736 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 53, NO. 3, JUNE 2006 Inverter EMI Modeling and Simulation Methodologies Jih-Sheng Lai, Senior Member, IEEE, Xudong Huang, Member,IEEE, Elton Pepa, Member,IEEE, Shaotang Chen, Senior Member, IEEE, and Thomas W. Nehl, Fellow,IEEE Abstract—A numerical prediction of electromagnetic interfer- ence (EMI) allows evaluation of EMI performances at the design stage and before prototyping. It can also help reduce the post- prototype electromagnetic compatibility cost by minimizing late redesign and modifications of a drive implementation. This paper describes two simulation approaches with time- and frequency- domain simulations and verifies them with experimental results. Both time- and frequency-domain simulation approaches are found effective as long as the noise source and propagation path are properly modeled. The three-dimensional (3-D) finite-element- analysis (FEA)-based parasitic parameter extraction tool—Ansoft Spicelink has been used substantially. To gain additional degree of confidence, the results obtained from FEA are verified with closed-form solutions and actual measurements. Index Terms—Electromagnetic interference (EMI) modeling and simulation, inverter EMI. I. I NTRODUCTION E LECTROMAGNETIC compatibility (EMC) has long been regarded as a “black magic” approach in power electronics study [1], [2]. Recently, more and more mathe- matical modeling and analysis on electromagnetic interference (EMI) sources and propagation had shed a light on a better understanding of the EMI production mechanism [3]–[9]. Although ultimately the radiated EMI is the concern of the environment, the conductive EMI is considered as the source of the radiated EMI. Thus, for the power electronics design point of view, conductive EMI modeling and simulation study is most desirable. Unfortunately, this is a complicated and difficult task since such a task has to deal with not only the fundamental electrical circuit behavior but also the complex EMI generation mechanism within the power converter. In spite of its difficulties, the EMI modeling and simulation study has attracted considerable research attentions in the past few years. A number of publications in the power electronics area have proposed a variety of EMI models for dc–dc con- verters and three phase ac motor drives [8]–[13]. These studies Manuscript received May 11, 2004; revised January 24, 2005. Abstract published on the Internet March 18, 2006. This paper was presented at the 29th Annual Conference of the IEEE Industrial Electronics Society, Roanoke, VA, November 2–6, 2003. J.-S. Lai is with the Future Energy Electronics Center, Virginia Polytechnic Institute and State University, Blacksburg, VA 24061-0111 USA (e-mail: laijs@vt.edu). X. Huang is with Linear Technology Corporation, Milpitas, CA 95054 USA. E. Pepa is with Aker Wade Power Technologies, Charlottesville, VA 22901 USA. S. Chen and T. W. Nehl are with Delphi Research Labs, Shelby Township, MI 48315 USA. Digital Object Identifier 10.1109/TIE.2006.874427 rely mainly on empirical models to help analyze the complex EMI generation process. They are limited to explain only the particular observed EMI phenomenon and are not suitable for predicting other unknown EMI mechanisms. Many papers have been focused on the impact to the motor itself [14], [15], but not much on the inverter, which is indeed the EMI source that tends to create the issues such as bearing currents, shaft voltages, winding voltage spikes, etc. In the conventional design methodology, EMC issues are addressed only after a prototype is built. At that stage, the traditional EMC remedies are confined to adding extra com- ponents, metal shields, metal planes, or even redesigning the entire system, with a potentially significant impact both on the cost and on the time-to-market of the products. It is therefore commonly recognized that EMC must instead be addressed as early as possible during the design phase, and not after. A numerical prediction of EMI/EMC has the potential to ad- dress EMI issues at the design stage and before prototyping. It provides significant benefits by avoiding late redesign and modifications of the drive implementation, leading to shorted design times and reduced postprototype EMC costs. In practice, however, for inverters with a complex circuitry, the modeling efforts and computational load to deal with all parasitics for every component within the device may exceed the capability of today’s computers. A simplification can be made by focusing only on major EMI components and circuits. Such a simplification would require some expert knowledge of the inverter electrical behavior and basic EMI characteristics. For example, a distinction between the power stage and logic circuits in a motor drive will justify the focus on salient EMI related circuitry and components, thus leading to considerably reduced efforts in parameter extraction. In this paper, two EMI simulation methodologies are in- troduced: 1) time-domain simulation followed by fast Fourier transform (FFT) and 2) frequency-domain simulation or com- putation. The time-domain simulation approach is straightfor- ward but requires substantial computation resource and lengthy simulation time. A frequency-domain approach requires knowl- edge on noise source and propagation path but significantly reduces computational effort, and thus is recommended as the preferred approach. A partial inverter leg is thus used as an example to show how to apply the frequency-domain approach, and the results are verified with both time-domain approach and hardware experiments. Although pulse width modulation (PWM) schemes will affect the spectrum of EMI, their major impact is in the PWM frequency region, which is relatively low compared with the wide EMI frequency range [16]. Thus, this 0278-0046/$20.00 © 2006 IEEE