Non-iterative SAW filter design using convolution techniques

It has been previously shown that an optimum, equi-ripple filter design can be achieved using the Remez algorithm design program. However, SAW filter requires the design of two tra~sducers whose time convolution yields the desired filter response. This has led to zero-extraction techniques which deconvolve filter design into the two required functions. The zero-extraction technique is both time consuming and often produces marginally implementable time functions. This paper presents a new, non- iterative, analytically exact method for deconvolving filter design based on truncated cosine series into the two time functions required for SAW filter design. The approach is simple, fast and results in transducers having equal time lengths. This design method produces results comparable to the Dolph·Chebyshev technique. The theoretical approach and design examples are presented.