control the insertion loss (IL) of the circuit, the inductor geometries are optimized based on area constraints using nonlinear mapping based on ANN. The approach has been verified for a 2-pole 3" X 3mm X 1.5mm, 9 component handpass filter centered at 2.45 GHz and scalability has been shown over a range of 2 3 GHz with bandwidth variation of 0.5-3% of center frequency (CF). The synthesized designs show good correlation with EM data.