Computational Magic And The EMC Engineer

By Dr. Glen Dash, Ampyx LLC

Using a computer to simulate EMC phenomena is a field full of promise.

In decades past, the EMC engineer had three basic methods for evaluating EM phenomenon: Maxwell's Equations, circuit models and fieldwork. Maxwell's Equations, solved for the boundary conditions and forcing functions at hand, serve as a practical tool only in relatively simple situations. Circuit models use simplifying assumptions to reduce radiated emissions problems to set of circuits. For example, to a first approximation an antenna can be modeled as a network of RLC circuits. Again, only relatively simple problems can be considered. Fieldwork provides real data, but is both expensive and time consuming. It is also subject to its own peculiar types of errors ranging from parasitics to broken cables.

The development of electromagnetic computational methods now provides us with another tool. In its current state of development, however, computational tools will not completely replace any of the methods above. Computer modeling of EM phenomenon in three dimensions requires a host of assumptions that make computational modeling a tricky business. To do it well, the engineer should not only have a working knowledge of Maxwell's Equations, but should be familiar with the equally complex field of numerical analysis as well.

There are three different modeling techniques typically used for EMC modeling, the Finite Difference Time Domain (FDTD) method, the Methods of Moments (MOM) and the Finite Element Method (FEM). Of these, the first two find the broadest application in EMC, although all three methods have their following.

The method we will be studying in this article is the Method of Moments, the method employed by the Numerical Electromagnetic Code (NEC) developed by Lawrence Livermore Laboratory. To use the Method of Moments, the user typically converts a conductive structure into a series of wires, creating a "wire frame model." These wires are then broken down into "segments," each segment being short compared to the wavelength of interest. Each of these segments will carry some current, and the current on each segment will affect the current on every other. To compute the currents on each segment, a set of linear equations is created and solved by the computer. Once the current on each segment has been calculated, both near and far fields can be calculated by superposition.