We propose a realizable RCLK-in-RCLK-out parasitic reduction technique. The method employs generalized Y-¡ transformation. In our method, admittances are kept in their original rational forms of ¢ , and their orders are reduced by truncating high-order terms. Therefore reduced admittances match the low-order terms in exact admittances. First-order realization of admittances is guaranteed, and higher-order realization is achieved by template optimization using Geometric Programming. The algorithm uniquely uses common-factor identification and cancelation operations to make Y-¡ transformation numerically stable. The experiment shows that our method can achieve higher reduction ratio than TICER and comparable simulation results with PRIMA.the expansion of determinants. However, enumerating all the trees in a large network is very difficult.Recently Ismail proposes DTT to approximate transfer functions in tree-structured RLC networks by direct transfer-function truncations [10]. The transfer functions are kept in the rational form so that low-order moments are matched implicitly. Since a truncated transfer function may not be a positive real function, the method is not compatible with general simulators either.Unsatisfied with these limitations, we devise a new RCLK-in-RCLK-out reduction method based on Y-£ transformation [11] [12] for general RCLK-VJ 1 linear networks. The idea is that, given such a network, we perform Y-£ transformation on all internal nodes in an efficient order. Y-£ admittances in each transformation are kept in the rational form, and admittances whose order is higher than a threshold¨will be truncated. Unlike topological analysis and other symbolic approaches, the new reduction method only evaluates the first¨terms in the denominator and numerator of £ admittances. 2 Discarded high-order terms, however, do not affect the preservation of the low-order terms, i.e., the low-order terms are precisely the firsẗ © U 7 ¤ ' 7 © 7 W V © m ¤

No abstract

No abstract

No abstract

Abstract-We propose a realizable RCLK-in-RCLK-out parasitic reduction technique. The method employs generalized Y-¡ transformation. In our method, admittances are kept in their original rational forms of ¢ , and their orders are reduced by truncating high-order terms. Therefore reduced admittances match the low-order terms in exact admittances. First-order realization of admittances is guaranteed, and higher-order realization is achieved by template optimization using Geometric Programming. The algorithm uniquely uses common-factor identification and cancelation operations to make Y-¡ transformation numerically stable. The experiment shows that our method can achieve higher reduction ratio than TICER and comparable simulation results with PRIMA.