Practical Papers, Articles and Application Notes Robert G. Olsen, Technical Editor

EMC Applications for Expert MININEC

Abstract

A brief description is given of the Expert MININEC wire antenna modeling code. Two examples of the EMC applications of this code are described.

Expert MININEC Description
The sample problems described in this paper were analyzed using Expert MININEC Classic. Expert MININEC Classic is available for free and can be downloaded from the following web site:
https://www.emsci.com/

Expert MININEC Classic is a limited version of the more general codes available on the same web site. Nevertheless, it is a powerful code useful for many practical problems such as the ones described in this paper. For example, it allows the use of up to 500 wires and 1250 unknowns; far more than needed to solve the problems described here. Other limitations of Expert MININEC Classic will be explicitly indicated within the text of this paper.
Expert MININEC is an advanced engineering tool for the design and analysis of wire antennas. The process of solution begins with several assumptions that are valid for thin wires. These assumptions include that the wire radius is very small with respect to the wavelength and is very small with respect to wire length. Because it is necessary to subdivide wires into short segments, the radius should also be small with respect to the segment length, so that currents can be assumed to be axially directed (i.e., there is no azimuthal component of the current).
Expert MININEC solves for currents on thin wires using a Galerkin procedure applied to an electric field integral equation. The electric field is formulated in terms of its scalar and vector sources. These sources are the vector magnetic potential and the scalar electric potential. The two potentials can be calculated from potential integrals, which are solutions of the Helmholtz vector and scalar wave equations. In the potential integrals, the integrands are the wire current and wire charge distributions. The current and charge are linked via the equation of continuity. Expert MININEC makes use of the boundary condition on tangential electric fields at the surface of a perfect conductor, namely that the electric field must be zero. Since the wires are assumed to be thin, this forces the total axial electric field on the wire to zero. The three sources of the tangential electric field on the wire are:

• Currents and charges on the wires and on nearby wires.
• Incoming waves from distance or nearby radiators.
• Local sources of electric field on the wire.

The local sources are usually in the form of voltage sources, current sources, or transmission lines that connect to the wires. By summing the tangential electric field components at each segment on the wire and enforcing the zero total value, an integral representation for the currents and charges is obtained.
The integral equation relating the tangential electric field at the surface of a perfect conductor and the vector and scalar potentials is

(1)

where

(2)

and

(3)

are the vector and scalar potentials respectively. The integration is along the length S of the wire and

(4)

r is the distance from the source point of the current to the observation point of the field. The integration is over the angular variation around the wire. From the continuity equation, the linear charge density is

(5)

Application of the method of moments to this formulation results in an unusually compact and efficient computer algorithm. A matrix equation is generated that is used to solve for the currents on the thin wires.
The user interface to Expert MININEC is through Microsoft Windows. Input data screens provide format sensitive entry boxes in individual windows with tabular data displays. Expert MININEC modeling geometry constructs include: (An * indicates that this feature is not available in Expert MININEC Classic.)

• Cartesian, cylindrical and geographic coordinate systems
• Meters, centimeters, feet or inches selection
• Straight, curved*, helix*, spiral*, and catenary* wires
• Wire meshes*
• Automated canonical structure meshing*
• Node coordinate stepping
• Symmetry options*
• Rotational and linear transformations *
• Numerical Green's Function*
• Automated convergence testing*

Electrical description options include:

• Free space, perfect ground, and imperfect ground environments
• Frequency stepping
• Passive circuits*
• Transmission lines*
• Voltage and current sources
• Plane wave source excitation*

Solution description options include:

• Near fields
• Two-port coupling
• Medium wave array synthesis*

In addition, the Expert MININEC includes a user-oriented capability to analyze finite arrays within the limits of Expert MININEC capabilities.*
Output products are displayed in both tabular and graphics forms. The integrated graphics of Expert MININEC include:

• 3-D geometry displays with rotation, zoom and mouse support.
• 3-D currents, charges and pattern displays.
• Linear, semilog and log-log plots of currents, coupling, near fields, impedance and admittance.
• Smith Chart plots of impedance and admittance.
• Linear and polar pattern plots.

Input and output data screens are fully interfaced to Windows printer drivers as well as other window applications, such as word processors and spread sheets. On-line, context sensitive help is also provided.
The computational intensive algorithms are implemented in FORTRAN for greater speed and make maximum use of available memory to set array sizes. The formulation has been changed from earlier versions of the MININEC to use triangular basis functions. This results in greater accuracy. The short segment limit is a function of machine accuracy. Square loops and Yagi antennas may be solved with confidence. In addition, a Fresnel reflection coefficient approximation improves the calculation of currents in the vicinity of real ground for wire segments more than one-tenth of a wavelength from the ground. As a summary Expert MININEC solves for

• Currents and charges on wires (peak or RMS)
• Impedance, admittance, S11 and S12
• Effective height and current moments
• Power losses and load voltages *
• Multi-port (antenna-to-antenna) coupling
• Near electric and magnetic fields
• Radiation patterns (dBi or electric fields, power or directive gain)
• Medium wave array design *
• Auxiliary calculations of ground wave, stub matching, and tower footing impedance*

Expert MININEC Development History
The original MININEC was written by John Rockway with a little prodding and support from Jim Logan. Over the years, the Rockway - Logan team has been responsible for the development of this code into one of the best known and most useful Method of Moments antenna modeling codes available. A number of other individuals have also contributed small, but not necessarily insignificant, pieces to the MININEC capability, but it has been the dual efforts of the Rockway - Logan team that has made MININEC into a powerful antenna design and analysis tool.
Because of the similarity in names, it is often stated that MININEC is but a personal computer (PC) version of its big brother, NEC [2]. However, this could not be farther from the truth. There are significant differences between these two codes. Both codes use the Method of Moments to solve for currents on electrically thin wires. However, each code starts with a different version of the integral formulation for the currents and fields for wires. Then, each follows significantly different algorithms for implementation of the Method of Moments.
In 1980, when the first version of MININEC was written, PCs had not been on the market for very long. They were relatively expensive and very limited in capability. PCs were generally regarded as mere novelties or toys. PCs were typically limited to 16K memory with a 8 bit word length. There was no FORTRAN. MININEC had to be written in BASIC. NEC was (and still is) a very powerful computer code, with tens of thousands of FORTRAN statements, originally written for use on large main frame computers. In those days PCs could not support such a large program size. The formulation had to be changed to allow a simpler implementation of the Method of Moments in order to produce a more compact code. It would not be possible to include many of the powerful modeling options provided by NEC. Following the advice of Professor Don Wilton at the University of Mississippi (now with the University of Houston), the first version of MININEC was written in 500 lines BASIC and required 32k of memory. Nonetheless, this version proved surprisingly accurate for dipoles and monopoles.
The first public release of MININEC occurred in 1982 [3]. The code was 550 lines of BASIC and would run on an APPLE II computer with 64 kilobytes of memory. It could compute the current distribution, impedance, and far field pattern of an arbitrarily oriented set of wires in free space or over a perfectly conducting ground plane. Lumped impedance loads were allowed at segment junctions except for segments intersecting with the ground plane. Also, wires intersecting the ground plane were restricted to right angles. In interpreter BASIC (there were no BASIC compilers then) the problem size was limited to 10 wires and 50 currents (or 70 segments with junctions).
MININEC was an instant success. Almost immediately, a small user group developed and began to grow. In 1984, partly to meet the demand for MININEC as well as share other computer algorithms, the authors teamed up with two colleagues, Peter Li and Dan Tam. They published a book that contained an improved version of MININEC along with some other useful algorithms [5]. MININEC2, as it became known, was not significantly different from its predecessor, but the limitation for wires intersecting the ground plane was removed. Wires could intersect the ground at any angle.
The power of PCs began to grow. Computers were getting faster, had more memory, and utilized math coprocessors. BASIC compilers also became available. These factors opened up new vistas for MININEC. In 1986, the authors released MININEC3 [6]. This code featured a new user interface which automatically determined wire connections from the user inputs for wire end coordinates. It could also read and interpret a limited NEC input data set. However, there was no way to save and edit geometry data. MININEC3 included near fields, a Fresnel reflection coefficient correction to the patterns for real ground, and an expanded lumped parameter loading option. MININEC had grown to just over 1600 lines of BASIC. With a math coprocessor and a BASIC compiler, MININEC3 could solve antenna problems up to 50 wires and 50 current unknowns.
The next MININEC effort by the authors produced the MININEC SYSTEM in 1988 [8]. This was a valiant effort by the authors to provide improved problem definition, save features, and on-line graphics. The release of the MININEC SYSTEM happened to coincide with the introduction of Microsoft Windows that took the PC world by storm. The authors were too close to publication to backtrack and implement a Windows system. However, there were many worthwhile innovations represented in this code. This was the first version of MININEC that required a compiler, a BASIC compiler. Previous versions could be run in interpreter BASIC. The solution time and storage requirements for rotationally symmetric antennas were greatly reduced. The transpose elimination algorithm was available as a user select option to allow computation of larger problems, up to 50 wires and 90 current samples or 190 segments were permitted without recompiling.
Many others have also attempted to improve on MININEC. Most notable are the innovative user interfaces and graphics displays offered by Roy Lewallen [4] in 1991 and Brian Beezley [1] in 1992.
In 1995, the authors published the first of a series of MININEC for Windows codes. These codes represented the development of a new version of MININEC. An improved solution of the potential-integral formulation for the currents resulted in a more accurate formulation in the solution for the currents on wires. In addition, FORTRAN was used for the computationally intensive portions of MININEC. This led to an increase in speed over previous versions of MININEC.
The first code was MININEC Professional for Windows [9]. Because it is a Windows application, text and graphical outputs are easily transferred to other Windows applications such as spreadsheets and word processors. Mouse support and printer drivers are also supplied by the Windows environment. The input is a node based geometry. That is, nodes define points in space (in Cartesian, cylindrical or geographic coordinates) and wires are defined between nodes. Entries are made in tables through individualized window screens. On line, context sensitive help is provided along with diagnostic preprocessing diagnostics. MININEC Professional is dimensioned for 1000 wires and 2000 unknowns.
In 1996, the authors published MININEC Broadcast Professional for Windows [10] which is similar to its predecessor, but more powerful. Additional features include an improved voltage source model, a plane wave source model, automated convergence testing, design analysis post processing, array synthesis, and ground wave calculations. MININEC Broadcast Professional is dimensioned for 2000 wires and 4000 unknowns.
Also in 1996, the authors published MININEC for Windows [11], a simplified version of MININEC Professional which is more suitable to first time users and their pocketbooks. This code is dimensioned for 400 wires and 800 unknowns.
In 1999, the authors published another improved set of codes, the Expert MININEC Series [12][13][14]. The new series features "Expert" assistance in selecting appropriate input dialog boxes while constructing a model. Context sensitive help is still an important feature. Accuracy and speed have also been improved.

Example 1 - Antenna Coupling
Antenna modeling codes, such as Expert MININEC can be used to accurately and efficiently calculate the coupling between antennas. Antenna coupling is a significant parameter in many electromagnetic compatibility (EMC) analyses. The specific approach is based on an N-port description and the application of the Linville method [15].

N-port Description
When two or more antenna systems are in proximity to each other, power from a transmitting system can be coupled into the other systems through an electromagnetic interaction. The problem of coupling among the antenna systems can be described using network "Y" parameters [16]. Any two ports of an antenna can be treated as a two-port network of the electromagnetic interactions. In the Figure 1 the terminals with voltage Vi are the feedpoint of antenna port i, and the terminals j are the feed point of the second antenna port. By convention, the currents Ii and Ij are assumed to be positive into the electromagnetic interaction network.
In general, of the four variables shown (Vi, Ii, Vj, Ij), only two are independent. Thus, the following functions may be written:

(6)

(7)

 Figure 1. Two-port Electromagnetic Interaction Network Definition.

Since the electromagnetic interactions are passive and linear, the functions can assumed to be linear and thus:

(8)

(9)

These admittance parameters are defined by:

(10)

(11)

(12)

(13)

Vi and Vj equals zero implies that the terminals associated with these voltages are short-circuited. An arrangement whereby these admittances may be computed is shown in Figures 2 and 3.

 Figure 2. Arrangement for Determining Yii and Yji.
 Figure 3. Arrangement for Determining Yjj and Yij.

The feed point of one electromagnetic port is excited, and the feed point of the second electromagnetic port is short-circuited. The calculated currents are used with equations (10) through (13) to determine Yii, Yji, Yij, and Yjj. It has been found that the admittance port parameters can be computed more effectively with the Method of Moments solution of the electromagnetic interactions. Once the admittance port parameters have been found, the maximum coupling between ports can be determined.

Linville Method
Again consider Figure 2 and assume a load ,Yload, on port j instead of a short circuit. The output power is

(14)

where Vj is the output voltage (port j), and Re [Yload] is the real part of the load admittance . The input power is

(15)

where Vi is the input voltage (port i), Re [Yin] is the real part of the input admittance of port i.
The power gain is then the ratio of (14) and (15)

(16)

This ratio is dependent on the input admittance. Yin can be calculated from Equation (8) and (9) by solving

(17)

with

(18)

then

(19)

The two-port admittance parameters in Equations (8) and (9) can be used also to find the output to input voltage ratio, , as follows

(20)

Substituting ((20)) into ((16)), the gain becomes

(21)

The load admittance, Yload, can be found that maximizes the gain. This gain is the maximum possible power transfer ratio, provided that the generator is matched to the resulting Yin as given in (19).
It is difficult to maximize the gain directly by taking the derivative of Equation (21) with respect to complex load admittance, Yload, and setting the derivative to zero. A more appropriate method is the Linville analysis approach used in the design of RF amplifiers [16]. The Linville method is a graphical based method. Using the Linville approach, the maximum coupling becomes

(22)

where

(23)

The matched load admittance on port j for the maximum coupling is

(24)

where

(25)

and * indicates complex conjugate of the product of Yij and Yji.
The maximum coupling between ports can be used to identify the greatest coupling paths. The maximum coupling calculation can be used to rank order the interactions. Possibly, this rank ordering can be used to eliminate the more weakly coupled paths as not being significant to the EMC analysis. This rank order provides insight into the design.

Example Calculation
As an example calculation, consider two monopoles. Each monopole has a length of .25 meters and a radius of .001 meters. The antennas are .1 meters apart. Six segments are used for each monopole in the Method of Moments calculation. The problem is to determine the maximum coupling between the two monopoles from 260 MHz to 350 MHz. As an example, at 260 MHz, the current computation for a one volt source on the first antenna is given in the following:

Current peak                                                             real               imaginary
no.              X                Y                    Z                (amps)             (amps)
GND         -.05              0                     0                5.57E-03        .0163887
2                -.05              0                     .0416667   5.41E-03        .0149296
3                -.05              0                     .0833333   4.93E-03        .0129975
4                -.05              0                     .125           4.15E-03        .0104939
5                -.05              0                     .166667     3.08E-03      7.49E-03
6                -.05              0                     .208333     1.74E-03      4.06E-03
END          -.05              0                     .25             0                   0

GND           .05              0                    0                4.78E-03     -6.53E-03
8                  .05              0                    .0416667   4.64E-03     -6.31E-03
9                  .05              0                    .0833333   4.23E-03     -5.68E-03
10                .05              0                    .125           3.57E-03     -4.67E-03
11                .05              0                    .166667     2.65E-03     -3.36E-03
12                .05              0                    .208333     1.5E-03       -1.83E-03
END           .05               0                   .25              0                   0

The current computation for a one volt source on the second antennas is given in the following:

CURRENT peak                                                          real          imaginary

no.              X                 Y                  Z                    (amps)         (amps)
GND         -.05               0                   0                  4.78E-03     -6.53E-03
2                -.05               0                   .0416667     4.64E-03     -6.31E-03
3                -.05               0                   .0833333     4.23E-03     -5.68E-03
4                -.05               0                   .125             3.57E-03     -4.67E-03
5                -.05               0                   .166667       2.65E-03     -3.36E-03
6                -.05               0                   .208333       1.5E-03       -1.83E-03
END          -.05               0                   .25               0                   0

GND           .05               0                   0                 5.57E-03      .0163887
8                  .05               0                   .0416667    5.41E-03      .0149297
9                  .05               0                   .0833333    4.93E-03      .0129976
10                .05               0                   .125            4.15 E-03     .0104939
11                .05               0                   .166667      3.08E-03      7.49E-03
12                .05               0                   .208333      1.74E-03      4.06E-03
END            .05               0                   .25              0                   0

The appropriate admittance parameters are calculated using Equations (10) to (13). Since there is obvious symmetry for this problem, the admittance parameters are
Y11 = Y22 = .00557 + j .0163887
Y12 = Y21 = .00478 -j .00653
Equation (22) is then used to calculate the maximum coupling. The load for this maximum coupling is given by Equation (24). Finally, the input impedance for this load is given by Equation (19). The results for an Expert MININEC calculation are given in the following table:

frequency input impedance          load  impedance         coupling
(MHz)    (ohms)        (ohms)       (ohms)     (ohms)          (dB)
260.        8.78226     -43.6793      8.78229  43.6792      -3.01957
270.      10.1627       -29.1595    10.1627    29.1594      -3.14462
280.      11.7534       -14.6924    11.7534    14.6925      -3.26634
290.      13.5772       -.184137    13.5772   .183732       -3.38805
300.      15.6652       14.4604     15.6652   -14.4604      -3.51159
310.      18.0553       29.336       18.0554   -29.3361      -3.63828
320.      20.7965       44.5426     20.7966   -44.5423      -3.7687
330.      23.9493       60.1861     23.9493   -60.186        -3.90311
340.      27.5894       76.38 27   .5894        -76.38          -4.04161
350.      31.8117       93.2499     31.8117   -93.2497      -4.18412

Example 2 - Common Mode Radiation
Common mode radiation is a concern for EMC engineers since it is often a much more serious problem than differential mode radiation. If the structure of interest is a multiconductor transmission line with a conductive return path (e.g., a two wire transmission line over a ground plane), it is possible to calculate the common and differential mode currents using conventional multiconductor transmission line theory [7]. This technique is well known to EMC engineers. Less well known, however, is how to determine these currents when the current return path is not conductive. For this case full wave electromagnetic theory must be used. The purpose of this example is to illustrate how an antenna analysis program such as Expert MININEC can be used to determine common mode current amplitudes.
Consider first the simple problem shown in the following figure. Here, a balanced transmission line is driven by a 1 volt sinusoidal source and terminated by a 408 _ resistor that matches the characteristic impedance of the transmission line. The transmission line is 25 cm in length and constructed with .2 cm diameter wires spaced 3 cm apart. The currents I1 and I2 represent the currents into and out of the top and bottom wires respectively.

 Figure 4. Open Wire Transmission Line Terminated in its Characteristic Impedance.

Note here that the definitions of common mode, Ic, and differential mode, Id, currents are respectively

(26)

(27)

Because the geometry of the problem is symmetric, I1 = I2 for all frequencies, the current is entirely differential. This result is consistent with two wire transmission line theory. This has also been demonstrated directly using Expert MININEC over the frequency range of interest, 10 to 300 MHz. For the Expert MININEC calculation, each horizontal wire was divided into 40 segments and the vertical wires divided into 3 segments.
If, however, another wire that might represent a ground lead that extends past the source is added, the symmetry of the problem is destroyed. This geometry is depicted in Figure 5. In this case, two wire transmission line theory is not appropriate and an analysis tool such as Expert MININEC must be used.

 Figure 5. Open wire transmission line with an additional wire to the left side.

The results of this calculation are shown in Figure 6. It is clear that at lower frequencies, the differential mode dominates since little current flows through an open circuited, electrically short wire. At these lower frequencies, conventional transmission line theory that assumes differential mode currents can be used despite the lack of symmetry. However, as the length of the structure approaches a half wavelength, the common mode current becomes significant. In fact, it can actually dominate the differential mode current. In this case, the structure behaves more like an antenna than a transmission line.

 Figure 6. Common and differential mode currents on the structure of Figure 5.

References
1. Beezley, B., The MN4 Manual, Brian Breezley, Vista, CA, 1992.
2. Burke, G. J. and A. J. Poggio, "Numerical Electromagnetics Code (NEC) - Method of Moments," Naval Ocean Systems Center Technical Document 116, January 1981.
3. Julian, A. J., J. C. Logan, J. W. Rockway, "MININEC: A Mini-Numerical Electromagnetics Code," NOSC Technical Document 516, September 1982.
4. Lewallen, R., "MININEC: The Other Edge of the Sword," QST Magazine, February 1991.
5. Li, S. T., J. C. Logan, J. W. Rockway, D. W. Tam, Microcomputer Tools for Communications Engineering, Artech House, Inc., Dedham, MA 1984.
6. Logan, J. C. and J. W. Rockway, "The New MININEC (Version 3): A Mini-Numerical Electromagnetic Code," NOSC Technical Document 938, September 1986.
7. Paul, C. R., Introduction to Electromagnetic Compatibility, Wiley, New York, 1992
8. Rockway, J. W., J. C. Logan, D. W. Tam, and S. T. Li, The MININEC System: Microcomputer Analysis of Wire Antennas, Artech House, Inc. Dedham, MA 1988.
9. Rockway, J. W. and J. C. Logan, MININEC Professional for Windows, EM Scientific, Inc., Carson City, Nevada, 1995.
10. Rockway, J. W. and J. C. Logan, MININEC Broadcast Professional for Windows, EM Scientific, Carson City, Nevada, 1996.
11. Rockway, J. W. and J. C. Logan, MININEC for Windows, EM Scientific, Carson City, Nevada, 1996.
12. Rockway, J. W. and J.C. Logan, Expert MININEC Professional for Windows, EM Scientific, Inc., Carson City, NV, 1999.
13. Rockway, J. W. and J.C. Logan, Expert MININEC Broadcast Pro for Windows, EM Scientific, Inc., Carson City, NV, 1999.
14. Rockway, J. W. and J.C. Logan, Expert MININEC for Windows, EM Scientific, Inc., Carson City, NV, 1999.
15. Rubin, D., "The Linville Method of High Frequency Transistor Amplifier Design," Naval Weapons Center, Research Department, NWCCL TP 845, Corona Laboratories, Corona, California, March 1969.
16. Van Valkenburg, M. E., Modern Network Synthesis, John Wiley and Sons, New York 1960.

John Rockway received the B.S. and M.S. degrees in electrical engineering and Ph.D. in engineering science from Washington State University in 1966, 1968 and 1971, respectively. He is currently the head of the technical staff of the Electromagnetics and Advanced Technology Division of the Space and Naval Warfare Systems Center - San Diego (SSC-SD). He has spent his entire engineering career with SSC-SD and its predecessors. The primary emphasis of this career has been on the development and evaluation of shipboard antennas, development and application of advanced computational electromagnetic and RF system design tools, and the development of advanced communication systems for the Navy. Government and professional recognitions include IEEE Fellow, the Lauritsen-Bennett Award (SSC-SD highest honorary award for Excellence in Engineering), the Department of the Navy Award for Meritorious Civilian Service, the National Society of Professional Engineers Federal Engineer of the Year Award and the Applied Computational Electromagnetics Society Mainstay Award.

Mr. James C. Logan earned his BSEE degree in 1967 and his MSEE degree in 1973, both from Syracuse University. He was a co-founder of the Applied Computational Electromagnetics Society (ACES) in 1984. During the first 10 years of ACES, Mr. Logan served as the first Vice President, the second President and the second Treasurer as well as numerous committees. He is also a Senior Member of the International Institute of Electrical and Electronics Engineers (IEEE) Antennas and Propagation Society and the IEEE Electromagnetic Compatibility Society. Mr. Logan is the author or co-author of many (more than 60) papers and presentations appearing in professional publications as well as numerous Government documents. Mr. Logan is also co-author of five commercial books; one on the design and analysis of RF communications systems and four on the design and analysis of antennas. Mr. Logan co-founded EM Scientific, Inc. in 1995. EM Scientific, Inc. is a publisher of scientific and engineering software and reference books. Mr. Logan retired from the Space and Naval Warfare Systems Center - San Diego (SSC-SD) in 1999. He is now a retired annuitant at SSC-SD.
Professional awards and recognitions include:

• Applied Computational Electromagnetic Society, 1990 Mainstay Award.
• Naval Surface warfare Center, Caderock Division, Group Award Citation for leadership and determination in defining and developing the Advanced Enclosed Mast/Sensor System (AEM/S), June 10, 1994.
• Applied Computational Electromagnetic Society, 1994 Founders Award.
• Naval Command, Control and Ocean Surveillance Center Exemplary Achievement Award, June 16, 1997.
• Naval Surface Warfare Center, Caderock Division, Incentive Award for Special Act or Service, December 10, 1997.
• Department of the Navy, Navy Meritorious Civilian Service Award, June 26, 1998.
• The Chief of Naval Research Dr. Arthur E. Bisson Prize for Naval Technology Achievement, May 16, 2000.

Robert G. Olsen (S'66 - F'92) received the BS degree in electrical engineering from Rutgers University in 1968 and the MS and Ph.D. degrees in electrical engineering from the University of Colorado, Boulder in 1970 and 1974 respectively.
Prof. Olsen has been a member of the electrical engineering faculty at Washington State University since 1973. During that time he has been a visiting scientist at GTE Laboratories in Waltham, MA at ABB Corporate Research in Västerås, Sweden and at EPRI in Palo Alto, CA and a Visiting Professor at the Technical University of Denmark.
His research interests include electromagnetic interference from power lines, the electromagnetic environment of power lines, electromagnetic wave propagation, electromagnetic compatibility and electromagnetic scattering. His recent work has been supported by the Bonneville Power Administration, the Boeing Defense and Space Group, the Electric Power Research Institute, the National Science Foundation and the U.S. Navy.
He is a Fellow of the IEEE and presently serves as chair of the IEEE Power Engineering Society Corona Effects Fields Working Group, as Technical Editor of the IEEE Electromagnetic Compatibility Society Newsletter and as USNC representative to CIGRE Study Committee 36 (Electromagnetic Compatibility). He is past chair of the IEEE Power Engineering Society AC Fields Working Group.

 Errata Art Glazar, author of the article, "A Software Implementation of TL Field-to Cable Coupling Equations," that appeared in the Fall 2000 issue of the Newsletter has informed us that the computer program (coax.exe) offered for free in that article has an error that affects problems where shields or signals are terminated by a complex impedance. He has corrected the error and will send an updated version of the program to anyone who requests it. If you would like to have this, please e-mail him at aglazar@ieee.org

NUMERICAL EMC SIMULATION FOR AUTOMOTIVE APPLICATIONS

Abstract: New electronic systems accompany most of the technical innovations in automotive industry. In consequence of the growing number of electrical equipment the electromagnetic noise level is rising in automobiles. The increased electromagnetic emissions however raise the risk to miss EMC standards and perturb the functional integrity of new and existing electronic systems. Hence, mastering potential EMC problems in new automobiles becomes increasingly important to the car industry.
Numerical simulation is an important key for the detection and rejection of potential EMC problems early in the design process. This paper presents a continuous concurrent EMC simulation process based on the exchange of EMC models between car manufacturer, electronic supplier and IC developer. The described process fundamentally influences the introduction of new technologies in a vehicle by cutting the risk of EMC failure and avoiding costly and time-consuming redesigns.

1. Introduction
In recent years there has been a significant increase in the amount of electronics that have been introduced into the car and this trend is expected to continue as car companies introduce further advances in safety, reliability and comfort. The accompanying increase of electronic noise emission and interference is a well-known problem in the automotive industry [1]. In modern cars, the expenses for the control of the electromagnetic emission per car can add up to 50 Euro for preventive measures such as filters and wire shielding. The costs of the chip manufacturer and the electronic supplier for electromagnetic compatibility (EMC) corrections are not included in the above amount.
In addition, future technical developments, such as:

• an increasing number of pulse-width-modulated (PWM) signal applications for electrical consumers (especially in 42Volt systems),
• wider penetration of the harness through the car,
• higher data traffic (new bus systems tend to data transmission rates up to 10Mbits/s using standard twisted pair cables), will aggravate EMC problems in automotive applications and increase the need for early detection and rejection of potential EMC failures.

Currently, system-level automotive EMC is controlled in two steps. In the first step, the electrical components inside a car are tested according the corresponding international standards (e.g. ISO7637 Part 2 and 3) and the standards of the car manufacturer. In the second and final step, system-level EMC tests (e.g. EN95/54/EG) on the car are performed. The mentioned standards describe measurements to qualify the electromagnetic noise emission and immunity. These measurements are performed on existing hardware at the end of the development process. The standards reflect the experience of the EMC engineers with existing electronics.
The latter two points, however, indicate the major drawbacks in the current EMC design flow. First, the existing standards are frequently insufficient for new electronic systems integrated in the car, e.g. a neon lamp can fulfill the standard for EM emission, but built in the rear window as a third brake light, it can inhibit any radio reception with the rear window antenna. Second, since the EMC problems are treated in a late stage of design or during prototype testing, correction measures are limited and usually cost-intensive. In the worst case, repeated redesign cycles for the automobile might stack up to several months and the introduction of new products to the market can be delayed dramatically.
Consequently, an accurate analysis of potential EMC problems in new automobiles, such as:

• coupling between wires inside a harness (crosstalk),
• radiation from the harness towards the environment and antenna (emission),
• immunity of equipment against external electromagnetic interference (immunity), in earlier stages of the development process is increasingly important to the car industry.

Numerical EMC simulation is an important key to reach this aim [2]. This paper describes a continuous, concurrent EMC simulation process developed under the COSIME project, granted by the European Commission. In Section 2, the general modeling and simulation strategy based on the exchange of EMC models between car manufacturer, electronic supplier, and chip developer is explained. Section 3 discusses different modeling approaches at sub-system level, while Section 4 presents how to perform the final system-level EMC simulation. In Section 5 the validation of the proposed process is discussed. Conclusions are drawn in Section 6.

2. Continuous Concurrent EMC Simulation
Previous considerations place the demand for a design process that comprises EMC analysis from the early development phase and guarantees that the final product performs as requested, without the need of adjustments during the prototype and production phase. Employing numerical EMC simulation throughout all design stages enables the targeted early identification of potential system malfunctions and most appropriate correction measures can be placed in time.
EMC simulation is well advanced in the development process of automotive electronics. Like the EMC validation by means of measurements, the EMC simulation is divided into the fields of electromagnetic emission (EMI) and electromagnetic susceptibility (EMS). Besides, it is distinguished between component simulations and system simulations at car level. The component simulations are less complex in principle, since no geometrical data of the car is needed and the test concerns only single components. The efficiency of this simulation process has already been proved [3, 4].
The complete car simulation, however, is still considered a challenge. One major difficulty encountered in automotive EMC simulation is to deal with very different relevant geometric scales, related to the three main parts of the problem: the car body (large 3D structures), the harness (2D incorporating ground plane effects), and the equipment (essentially 0D). This scattering of geometric scales corresponds to different physical behaviors, which consequently call for different modeling approaches and different simulation environments [5] such as:

• full-wave 3D resolution at the car body level (incl. antenna),
• transmission line propagation at harness and bundles level, and
• circuit formulation for equipment of negligible size with respect to the wavelength of interest.

Applying to a complex car model Maxwell's equations only - in order to solve the electromagnetic simulation problem - requires a very fine discretization with respect to the geometrical size. This implies, however, unaffordable computer resources in terms of computation time and memory. Thus, computational efficient numerical simulations can only be performed successfully by decoupling the problem hierarchically and employing suitable simulation techniques that consider the different levels of details (multi-level strategy).
Experience proved [6] that the accuracy of simulation results for a complete system-level automotive EMC simulation does not only depend on the simulation model of the car body and harness, but also on the utilized simulation models of the control modules at equipment level. Models of the control modules in turn are based on models of the integrated circuit devices (IC). Hence, improved EMC simulation results can be achieved using a continuous simulation process with contributions from car manufacturer, electronic supplier and IC company. This concurrent approach is based on the exchange of EMC behavior models from the concept phase towards the prototyping in the car development process. Currently, models for the electrical equipment are derived from measurements on existing hardware or from experience gained from existing electronics. In the proposed process, electrical behavior models (e.g. IBIS, transistor models of reduced complexity, etc.) representing the EMC behavior of the components to be modeled are used. As exchange format, standard SPICE syntax is employed. The SPICE format enables a tool-independent interchange of EMC models among the different partners involved in the design process, and simplifies the integration of individual EMC behavior models for control modules and ICs, respectively, into the complete car model. Fig. 1 depicts the proposed EMC simulation strategy, starting from chip level EMC design up to system level EMC analysis, as applied in the car development process. The concurrent approach is employed continuously at all stages in the design flow, from the initial idea up to the prototyping and the production phase.

 Fig. 1. Concurrent engineering approach in EMC design.

Besides the use of the explained multi-level strategy, the described simulation process enables the continuous exchange of models and simulation results in both directions, e.g. IC models from IC manufacturer to the electronic supplier and the car manufacturer but also chassis or antenna models from the car manufacturer to the electronic supplier and IC manufacturer. For this reason, the validation of new concepts can be performed at earlier stages of the development process and additional measures to fulfill the EMC standards or to reject system malfunctions can be implemented in time. This way, the time to market for new electronic products (not only in the car industry) will be reduced and the reliability of the system can be improved.

3. Subsystem Modeling

The simulation process presented in the prior section is based on the multi-level modeling approach and on the exchange of EMC behavior models. This Section discusses the different subsystem models used for IC, control module and vehicle. In parallel, the different modeling techniques applied in the quoted multi-level modeling approach are explained in detail.

3.1 Active components
Regarding active components used in modern electronics, micro-controller chips (µC) and bus drivers (e.g. CAN) are of primary interest in EMC design in automotive industry. Thereof only the mains-borne disturbance through supply lines from the µC and conducted emission through output ports from the bus-driver are modeled. Evaluating an accurate EMC behavior model for all pins of an active component (including the radiated coupling) would blast the simulation process.
In order to generate pre-mentioned EMC behavior models, different modeling techniques can be applied. One commonly used standard in chip industry is IBIS (Input/Output Buffer Information Specification) [7]. The IBIS-standard is based on measurements and/or transistor level simulations of static and dynamic characteristics of the IC. Additionally, IBIS files include package data (R, L, C), driver capacitances and clamping diodes behavior. A complete IC contains different models, depending on the number of different types of drivers and receivers of the device. Yet it allows no insight into the actual design, but describes the EMC behavior on the circuit I/O-terminals. From the circuit information in IBIS format an electrical circuit model in SPICE syntax is generated as described in [8, 9].
Guidelines for the extraction of an EMC model describing the conducted emissions through power-supply lines are presented in [10, 11]. The core model consists of a current generator modeling the main source of parasitic emissions. The current shape is either extracted from standardized measurements or computed by means of simulation tools. Further, the model takes into account first order effects due to package parasitics and on-chip capacitance, and second order effects caused by die capacitance, bonding and die connection inductance. If requested the core coupling to the I/O pins of the chip can be included in the model.
Both model formulations, IBIS and integrated circuit electromagnetic models (ICEM), are given as electrical circuit networks in SPICE syntax. Hence, the requested integration into most of the network simulators is easily possible.
Additional modeling methods are presented in [12] and [13]. The first approach is based on the identification of black-box nonlinear dynamic models. It is suitable for characterizing poorly documented devices from standard input/output transient measurements. The latter approach proposes a model for the µC-core based on supply-current and impedance measurements.

 Fig. 2. 3D models for (a) chassis, (b) harness, and (c) rear window antenna.

3.2 Interconnects and passive components at PCB level
Passive components are characterized satisfactorily in most cases by the corresponding high frequency models. Of particular interest however is the modeling of interconnects. Depending on the highest operating frequency, signal rise times and the nature of structure, interconnects can be represented by lumped, distributed, or full-wave models [14]. Lumped RLC-circuit models are used for electrical short interconnects only, whereas distributed transmissions line (TL) models described by Telegrapher's equations are applied at higher switching speeds. The distributed R, L, C, G per unit length parameters can also take into account frequency-dependent effects (e.g. skin effect). Spatial EM effects are successfully accomplished using partial element equivalent circuit (PEEC) models [15]. The PEEC models are coupled RLC-circuits extracted from the geometry using the quasi-static or the full-wave solution of Maxwell's equations.
Concerning the multi-level strategy, the generated models for each control model are incorporated in the network simulator environment.

 Fig. 3. Statistical variations of the transmission and crosstalk parameters of a 9-wire harness due to mutual position of wires.

 Fig. 4. Recursive harness modeling process incorporating geometrical data and statistical bundling variations.

3.3 Chassis
The geometry extraction process starts with the meshing of the car body. Very detailed geometry data of the car body is available from the CAD-framework of the car manufacturer. However, with respect to EMC analysis it is an important task to simplify the detailed geometry data to a computable mesh without loss of essential information, and to translate the data into an interchangeable format for field solvers. To complete the geometrical model, the centerlines of the interesting harness sections and antenna are extracted from the CAD-framework and added to the 3D mesh. Fig. 2 exemplifies obtained geometrical models for (a) chassis, (b) harness, and (c) antenna. The generated geometrical models are also used for the 3D electromagnetic field calculations (described in Section 4).

3.4 Cable harness

One major problem encountered in numerical analysis of high frequency electromagnetic interference in automobiles is the high complexity of chassis and harness. The harness with its overall wire length of several kilometers, its small cross-section diameter compared to its length and its tolerances in geometry and electrical parameters, limits the accuracy of the EMC simulation.
Likewise, statistical variations need to be considered when generating the harness model, since the mutual positions of the wires inside the bundle vary within certain limits. Fig. 3 demonstrates the influence of the mutual positions of the wires by means of comparison of far end transmission and crosstalk of a generic harness [16].

The complexity of the problem can be accomplished by applying a recursive modeling process as outlined in Fig. 4. In a first step, the centerline polygon of the geometrical model of the harness is separated in different segments. Thereafter, the cross-sections of these segments consisting of the center point of the polygon and its surrounding chassis shapes are mapped in 2D coordinates. Some sample sections are plotted in Fig. 5.
Next, the wires inside the automotive harness are positioned statistically around the 2D center point of each cross-section. Using the 2D coordinates and the statistical data as input, the TL parameter matrices R, L, G, C of the harness are calculated for each section. The obtained multi transmission line model (MTL) is employed to calculate the current distribution along the wires. This process is repeated with statistically repositioned wires. Out of the simulation results a characteristic current distribution is incorporated in the harness model by means of equivalent current sources for further calculations.

 Fig. 5. 2D model of harness for TL parameter extraction.

3.5 Antenna
One particular application of the EMC simulation in automobiles is to calculate the voltage at the base of the car antenna caused by the signal and power transmission via the harness. For this purpose, the frequency-domain transfer function representing the coupling from the investigated ports of the harness towards the antenna is calculated. This transfer function is computed by means of a 3D field solver with the geometrical models of chassis, harness and antenna as inputs.
Fig. 6a depicts a practical example where the coupling of a harness located above the hatrack towards the rear window antenna of the car is investigated. The calculated transfer function given by the corresponding scattering parameters is plotted in Fig. 6.
For the frequency-domain simulation the tabulated data of the coupling is directly incorporated in the multi-level simulation process. In case of performing a transient analysis an equivalent circuit model is generated from the scattering parameter data and integrated in the network simulator as described in [17].

 Fig. 6. Harness-antenna-coupling: (a) geometrical model (b) scattering parameters.

4. System Simulation
Whereas in previous sections different modeling approaches for individual subsystems in automotive applications were presented, this section discusses the system-level simulation of automotive EMC problems. It focuses on the numerical prediction of the major EMC challenges in automotive applications: crosstalk, emission, and immunity. In the following, specific simulation strategies for the different phenomena are proposed. All strategies are summarized in Fig. 7.

 Fig. 7. Simulation flow for Crosstalk (CT), Emission (EMI) and Immunity (EMS).

4.1 Crosstalk
As partly described in Chapter 3.4 the crosstalk simulation is based on a MTL model, statistical variations of the bundle, and terminations at the ports of the harness. The latter are modeled by the corresponding equipment models consisting of IC models and circuit models of passive components and interconnects. Connecting the MTL model to the EMC behavior models for the equipment the crosstalk simulation can be performed (compare Fig. 7). In case of linear loads the joined model is analyzed in frequency-domain with TL simulators. For non-linear terminations network simulators that support MTL models are employed to calculate the time-domain solution.
For the purpose of illustration the crosstalk occurring in four twisted pair cables is investigated [16]. The twisted pairs are excited by simultaneously switched PWM-signals. Applying the modeling procedure described in Fig. 4 and performing the simulation process outlined in Fig. 8 the crosstalk for ten statistical positioned wire bundles is calculated.

 Fig. 8. Crosstalk simulation: Transient port response using ten different wire bundles.

4.2 Emission
The calculation of EMI starts again with the generation of the subsystem models for chassis, harness and equipment. Thereafter, the harness and equipment models are joined together and the current distribution along the wires of the harness is calculated using a network solver and TL models. The obtained current distribution is implemented as impressed current sources (Huygen's principle) in the meshed space or surface of the geometrical model for the harness defined for the field solver. In the final step, the geometrical models for chassis and the impressed current sources are imported to a 3D field solver and the radiation from the harness is calculated. Fig. 9a illustrates the EM field in a cut plane during this 3D radiation.
Rather than the 3D fields, in many practical applications the voltage obtained at the antenna feeding point (e.g. located in the rear window) is of particular interest [18]. The incorporation of the antenna model into the car environment is described in Chapter 3.5. At this point, the system configuration depicted in Fig. 6 is excited with a non-linear driver at port 1 (harness). The calculated transient response is plotted in Fig. 9.

4.3 Immunity
For EMS analysis, basically the reversed simulation process than for EMI is applied. First, using the geometrical model of the chassis a 3D field solver is employed to calculate the electromagnetic field within the car body caused by the EMI source. The obtained E- and H-field is analyzed at the position of the harness, and the corresponding current and voltage values are calculated. Next, the currents and voltages are impressed as distributed controlled sources in the TL network [19]. Finally, the obtained electrical circuit including the equipment models is imported in the network simulator and the port response is simulated.

 Fig. 9. Emission simulation: (a) electromagnetic field (b) voltage at the antenna feeding point.

5. Validation
To prove the feasibility of the proposed continuous concurrent EMC simulation process, EMC measurements need to be performed simultaneously. The validation is carried out at the different partners (IC manufacturer, electronic supplier, car manufacturer) contributing to the vehicle design. As a result of the validation the simulation process can be improved perpetually. Nevertheless, EMC simulations will not replace EMC measurements and EMC standards. Both still play an important role to guarantee the quality and functionality of the final product. However, the proposed simulation process provides a useful tool for the evaluation of new concepts regarding their EMC characteristics at an earlier stage of the design process.

6. Conclusion
In recent years, most innovations in the automotive industry are accompanied by new electronics. In consequence of the growing number of electrical equipment the electromagnetic noise level is rising in automobiles. The increased electromagnetic emissions place high demands on the EMC engineer to fulfill requested EMC standards and guarantee the functional integrity of the electronic systems. The early detection and rejection of potential EMC problems becomes mandatory for the success of new technologies.
This paper presented a continuous concurrent EMC modeling and simulation process for automotive applications. Essential guidelines for the generation of EMC behavior models of components and structures from IC designer, electronic supplier and car manufacturer involved in the design were proposed. Furthermore, specific simulation strategies for the different phenomena crosstalk, emission, and immunity were discussed.
Employing EMC simulation, the goal of early design-concept validation and on-time implementation of EMC measures can be achieved. Although EMC simulation will not replace the validation of the final design by measurements, it reduces the risk of EMC failure and aids a scheduled launch of new products.

Acknowledgement
This work was supported by the European Commission under contract number G3RD-2000-00305.

References
1. I. E. Noble, "Electromagnetic Compatibility in the Automotive Environment," in IEE Proc. Science, Measurements and Technology, vol. 14(4), 1994, pp. 252-258.
2. J.C. Kedzia. "Numerical EMC in Ground Transportation: How to Manage Efficiently Realistic Automotive Problems," in Proc. PAM Users Conference in Asia - PUCA'99, Nov. 1999.
3. J. C. Rautio, "MIC Simulation Column - A Standard Stripline Benchmark," Int. Journal of Microwave & Millimeter-Wave Computer-Aided Engineering, vol. 4, no. 2, April 1994, pp. 209-212.
4. A. Englmaier, "Methods and Models for EMC-Simulation," PhD Thesis, Technical University Munich, June 1998.
5. F. Canavera, J.C. Kedzaia, P. Ravier, B. Scholl, "Numerical Simulation for Early EMC Design of Cars," in Proc. CEM 2000 Symposium, Brugge, Sept. 2000.
6. B. Scholl, "Automotive EMC Problems & Research," in Proc. AutoEMC Workshop, Paris, France, June 2000.
7. IBIS I/O Buffer Information Specification, URL: https://www.eigroup.org/ibis/ibis.htm.
8. F. Haslinger, B. Unger, M. Maurer, M. Tröscher, R. Weigel, "EMC Modeling of Nonlinear Components for Automotive Applications," in Proc. 14th Int. Zurich Symp. on EMC, Zurich, Switzerland, Feb. 2001.
9. I. A. Maio, I. S. Stievano, F. G. Canavero, "Signal Integrity and Behavioral Models of Digital Devices," in Proc. 13th Int. Zurich Symp. on EMC, Zurich, Switzerland, Feb. 1999.
10. S. Baffreu, C. Huet, C. Marot, S. Calvet, E. Sicard, "A Standard Model For Prediction The Parasitic Emission of Micro-Controllers," in Proc. EMC Europe 2002, Sorrento, Italy, Sept. 2002, pp. 481ff.
11. ---, "Cookbook for Integrated Circuit Electromagnetic Model (ICEM)", Release 1.d, Oct. 2001, URL: www.ute-fr.com.
12. F. Canavero, I. A. Maio, I. S. Stievano, "Black-Box Models of Digital IC Ports for EMC Simulations," in Proc. 14th Int. Zurich Symp. EMC, Zurich, Switzerland, February 2001.
13. J. Held, B. Unger, L. Girardeau, "Measurement of the Supply-Current and Internal Impedance of VLSIs in a wide Frequency Range," Int. Symp. on EMC 2003, Istanbul, Turkey, May 2003.
14. Achar, R., Nackla, M. S., "Simulation of High-Speed Interconnects", IEEE Proceedings, vol. 89. no. 5, May 2001, pp. 693-727.
15. A. Ruehli, "Equivalent Circuit Models for Three-Dimensional Multiconductor Systems," IEEE Trans. Microwave Theory and Techniques, vol. 22, no. 3, March 1974, pp. 216-221.
16. A. Englmaier, B. Scholl, "EMC Modelling Strategy For Automotive Applications," Int. SimLab User Conference 2000, Munich, Nov 2000.
17. R. Neumayer, A. Stelzer, F. Haslinger, R. Weigel, "On the Synthesis of Equivalent Circuit Models for Multiports Characterized by Frequency-Dependent Parameters", IEEE Trans. Microwave Theory and Techniques, vol. 50, Dec. 2002.
18. B. Scholl, W. Kühn, P. Malnoult, H. Luzet, J.C. Kedzia. "Electromagnetic Interferences, Conducted and Radiated Emissions of a Harness towards a Receiving Antenna: Comparison of Numerical Results with Measurements," in Proc. Euro-PAM'98 Conference, Tours, France, Oct. 1998.
19. A. C. Cangellaris, "Distributed Equivalent Sources for the Analysis of Multiconductor Transmission Line Exited by an Electromagnetic Field", IEEE Transactions on Microwave Theory and Technique, vol. 36, no. 10, Oct. 1988, pp. 1445-1448.

Roland Neumayer (S'01) received the M.S. degree from Loughborough University, England, UK, and the Diploma degree from Johannes Kepler University, Linz, Austria, both in mechatronics, in 1999, and 2000, respectively. He is working towards the Ph.D. degree in the department of communications and information engineering at Johannes Kepler University, Linz. Currently, he is engaged with the European research project for continuous simulation of EMC in automotive applications (COSIME). His research interests include network synthesis, modeling and simulation techniques for EMC analysis.

Andreas Stelzer (M'00) received the Diploma Engineer degree in electrical engineering from the Technical University of Vienna, Austria, in 1994. In 2000, he received the Dr.techn. degree in mechatronics with honors sub auspiciis praesidentis rei publicae from the Johannes Kepler University. Since 2000 he is with the Institute for Communications and Information Engineering. His research work focuses on microwave sensors for industrial applications, RF- and microwave subsystems, EMC modeling, DSP and micro controller boards as well as high resolution evaluation algorithms for sensor signals.

Friedrich Haslinger received the Dr.techn. degree in mechatronics from the Johannes Kepler University in Linz, Austria in 2001. Mr. Haslinger then joined the BMW Group in Munich, Germany, where he is engaged with electromagnetic compatibility in cars. His main interests are various aspects of simulation of electromagnetic compatibility effects, especially the integration of non-linear noise sources.

Gernot Steinmair (S'01) received the M.S. degree from Loughborough University, England, UK, and the Diploma degree from Johannes Kepler University, Linz, Austria, both in mechatronics, in 1999, and 2000, respectively. He is working towards the Ph.D. degree at the department of communications and information engineering at Johannes Kepler University, Linz. Currently, he is engaged with the department for electromagnetic compatibility of Bayrische Motorenwerke (BMW AG), Munich, Germany. His research interests include EMC modeling, model order reduction and simulation techniques for EMC analysis.

Matthias Troescher received a Diploma in physics from the Technical University Munich, Germany, in 1994 and a Ph.D. degree in the Doctoral Program of Engineering Sciences from the Johannes Kepler University Linz, Austria, in 2000. From 1991 to 1994 he assisted in a European project for EMC simulation at the Fraunhofer Institute for Solid State Technology in Munich, Germany. In 1994 and 1995, he worked with the Institute for Radiation Protection in Munich, following which he joined the research department of BMW AG Munich. He joined SimLab Software GmbH (Munich) in 1999, where he is responsible for publications, technical support and product development.

Joachim Held was born in 1960 in Germany. He graduated in 1986 in electrical engineering at the Technical University of Erlangen, Germany. 1996 he joined the Siemens AG delivering EMC-support, where he works on innovative principles of inductive current dividing for supply-systems, and special measurement-methods for VLSI supply-currents.

Bernhard Unger was born in 1940 in Germany. After studies of physics and graduation he joined Siemens AG in 1972. In the first years he was mainly concerned with ECL-gate array development. Presently he is working on Signal Integrity and EMI issues.

Robert Weigel (F'01) received the Dr.-Ing. and the Dr.-Ing.habil. degrees, both in electrical engineering and computer science, from the Munich University of Technology in Germany, in 1989 and 1992, respectively. From 1994 to 1996 he was a Professor for RF Circuits and Systems at the Munich University of Technology. Since 1996, he has been Director of the Institute for Communications and Information Engineering at the University of Linz, Austria. In August 1999, he co-founded DICE - Danube Integrated Circuit Engineering, Linz, meanwhile an Infineon Technologies Development Center, which is devoted to the design of mobile radio circuits and systems. In 2000, he has been appointed a Professor for RF Engineering at the Tongji University in Shanghai, China. In 2002, he moved to Erlangen, Germany, to accept the Directorship of the Institute for Technical Electronics at the University of Erlangen-Nuremberg. EMC