Hybrid modeling of electromagnetic interference on circuit and structure topologies.

Abstract

The topic of electromagnetic interference and compatibility (EMI/EMC) has been pursued for some time. However, much of the effort has focused on the analysis of simplistic structures, with realistic electronic systems primarily investigated via experimental means. More specifically, the situation of some structure enclosing an electronic device has yet to be considered via numerical or analytical methods. Moreover, design techniques aimed at mitigating EMI/EMC coupling have not been considered before. This dissertation considers the adaptation, further development and hybridization of numerical methods such as the moment method and finite element-boundary integral (FE-BI) technique for the analysis and design of electromagnetic interference on structures enclosing passive and active circuit topologies. Several experiments are also conducted to validate the proposed analysis methods. In this dissertation, we employ the moment method (MoM) in conjunction with curvilinear bi-quadratic basis functions and accelerated via the multilevel fast multipole method for the analysis of EMI on complex and realistic surface geometries. Several example applications of this analysis method are presented to illustrate the EMI effects onto cavities, automobile chassis and to also demonstrate EMI mitigation designs. For the analysis of EMI onto passive and active analog circuits, we employ the hybrid finite element boundary integral (FE-BI) algorithm. For this unique adaptation of FE-BI, the numerical algorithm is coupled with the harmonic balance method to investigate operational dysfunctions and non-linearity effects on an analog circuit when subjected to external interference. An important component of the dissertation is the integration of the associated numerical solvers with fast parameter optimization tools to mitigate EMI/EMC. For our optimization algorithms, we employed two approaches (new to the electromagnetic community) to create a reduced order model (using a dataset from simulation) for carrying out design in the presence of complex geometries. Several optimization examples are given showing a design convergence that is both fast and robust. A significant contribution of this dissertation lies in the integration of various modeling techniques to analyze increasingly complicated geometries. In this regard, this dissertation demonstrates the potential of hybridizing appropriate modeling approaches towards the analysis and design of realistic EMI/EMC problems.