Sensitivity Minimization in Active RC Synthesis

Abstract

Since the majority of active RC synthesis techniques are based upon the partitioning of network functions into subnetwork functions, it was thought to be of interest to discover how this decomposition might be achieved to minimize sensitivity to the active element. This is, then, an extension of the work of Horowitz, Sipress, and others. Two types of active decompositions were considered: RC-NIC and RC-RL. It was shown that, regardless of the degree of the polynomial decomposed, when RC-RL decomposition is possible at all, an RC-RL decomposition can be found which results in a lower zero (pole) sensitivity figure than any RC-NIC decomposition. In the process of obtaining this result, the form of the RC-RL decomposition(s) which minimize(s) zero sensitivity is obtained. It is shown that, in general, there are an infinite number of these RC-RL decompositions with the same minimum sensitivities at all zeros (poles). This is in contrast to the Horowitz decomposition, which uniquely minimizes sensitivity in the RC-NIC case.