Optimization of the frequency characteristics in saw filter design

Abstract

The transmission characteristics of surface acoustic wave (SAW) filters is considered. In the process of the design, the bandpass characteristics is requested to meet a specification in the process of the design. The apodization in the input interdigital transducer (IDT) is adopted to achieve the desired characteristics of the frequency bandwidth and the attenuation of the passband for the central frequency given. In the present paper, optimization procedures are discussed and some demonstrations are presented. The frequency characteristics of a SAW filter is characterized by the delta function modelling. The apodization is made by varying the length of each electrode with the input IDT. The objective function is defined as the error norm between the required frequency characteristics and the calculated one for the assumed length of each electrode. The calculation is repeated to minimize the objective function. Three optimization methods, the Least-Mean-Square (LMS) method, the Davidon-Fletcher-Powell (DFP) method and the genetic algorithm (GA) are examined for this purpose. The paper demonstrates the capability and efficiency of the above three approaches for the SAW filter design with a couple of worked examples.