An efficient computer-aided design procedure for interpolating filter dimensions using least squares methods

This paper proposes a very efficient algorithm based on least squares (LSQ) methods to provide final dimensions of microwave filters. This algorithm just needs to start from very few accurate designs of practical filters, in order to provide the dimensions of any similar filter (same order and return loss values) with different center frequency f₀ and bandwidth BW. Variations in f₀ and BW have been studied to give models for the filter physical dimensions. Finally, both variables are included in a bivariate model allowing to directly obtain the filter dimensions for any combination of f₀ and BW of the filter.