IEEE TRANSACTIONS ON ELECTROMAGNETIC COMPATIBILITY, VOL. 42, NO. 4, NOVEMBER 2000 477 Short Papers_______________________________________________________________________________ Broad-Band Modeling of a Realistic Power Converter Shield for Electric Vehicle Applications Franck Briault, Marc Hélier, Dominique Lecointe, Jean-Charles Bolomey, and Richard Chotard Abstract—This paper deals with the broad-band modeling of a realistic power converter shield used in an electric vehicle. To validate such a model, comparisons between computation and measurements are required. As it is not practical to characterize experimentally the true radiation of a power converter through its enclosure, a reciprocal approach has been defined in which the enclosure is illuminated by a plane wave (receiving configu- ration). The associated experimental setup has been carefully defined and tested. At low frequency, the model is based on radiation through aper- tures in a simple cavity. At higher frequency, the Aseris-BE code, which is based on the boundary integral equations technique, has been used for computation. An optimal and realistic mesh of the enclosure suitable for the Aseris-BE code has been built by making numerous comparisons be- tween computation and measurements, which have shown how to improve the mesh. Index Terms—Aseris-BE code, boundary integral equation, electric vehicle, electromagnetic (EM) field simulator, power converter shield, shielding effectiveness. I. INTRODUCTION In the automotive industry, the handling of electromagnetic compat- ibility (EMC) problems has so far been mostly empirical and experi- mental. However, in an attempt to reduce the cost of EMC design and protection equipment, EMC problems have to be solved from the very first step of a new vehicle design. Therefore, the use of numerical tools has become increasingly common in the automobile industry in order to check the protection equipment and to estimate the radiated or con- ducted emissions and immunities. In the long run, the aim is to integrate the handling of EMC problems into the computer-aided design (CAD) of a vehicle. The EM problems encountered in a vehicle are very in- tricate [1]. The numerical tools contribute to predicting if the vehicle is below the acceptable limits for radiated and conducted immunity or emission, to evaluating the efficiency of EMC protections and to de- signing immune equipment. There are some potential victims of emis- sions from the converter: radio receivers, mobile phones, and micro- processors. These difficult problems can be solved with the help of nu- merical EM codes already developed for the military or for the aircraft industry. For example, PSA Peugeot Citroën uses the Aseris-BE code developed by Aerospatiale Company ([2], [3]). However, some modi- fications are required to apply such codes to specific EMC problems in automobiles in order to obtain more precise near-field results, even at low frequency (below 100 MHz) and to take into account metallic sur- faces with a finite conductivity. The proper description of the internal perturbation sources is still a major difficulty. In the case of the power Manuscript received May 19, 1998; revised March 21, 2000. F. Briault is with PSA Peugeot Citroën, DRAS/RVE, Centre Technique de Vélizy, 78140 Vélizy-Villacoublay, France. M. Hélier, D. Lecointe, and J.-C. Bolomey are with the Département de Recherche en Électromagnétisme, SUPÉLEC-CNRS, 91192 Gif-sur-Yvette CEDEX, France. R. Chotard is with PSA Peugeot Citroën, DETA/CAR/ELE/QEV-LG, 92250 La Garenne-Colombes, France. Publisher Item Identifier S 0018-9375(00)06651-5. converter, one issue is to estimate the field radiated by a converter and its enclosure. On the one hand, it is not feasible to study the radiation through the shielding of a fully operational converter located inside the vehicle. Such an experiment would require a very specialized range like a road simulator to simulate the varying load of the converter and obtain the nominal values of its currents and voltages. Moreover, the measured field in such a case is the field radiated by the whole vehicle and not the field radiated through the shielding of the converter alone. On the other hand, if only the shielding is considered, an equivalent set of antennas would have to be placed inside it in order to simulate the converter EMI. It is not mandatory to reproduce the actual inten- sities of the currents and voltages of the converter, as lower levels are acceptable using a linearity principle, but the field radiated by these antennas must have a frequency spectrum with the same shape as the original one. Similarly, the set of antennas should be chosen in order to replicate, as much as possible, the presumed field distribution and po- larization of the field inside the converter. The set of antennas should be fed but the field radiated in free-space by its feeding cables is un- wanted. Therefore, the antenna, its feed, and cables should be placed inside the shielding. At low frequency, it would be extremely difficult to find an equivalent efficient set of antennas and difficult to feed it because of the low value of the radiation resistance and efficiency of small antennas. If the feeding intensity and the radiated power are too low, the radiated field will not be measurable due to effectiveness of the shielding and the field will remain upper bounded but unknown. As the practical study of an emitting empty enclosure is not feasible (emission problem), the comparison has been made in a reciprocal configuration, with a receiving enclosure. In such a case, the shielding is illuminated by an EM simulator and works as a receiving device. By reciprocity, the shielding effectiveness will be determined by measuring the ratio of the illuminating field over the field induced inside the enclosure under test (EUT). This configuration, in which the shielding is illuminated from the outside, is the test configuration used for validating the Aseris-BE code since it allows comparisons between simulation and experiment. The goal of such an approach is to optimize the mesh of the converter shielding, but not to address the emissions of the converter electronic components. This paper deals with a broad-band model of a realistic power converter shield defined using this configuration and assumed to be a suitable basis for handling an emissions problem. The paper is organized as follows. After this introduction (Section I), the EUT is presented in Section II. It is illuminated by an incident plane wave gen- erated by an EM field simulator. The measurement of a low intensity field with a passive probe must be done carefully. The experimental procedure is described in Section III. Available theoretical and numer- ical tools and the associated models are presented in Section IV. The difficulty is to model the enclosure and, more specifically, to obtain an optimal mesh of it suitable for the Aseris-BE code. Section V is devoted to such an approach and presents comparisons between the theoretical and experimental results. Finally, Section VI is the conclusion. II. CONVERTER SHIELD The shield, studied from 10 MHz to 1 GHz, is the power converter enclosure used in the electric vehicle built by the PSA Peugeot Cit- roën company. This metallic shield is roughly a parallelepiped box, which contains the various power converters required in an electric ve- hicle. The average dimensions of this box are length mm, width 0018–9375/00$10.00 © 2000 IEEE