Ensuring passivity and automatic order selection for global rational approximation macromodeling

Today digital and wireless designs demand both increasingly higher frequencies and better performance. As overall model complexity grows, greater importance is placed on the development of accurate, efficient interconnect simulation tools. One common approach is to use frequency domain macromodels of the network. This paper concentrates on the global rational approximation macromodeling technique as in the method developed by Elzinga (Elzinga, 2000). Here the technique is improved upon by enforcing passivity for macromodels where only minor passivity violations occur. In addition, an improved automated order selection method for the rational polynomial was developed and applied to the specific program of (Elzinga, 2000) to test the method