Sensitivity Analysis of Microstrip Line Parameters on Total Radiated Power Through Surrogate Modelling

EMC standards are defined at system level. A system is composed of equipment that include printed circuit boards. Lowering the total radiated power by the latter contributes to minimize the risk to overcome the radiated emissivity limits at system level. We aim at optimizing printed circuit board design with regard to total radiated power while fulfilling other constraints such as signal integrity. Impact of electrical tracks topology on radiated field should be controlled. Electromagnetic solvers help to do so, but deal with some shortcomings. First, the output of a model of radiated power from a conductive track must be close enough from the actual radiated power of the corresponding printed track. Second, a parametric study is out of reach since we face a large number of variables. This paper presents a methodology to overcome these obstacles. We first assume that the model response and measurements are sufficiently correlated. We check this assumption for a few test cases. Then, the radiated power of a microstrip line is evaluated according to its parameters and frequency through a surrogate model built from a limited set of simulations. At the end, critical parameters are identified. We apply this methodology to a simple scenario for illustration.