EMC Modeling for Large Electronic Systems Thomas WEILAND#1 and Min ZHANG*2

#TECHNICAL UNIVERSITY DARMSTADT, Institute for Theoretical ElectromagneticField (TEMF) Department of Electrical Engineering and Information Technology, Schlossgartenstrasse 8, 64289 Darmstadt, Germany

1thomas.weiland@temf.tu-darmstadt.de *TONGJI UNIVERSITY, Modern Integrated Electromagnetic Simulation R&D Center (MIEMS)

School of Electronics and Information Engineering, 4800 Cao’an Road, Jiading District, Shanghai 201804, China 2min.zhang@tongji.edu.cn

Abstract— EMC simulations become more and more important because of the continuing increase of simulation capabilities. Even large electronic systems including small details may be simulated today. The key technology is to provide and integrate sub-models for modelling e.g. thin slots without a full 3D resolution. This paper will demonstrate the capabilities of state of the art simulation tool for practical problems.

Key words: System-level EMC simulation, compact models, EMS and EMI, PCB, cable harness, cabinet.

I. INTRODUCTION

Over the past three decades, electromagnetic (EM) simulation makes a rapid development from a mere single component-level to board-level [1], to cabinet-level, and nowadays, gradually even to a full system-level simulation. The increase of computing power and the development of new computational electromagnetic (CEM) methods both make it possible.

Larger electrical and electronic systems come up on the simulation list in terms of EMC. An accurate EMC simulation of working systems is always coming along with a prior knowledge of their behavioural functions, while what is actually interested in is their side-effects: conducted or radiated noises. This makes the simulation even more challenging, which involves both behavioural, normally circuitry, and side-effect, mostly EM, simulations.

We commonly accept the following classification of EMC simulations : a) PCB board level [2]: includes SI, PI, EMI, and EMS; b) Cable harness level [3]: involves SI, EMI, and EMS; c) Cabinet level [4]: EMI and EMS from PCBs and cables in

cabinet or enclosures; d) System level [5]: EMI and EMS of a standalone full-

functional device or equipment, involving in turn cabinets, enclosures, cables, and PCBs.

The above classification is based on the computational algorithm and computing effort. There exists another classification in practice which is in terms of their physical characteristics.

In narrow sense, EMC covers “CE” for conducted emission; “CS” for conducted susceptibility; “RE” for radiated emission; and “RS” for radiated susceptibility.

In a broad sense, however, all phenomena, as long as they are not desired or unexpected, may be tagged as EMC problems, like signal integrity (SI), power integrity (PI),

lightning and protection, high power electromagnetic pulse (HEMP), special absorption rate (SAR), hearing-aid compatibility (HAC), electrostatic discharge (ESD) , antenna placement, cross-coupling and interference among multiband antennas, all kinds of radio interference, and so forth.

In the following sections, we will demonstrate a series of practical EMC simulation examples by using the commercial software package – CST STUDIO SUITE™.

II. EMC SIMULATION SOFTWARE, METHODS, AND

THEORETICAL PREPARATIONS

CST STUDIO SUITE™ is a general purpose, commercially available comprehensive set of electromagnetic simulation software [6]. It consists of eight functional packages which are integrated in a uniform design environment. Five of the eight packages are useful in EMC simulations. They are: a) CST PCB STUDIO™ (PCBS): A PCB board level field

simulator, covering SI/PI/EMS/EMI tasks; b) CST CABLE STUDIO™ (CS): A cable harness level field

simulator, covering SI/EMS/EMI tasks; c) CST MICROSTRIPES™ (MS): A dedicated tool for

general EMS and EMI analysis for enclosures and cabinets with field and compact models;

d) CST MICROWAVE STUDIO® (MWS): A general purpose field simulation tool for system level EMS and EMI analysis;

e) CST DESIGN STUDIO™ (DS): A circuit and system simulator.

There are more than 11 different field simulation algorithms employed in the above package, namely, the finite integral time domain method (FITD), transmission line matrix method (TLM), finite integral frequency domain (FIFD), finite element method (FEM), method of moments (MoM), multilayer fast multipole method (MLFMM), modal order reduction (MOR), eigenmode solver, shooting and bouncing rays (SBR), partial element equivalent circuit method (PEEC), boundary element method (BEM), and various circuit simulation tools like linear and non-linear time domain and frequency method, IBIS, SPICE, etc.

One of the most appealing features in this software suite is the compact model library integrated with MS. This library offers a wide spectrum of so-called compact models for electrically small but EMC-significant geometrical features like slit/slot, thin film, vent, seam, etc. With the compact models, the simulation will become much faster, and at the

same time, much more accurate, for the numerical noise due to bad conditioned matrixes is reduced to the minimum.

Fig. 1 Compact models for vent, slot, contact, cable bundle

III. PRACTICAL EMC SIMULATION EXAMPLES

EMC covers almost all fields in electrical and electronic industry and application. In this section, we will demonstrate various simulation examples by using CST software: a) PCB board level: a 1 kVA buck-boost rectifier with

enclosure; b) Cable harness level: HUMVEE combat vehicle EMI and

EMS under EMP strike. Induced current in cables exposed in a broadband uniform EM field in a shielding chamber;

c) Cabinet level: a 436kVA high power convertor EMI； d) System level: mobile phone SAR; e) Direct lightning strike on A320; f) Antenna placement: multiband multiple antenna on M17

helicopter for interference and coupling study.

A. 1 kVA Buck-Boost Rectifier

Specifications are as follows: input AC380V/50Hz, output ±DC270V, nominal power 1kV, 3-phase full bridge rectifier with buck-boost PWM regulator at 48kHz, four MOSFETs used as two switchers, two in parallel for each switcher. The schematic and the real rectifier together with its simulation model can be found in Fig.2.

Fig. 2 Schematic, simulation model, photo

Fig. 3 3D radiation patterns at 3m away from the rectifier at 10MHz for

four different cooling slot patterns: circular and rectangular holes

The goal is to diagnose and minimize the maximum radiation to fulfil the ISO EMC class A standard, on condition that the area of cooling holes and their locations are fixed.

A novel linearization method is used in the simulation [7], which is conducted with CST DS in conjunction with CST MS. Radiation pattern and strength 3 meters away from the enclosure for different patterns of ventilation holes can be found in Fig. 3.

B. HUMVEE Cable EMI and EMS

Three different kinds of wires are laid along two routes in a HUMVEE car (Fig. 4): 2 bare wires, 2 twisted pairs, and 1 coaxial cable. They are terminated with 1 k for the bare wires, 100 for the twisted pairs, and 50 for the cable. Voltage sources of three different signals are fed to the wires: AC100kHz/pp24V for the wires, 200kHz square wave/duty factor of 20%/pp12V with 0.2us risetime and 0.5us falltime for the twisted pairs, and 5MHz/pp20V triangular signal for the cable. Cable and wires are bundled together along the chassis. Cross talks among all signal lines are to be investigated together with a prediction of the EM radiation level according to the CISPR 25 EMC standards.

Fig. 4 HUMVEE with cable harness and their radiation at 10MHz

CST CS and CST MWS are used in this simulation. The farfield radiation pattern at 10MHz is shown in Fig.5. On the left-hand side of Fig. 5, the SPICE modeling schematic in CST DS is presented. The right-hand part of Fig. 5 shows the input signal of the twisted pairs, two distorted signals at the loads of two twisted pairs.

Fig. 5 Cable SPICE model and signal distortion in twisted pairs

Fig. 6 HUMVEE under HEMP attack

Another reciprocal example with the precious EMI problem is EMS. Figure 6 demonstrate a simulation of HEMP exposure to HUMVEE. Inside the car there is bundle of cables. All windows are covered with shielding thin films. The simulation is carried out with CST MS. Induced current and probe signal inside the car are presented in Fig. 6.

C. 436 kVA AC-AC Power Convertor

The convertor works for two driving motors in an express train. Specifications are input 3-phase AC380/50Hz, output 3-phase AC420V/400Hz, nominal power 436kVA. The cabinet measures 1875mm x 1100mm x 640mm, built with 2mm-thick metal sheets (Fig. 7).

Fig. 7 Convertor cabinet photo and simulation model

The whole convertor together with its low voltage control unit is packed in the cabinet. The schematic is shown in Fig. 8.

Fig. 8 Schematic of convertor

There are six heavy duty IGBTs located at the lower central compartment of the cabinet, which is completely enclosed with metallic nets of 1mm thick wires and a cheque size of 2mm x 2mm. Copper buses are used to connect all the components. The cabinet is cooled with a ventilator at the top right deck with a cover made of thin shielding nets. Shielding nets are used elsewhere except for the solid wall areas. The contact tolerance of all the movable parts like doors is 1mm with gasket profiles.

Fig. 9 Near E field distribution and strength at three frequencies

CST MS and CST DS are used for the simulations. All the small geometrical features like nets, ventilator, seams of the doors and the cabinet itself are replaced by the corresponding compact models provided by MS. Stored and radiated power at three specific frequencies are shown in Fig. 9. From a previous simulation we found the strong radiation was attributed to the badly shielded ventilator. After redesigning a new shielding nets for the ventilator, the radiation is reduced considerably and overall radiation level is below the limit.

D. Complete Mobile Phone with Human Head

Mobile phones become more and more popular recently. The fear about being irradiated by its RF power gradually draws public attentions. People now tend to simulate a real phone instead of early day’s simple metal block with a stub antenna on it. This increases the computational effort tremendously. Every single detail must then be resolved to order to achieve more realistic and convincing simulation results, specifically, the special absorption rate – SAR, which is a direct measure of how much power human tissue absorbs during the phone calls.

Fig. 10 Complete mobile phone placed at human head wearing glasses (SAR)

Figure 10 shows a complete mobile phone placed against a human model. The model used here is a Chinese female with the voxel resolution of 0.67×0.69×0.5 mm3. The mobile phone is equipped with a dual band antenna, powered at the peak power of 1W. The two plots at the bottom of Fig. 10 show the SAR distribution at 900MHz (left) and 1800MHz (right) under the 10g averaging scheme. Details about this simulation can be found in [8]. This simulation reveals that glasses with metallic frames can eventually increase SAR at the human head.

E. Direct Lightning Strike on A320

Lightning strike is a severe natural phenomenon to in-flight aircrafts. Simulation is based on the MIL-STD-464A lightning waveform as shown in Fig. 11. The aircraft A320 measures about 37.57m long, 34.09m wide, and 11.76m high, while the thickness of the aluminium skin only around 1~1.5mm. Such a huge aspect ratio and the low frequency eventually prevent us from using the common full-wave EM methods like FDTD or FEM and the alike.

Fig. 11 MIL-STD-464A lightning waveform (upper left), strike route on A320 (upper right), contact seams (lower left), and cables (lower right)

We use CST MS to accomplish this simulation by applying the thin film compact model for the thin aluminium cover,

which is penetrable at this frequency range, seams for all the doors and windows.

The strike route, doors and windows, and wire and cable layout are presented in Fig. 11. a detailed description of this simulation can be found in [9]. The intensity of the discharging currents in the outer and inner surface of the aircraft is shown in Fig. 12. Obviously the inner current density is much lower than the outer one. A further study of induced current in wire and shielded cable on board confirms once again the strong shielding effect by the skin. It is also concluded that bare wire is much more vulnerable to the intense lightning by several orders of magnitude (Fig. 13).

Fig. 15 Coverage of surveillance radar pattern at 12GHz

IV. CONCLUSION

EMC simulation of large electronic system can be broken down to the four major levels: PCB board level, cable harness level, cabinet level, and finally system level. This classification is made based on the numerical methods to be used. In a broad sense, EMC covers almost all kinds of electromagnetic side-effects, as long as they are undesired or unexpected.

The complete EMC simulation suite – CST STUDIO SUITE proves to be one of the most powerful tools which can address all levels of EMC problems. Practical EMC simulation examples are presented in this paper to support this statement.

Compact models for electrically small but EMC-significant geometrical features are very useful and efficient in conjunction with the other full-wave simulation methods in the CST software.

Fig. 12 Temporal development of transient surface current on outer side of the aluminium shell (upper two) and on inner side of the shell (lower two)

REFERENCES

[1] Shogo MIYATA, Yoshiki KAYANO, and Hiroshi INOUE , “EM Radiation through Aperture of Metallic Enclosure with a PCB inside” APEMC, pp. 658-661, May. 2008

Fig. 13 Induced current in bare wire (left) and shielded cable (right)

[2] Huanghui Shen, Zhensong Wang, and Weimin Zheng, “PCB Level SI Simulation Based on IBIS Model for High-speed FPGA System”, ICEMI, pp. 75-79, 2009

F. Antenna Placement on M17 Helicopter

An optimum antenna placement on large military objects like aircraft and warship is an important factor for a high reliability. This also belongs to the EMC category.

[3] Burghart, T.; Rossmanith, H.; Schubert, G., Evaluating the RF-emissions of automotive cable harness, Electromagnetic Compatibility, 2004. EMC 2004. 2004 International Symposium on Volume 3, 9-13 Aug. 2004 Page(s):787 - 791 vol.3

[4] Martin Paul Robinson, Trevor M. Benson, Christos Christopoulos, John F. Dawson, M. D. Ganley, A. C. Marvin, S. J. Porter, and David W. P. Thomas, “Analytical Formulation for the Shielding Effectiveness of Enclosures with Apertures” IEEE Transaction on Electromagnetic Compatibility, vol. 40, No. 3, pp 240-248, Aug. 1998

[5] Renko,A, Arsian, A.N, Yuferev.S, Uusimaki.M, “3D Electromagnetic Modeling and Design Flow in System Level,” in IEEE International Symposium on Electromagnetic Compatibility, 2003, pp.1077-1080.

[6] CST AG, Germany, CST STUDIO SUITE Reference Manual, 2009. www.cst.com

Fig. 14 M17 helicopter simulation model and antenna locations

More than antennas covering a broad frequency band are attached to the M17 helicopter. To determine the cross interference level, an accurate simulation by using full-wave simulation methods is necessary. We use CST MWS to accomplish this task successfully. The simulation time ranges from a few minutes for low frequency up to 10 hours for higher band, occupying up to 17GB RAM. The simulation results agree very well with the measurement, delivering a sound insight for the placement design engineers.

[7] Min Zhang, Rui Wu, “A Novel Linearization Method for Full Wave EMI Simulation of Switching Power Supplier”, this conference

[8] Min Zhang, Xiao Wang, “Influence on SAR due to Metallic Frame of Glasses based on High-Resolution Chinese Electromagnetic Human Model”, this conference

[9] Min Zhang, Zhiyong Huang, “Transient Current Burst Analysis induced in Cable Harness due to Direct Lightning Strike on Aircraft”, this conference