Improving Accuracy in Solving Feldtkeller Equation

One major obstacle in filter and multiplexer synthesis is the challenge of accurately solving the Feldtkeller equation to compute the common denominator polynomial. In general, the numerical error can be detrimental to the direct computation using double floating precision when the network order exceeds 24. This letter introduces an iterative computational method for enhancing accuracy in solving the Feldtkeller equation without requiring extended precision in number representation. The proposed approach allows for the synthesis of high-order general asymmetric filters and multiplexers without the restriction of placing the reflection zeros (RZs) solely on the imaginary axis. Experimental results demonstrate the method’s ability to accurately compute equi-ripple filter and multiplexer characteristics with degrees up to 39.