Proc. of the 2013 International Symposium on Electromagnetic Compatibility (EMC Europe 2013), Brugge, Belgium, September 2-6, 2013 Influence of a Traction Battery’s Input Impedance on Conducted Emissions of an Automotive HV Inverter M. Reuter1*, S. Tenbohlen1, W. Köhler1 1 Institute of Power Transmission and High Voltage Technology (IEH), University of Stuttgart, Germany * Martin.Reuter@ieh.uni-stuttgart.de impedance and the value of the load impedance [6]. As the disturbance sources within a component are constant at a given operating point, the load impedances dominate the amplitudes of conducted emissions in HV systems [2]. Thus, comparable conducted disturbances on component and system level are only achieved, if the load impedances of the disturbance sources within the tested device are similar. But the load impedances in the standardized test setup differ significantly to the impedance situation within the automotive HV system’s environment. This contribution quantifies the impedances of both situations and their difference in measured conducted emissions of an automotive traction power inverter. Abstract — Conducted emissions of automotive high voltage (HV) components recently are measured according to CIPSR 25. If the tested device conforms to specific limits on component level (set by customer and supplier agreements) it is inserted into the automotive system. Thus, the EMC performance of the investigated device has to be independent of the test level, e.g. on component or vehicle level. Comparing the standardized test setup with an automotive HV system, the main difference consists in the termination of the cable harness. In a component level EMC test setup the HV harness is terminated with line impedance stabilization networks (LISNs), whereas implemented in a vehicle the cables are connected to a traction battery. This contribution investigates the difference in the conducted emissions of an automotive power inverter, when it is introduced into an electromagnetic environment that is similar to vehicle level. Therefore, the input impedances of a HV battery are characterized and modeled. This battery emulation is introduced into the component level test setup and the deviations in the conducted emissions of a traction inverter are determined. II. In lack of standardized EMC test setups for HV systems, CISPR 25 is recently also applied to determine the conducted emissions of HV components. This standard prescribes a measurement setup consisting of the device under test (DUT), its cable harness with a rest bus simulation and LISNs for termination of the supply cables [7]. It was developed to emulate the impedance situation within a typical low voltage (LV) harness, based on impedance measurements [8]. DUT and cable harness are laid up as close as possible to their later configuration in the vehicle, and the LISNs are used to emulate a standardized car environment. Keywords – Conducted Emissions, CISPR 25, Traction Battery, Automotive HV Network Impedance, Electric Vehicles. I. INTRODUCTION The electromagnetic compatibility (EMC) of electric traction systems is a major task in the development process of electric driven cars. Most electric vehicle (EV) concepts position the required traction battery underneath the rear trunk, whereas power electronics and electric machine are placed in the front motor compartment. Thus, a transmission of electric power is required through the entire car, with a cable harness of about 3 – 4 m length. An increase of the operating voltages up to 120 – 400 VDC reduces ohmic losses of the power transmission. The battery, power electronics and cable harness form a high voltage traction network, which is assembled entirely shielded to reduce electromagnetic interferences (EMI). The applied coaxially shielded HV cables act as transmission lines in the EMC frequency range [1]. Thus, the termination of the traction harness effects the conducted emissions of HV components [2]. The LISNs are needed for three reasons: First to power the DUT and simultaneously providing a radio frequency (RF) decoupling of the supply from the cable harness. This is required to prevent the DUT’s disturbances from propagating along the supply cables. The supply lines are usually much longer than the harness itself and their lengths differ in each EMC laboratory. Therefore, only with an effective decoupling, comparable test results can be achieved, independent of power source and supply lines. Second task is to emulate the input impedance of a LV cable harness having a length of approximately 5 m. LV cable harnesses consist of unshielded wires, bundled close to each other. Their termination does not impact the input impedance of the harness above a length of ℓ > λ/6 because of high series inductance and crosstalk [9]. In contrast, the coaxially shielded HV cables act as transmission lines in this frequency range and transform the termination impedance into the cable’s input impedance. Thus, the termination affects significantly the input impedance of the HV harness [2]. The third reason for using LISNs is that they provide a voltage measurement port for EMI test receivers. Thévenin's theorem states that any combination of linear electrical circuits is equivalent to one ideal voltage source and a series resistor [3], [4]. Applied to the field of EMC, the sources of conducted emissions can be approximated by an ideal voltage source and a series impedance [5], [6]. Therefore the amplitudes of conducted emissions are determined by three factors: The amplitude of the ideal voltage source, its source 978-1-4673-4980-2/13/$31.00 © 2013 IEEE MEASUREMENT OF CONDUCTED EMISSIONS ON COMPONENT LEVEL 229 Proc. of the 2013 International Symposium on Electromagnetic Compatibility (EMC Europe 2013), Brugge, Belgium, September 2-6, 2013 III. local minima, the transmission coefficients reach local maxima (5, 20, 30 MHz). Thus, the HV battery can be described as a symmetrical and reciprocal two-port device. SCATTERING PARAMETERS OF AN AUTOMOTIVE TRACTION BATTERY As the main difference between standardized test setups and automotive environment is the termination of the HV cables, this termination impedances needs to be characterized. The input impedance of LISNs is known and prescribed in CISPR 16 [10]. The RF properties of an automotive HV battery are determined with the test setup described in [1]. A photo of the measurement setup is shown in Figure 1. c DUT IV. CALCULATION OF BATTERY IMPEDANCES Scattering parameters are able to describe the complete RF behavior of a DUT as black box. They are not suitable to develop a physical model, as they do not visualize internal structures of the tested device. To receive a physical model suited for EMC issues, a calculation of common mode (CM) and differential mode (DM) impedances is more useful. Figure 3 shows a simplified EMC equivalent circuit of the HV battery with its CM impedances ZG1 and ZG2 to ground and one DM transmission impedance ZX. d a b Figure 1. Measurement setup to determine the scattering parameters of an automotive HV battery, consisting of DUT, coaxial adapter (a), DC block (b), battery controller (c) and Network Analyzer (d) [1] The tested device is an automotive HV traction battery with an operating voltage of 130 VDC, a power rating of 25 kW and a total capacity of 9.6 Ah. A DC blocking device (b) enables the measurement of the RF behavior of the DUT using a Network Analyzer (d) in the frequency range of 300 kHz – 200 MHz. The coaxial adapter (a) allows a connection of Port 1 of the VNA to HV- at the battery and Port 2 to HV+. Figure 2 shows the resulting scattering parameters of the HV battery. Figure 3. EMC equivalent circuit of the battery and calculation methodology of CM and DM impedances using Z- parameters The measured scattering parameter matrix [S] of the tested HV battery can easily be converted into an impedance parameter matrix [Z] as e.g. shown in [11]. The impedance parameters [Z] of the simplified circuit in Figure 3 are known as function of the equivalent circuit impedances , , . An inversion of this equation system leads to the CM and DM impedances in dependency of the ZParameters as displayed in Figure 3. The calculation result of CM and DM impedances is plotted in Figure 4. 0 S11, S22 -5 1k |S| / dB -10 -15 S21, S12 |Z| / Ω -20 -25 -30 1 ZG1, ZG2 100 10 f / MHz 10 1 100 0.1 Figure 2. Measured scattering parameters [S] of investigated HV-Battery The scattering parameter curves of the HV battery in Figure 2 present the reflection coefficients S11 and S22, which are very similar to each other. Thus, the battery is a symmetric device up to frequencies of ~30 MHz. Above this frequency the reflection coefficients differ, but show similar trends. In the transmission parameters S21 and S12 there are no visible deviations in the considered frequency range. A comparison of reflection and transmission parameters presents that they have a diametrically opposed behavior: At frequencies, where the reflection coefficients peak (3.5, 11, 25 MHz) the transmission coefficients minimize. Vice versa where the reflections have ZX 1 10 f / MHz 100 Figure 4. Common mode impedances (ZG1 / ZG2) and differential mode impedance (ZX) of the investigated HV Battery; a main series resonance at 5 MHz occurs The CM impedances ZG1 and ZG2 have a series resonance at fR = 4.5 MHz. ZG1 and ZG2 show capacitive behavior below the resonance frequency fR and an inductive trend above. The DM impedance ZX starts at low resistive values, increasing to a maximum at ~3.5 MHz before minimizing to the main series 230 Proc. of the 2013 International Symposium on Electromagnetic Compatibility (EMC Europe 2013), Brugge, Belgium, September 2-6, 2013 resonance at 4.5 MHz. Above 5 MHz several minor resonances can be observed, overlaying a mainly inductive trend. Ground capacitance This impedance behavior can be explained with the internal structure of the HV battery: At low frequencies the CM impedance is dominated by the capacitance to the shielded housing. The capacitance and the line inductances of the cabling form a series resonance circuit. Thus, for high frequencies the line inductances dominate the CM input impedance. As these parasitic elements are distributed parts, they can be diverted into smaller elements, which form further resonances in the higher frequency range. But these resonances are way less dominant influencing the CM input impedance. T+ Model of aggregated single cells HV connector / main contactor TGround capacitance The DM impedance ZX is low resistive for low frequencies. Underneath the investigated frequency range it will converge to the DC resistance of the battery [12]. Each single battery cell can be modeled as plate capacitor parallel to a conducting electrolyte [12]. All cells are interconnected in series. Thus, the total wiring inductances add, whereas the cumulative capacitance diminishes. With increasing frequency the electrolyte becomes less conductive. This leads to an increasing trend of the DM impedance. Above frequencies of 1 MHz the electrolyte does not influence ZX any more, as the velocity of the electrolyte molecules is too low. The effect was verified using a battery dummy without electrolyte fluid, having the same internal assembly as the investigated battery. The main series resonance of the CM impedance also appears in the DM. This is caused by the two interconnected CM impedances, which are more conductive than the series interconnection of the battery cells at the CM resonance frequency fR. Above the series resonance frequency the DM impedance also shows inductive behavior, caused by the internal wiring of the battery cells. In the higher frequency range overlaying minor resonances can be observed, again caused by smaller parts of the distributed wiring structures. V. Internal wiring External wiring Figure 5. SPICE Simulation Model of HV Battery with external connection inductivities The battery model of Figure 5 consists of a ground capacitor (CG = 3.8 nF) in series to the wiring inductances with a total value of LS = 370 nH. The aggregation of all single cells is modeled as a total capacitance of CC = 6.7 μF and a series inductance of LC = 320 nH. The cumulative wiring inductances represent cables of a length of approximately ℓ ~ 1.06 m. As within the main contactor the HV conductors are nearby, a small capacitor (CMC = 330 pF) is needed to reproduce the capacitive cross coupling. As a simulation of presented SPICE model showed similar RF properties as the HV battery, the model’s network was built up to emulate the battery within a component level test setup. VI. BATTERY EMULATION FOR EMC TESTS OF HV COMPONENTS As HV battery prototypes are very rare, expensive and dangerous to handle, a battery emulation is used instead. Using the SPICE simulation model of the investigated HV battery, a network of concentrated elements is built up. It simulates the input impedances of the automotive HV battery within a component level EMC test setup. Thus, for the measurement of conducted emissions, it behaves in a similar manner as a HV battery, without having the drawbacks of real HV batteries: There are no flammable chemicals inside and no energy storage. When the power supply is switched off, the device is inherent safe. As the system is powered by an external supply, there is no recharging required and the state of charge (SOC) does not change within the duration of the test. Also, there is no need for a CAN communication or rest bus simulation, whose emissions would interfere with the test requirements. MODELING OF BATTERY IMPEDANCES The main advantage of regarding CM and DM impedances instead of scattering parameters, is that the modal impedances allow a “grey box” modeling approach. Whereas a complete model of all internal structures is highly complex and very difficult to achieve, a black box model does not represent physical expressions of the internal structures. A compromise of these two opposed modeling approaches is to recognize dominant structures and to abstract less influent parts. This leads to a “grey box” model, which is less complex and faster to achieve, but has physical explanations for the major structures of the modeled device. In the case of presented HV battery a simplified model was developed, that represents the capacitances of the cells, the internal wiring, the capacitance to ground and the capacitive coupling within the HV main contactor. Figure 5 shows this simplified RF model of the HV battery, showing similar impedance properties as the tested device. This model is intended to emulate the HV battery’s input impedances in a frequency range of 300 kHz – 30 MHz. The developed battery emulation supplies the tested device with DC currents up to 20 A and voltages up to 400 V. It also decouples the supplying power source and DUT within the EMC test frequency range. Additionally it enables the measurement of disturbance voltage with an EMI test receiver, which is not directly possible using a real HV battery. A photo of the battery emulation is shown in Figure 6. 231 Proc. of the 2013 International Symposium on Electromagnetic Compatibility (EMC Europe 2013), Brugge, Belgium, September 2-6, 2013 The CM input impedances of HV battery and battery emulation are equal up to a frequency of ~10 MHz. Above this frequency the higher order resonances of the distributed elements could not be represented with the simple circuit shown in Figure 5. Nevertheless, the dominant inductive trend of the CM impedance is obtained. A comparison of the battery’s DM impedances and battery emulation shows, that the battery impedances can be emulated up to ~9 MHz. Again, the higher order resonances cannot be represented because of the simplified network model. Within a component level EMC test setup the DUT is connected via coaxially shielded HV cables to the LISNs. These cables act as wave guides at frequencies where the cable’s length is more than ℓ > λ/6 [1]. Thus, their termination impedance is transformed according to the transmission line theory to the cable’s input impedance. In this frequency range a higher accuracy of the battery emulation is not required, as long as the impedance trend is represented. Figure 8 shows a comparison of the HV cable’s input impedance, when it is terminated with a HV battery and the battery emulation. DUT T+ a Supply Filter Battery impedance emulating network a Supply HV- DUT T- Figure 6. Battery Emulator inside a shielding box. It consists of three functional blocks: A supply filter for decoupling of power supply and DUT, battery impedance emulating network and two impedance converters (a) for voltage measurements on the traction lines (T+/T-). Figure 6 shows the battery emulation device with its three functional blocks. On the left hand side there is a supply filter to decouple the power lines from the conducted disturbances of the DUT. The supply filter provides an attenuation of > 50 dB up to 80 MHz between power supply inputs and DUT output ports. Thus, there is no influence of the supply’s input impedance on the battery emulation’s input impedance. Additionally, the filter yields an effective decoupling of RF disturbances between supply and DUT, respectively the measurement port. The battery impedance emulating network is comparable to the schematic shown in Figure 5. To quantify the conducted emissions, a measurement port is required. This measurement device is connected in parallel to the battery impedance emulating network. To avoid a change of the emulation’s input impedance, the measurement device needs a high ohmic input. For RF measurement purposes a 50 Ω source impedance of the measurement port is required. Thus an impedance converter circuit (a) was used for each traction power line (T+ and T-). The 3 dB corner frequency of the measurement port lies at ~67 MHz, because of a limited gainbandwidth product of the applied operational amplifiers. Figure 7 shows a comparison of the CM and DM impedances within the HV battery and its emulation. 1k 10 100 |ZDM| / Ω |ZCM| / Ω 100 1 10 f / MHz 100 10 f / MHz |ZIn,DUT| / Ω 10 100 HV Battery 1 . 0,1 1 10 f / MHz 100 Battery Emulation 10 1 . 0,1 HV Battery 1 10 f / MHz 100 Figure 8. Input impedances of HV battery and battery emulator before and after impedance transformation The curves of Figure 8 show the dominance of the impedance transforming behavior above 10 MHz, where the HV cables begin to act as transmission lines. Although the battery emulation is not able to reproduce the battery resonances at frequencies higher than 20 MHz, the input impedance of the HV harness is similar. Both devices, which terminate the cable, are similarly mismatched to the wave impedance and therefore result in a similar HV cable’s input impedance. Only smaller deviations can be observed at ~10 MHz and around 15 MHz. In this region of frequency the HV cable begins to behave as wave guide and its impedance transforming behavior is not yet dominant enough to cover the differences. HV Battery 1 ZIn,Cable 1k Battery Emulation 100 1 . 0,1 HV cable ZL = 18.5 Ω, ℓ = 3.7 m ZIn,DUT 1k Battery Emulation 10 HV Battery 1 . 0,1 1k Battery Emulatio HV Battery / Battery Emulation |ZIn,Cable| / Ω Supply HV+ As the mentioned deviations are very small, it can be stated that the battery emulation represents a valid HV battery model within a component level test setup. The emulation meets the battery’s impedance in the lower frequency range. Above 10 MHz it is sufficient to reproduce the impedance trend of the battery for the tested device. 100 Figure 7. Comparison of CM impedance (left) and DM impedance (right) of HV traction battery and battery emulation 232 Proc. of the 2013 International Symposium on Electromagnetic Compatibility (EMC Europe 2013), Brugge, Belgium, September 2-6, 2013 resonances following. In contrast, the input impedance of the cable terminated with a HV battery starts at ~100 Ω declining to the main series resonance of the battery at 3 MHz. This series resonance is followed by a parallel resonance at 9 MHz at ~400 Ω and regular line resonances in the higher frequency range. Above 40 MHz there are no visible deviations in the input impedance of the cable, as for both devices the termination impedance of the cable is higher than the wave impedance. Below 10 MHz there are significant deviations between 20 log(ZLISN/ZBat) = 33 dBΩ at 3 MHz and -25 dBΩ at 7.5 MHz. VII. INFLUENCE OF THE BATTERY IMPEDANCE ON CONDUCTED EMISSIONS OF A TRACTION INVERTER To investigate the influence of the battery impedance on conducted emissions of HV components, a power inverter for a hybrid driven car is tested. Figure 9 shows a photo of the test setup according to CISPR 25. HV LISNs / Battery Model DUT: Inverter M In Figure 11 the recorded disturbance voltages of the power inverter depending on the HV cable termination are plotted. Cable Harness LV LISNs 120 Figure 9. Measurement setup on component level according to CISPR 25 for a HV traction inverter U / dBμV The setup consists of the inverter (DUT), which generates three phase AC to supply a synchronous machine (M). The inverter is fed with DC via two HV traction cables of ℓ = 3.7 m length. The HV and LV cables are bundled into one cable harness and the supply lines are terminated with LISNs. The electric traction system is operated at a typical working point driving a generator with a mechanical power of Pm = 2.8 kW (outside the anechoic test chamber). To investigate the EMC performance of the inverter on system level, the HV LISNs are replaced with the presented battery emulation. The conducted emissions are recorded with an EMI test receiver (IFBW 120 kHz, peak detector). |ZIn,Cable| / Ω 100 By changing the cable termination to the battery emulation at the same operating point of the inverter a different spectrum occurs: At low frequency range the disturbances are reduced by 5 – 10 dB. But at the battery’s main resonance (3 MHz) the disturbance voltage peaks and superposes the disturbances according to CISPR 25 by ~10 dB. This peak is followed by a disturbance minimum, where the voltage emissions are up to 33 dB (at 6.3 MHz) lower as measured with LISNs. From 10 MHz the voltage disturbances emerge with increasing frequency. Above 40 MHz the disturbance voltages are below the noise floor due to the limited bandwidth gain product of the used operational amplifiers within the impedance converters. Because the load impedances of both cases are identical above ~40 MHz the conducted emissions in this frequency range will be similar. Battery Emulation 10 f / MHz 10 f / MHz Figure 11 shows that the conducted voltage disturbances measured according to CIPSR 25 begin at 110 dBμV. They minimize at around 1 MHz with a value of 90 dBμV before peaking at 6 MHz with a value of 115 dBμV. For higher frequencies a declining trend can be observed with overlaid line resonances caused by the termination impedance mismatch of the HV cables [2]. LISN 1 1 Figure 11. Disturbance voltage of the power inverter with LISNs and battery emulation as cable termination 10 0.1 Battery Emulation 60 20 1k 1 80 40 The traction inverter is supplied via two HV cables, whose input impedances dominate the conducted emissions of the tested device. A comparison of the HV cable’s CM input impedance within the test setup is plotted in Figure 10. These impedances represent the load conditions for the disturbance sources inside the tested power inverter. 100 LISN 100 100 Figure 10. CM input impedance of the HV cables according CISPR 25 (marked with LISN) compared to the presented battery emulation as cable termination. This impedance loads the CM disturbance sources within the tested inverter As for radiated emissions the CM current is one dominant influencing factor, a current clamp is used to determine the current emissions of the power inverter. The results are plotted in Figure 12. CIPSR 25 specifies that the HV cables need to be terminated with LISNs, resulting in the green impedance curve of Figure 10. This load impedance for CM sources starts at around 8 Ω (300 kHz) and follows the LISN’s input impedance up to ~3 MHz. At higher frequencies the impedance decreases to the first line resonance at ~15 MHz with regular line 233 Proc. of the 2013 International Symposium on Electromagnetic Compatibility (EMC Europe 2013), Brugge, Belgium, September 2-6, 2013 100 80 I / dBμA emissions on the traction lines. When a HV component, whose EMC performance is validated according to the only existing component level test setup, is connected to a HV battery its EMC performance changes dramatically. The deviations are up to 20 dB in the disturbance current and even 33 dB in the voltage emissions. Battery Emulation 60 LISN 40 These changed conducted emissions, caused by the changed impedance situation, can lead to ineffective EMC countermeasures. To ensure the EMC of electric traction systems the cable’s termination impedances need to be considered in the development of EMC concepts. 20 0 1 10 f / MHz 100 ACKNOWLEDGMENT Figure 12. Disturbance current of the power inverter with LISNs and battery emulation as HV cable termination The authors would like to thank the Robert Bosch GmbH for funding and supporting the EMC measurements of an automotive electric traction system. Within the standardized setup at low frequencies the disturbance currents are higher, because of the lower CM input impedance. Above 600 kHz the currents with the battery emulation exceed the ones of the LISN setup, because the input impedance of the emulation is lower. A maximum deviation of 25 dB occurs at the battery’s main series resonance (3 MHz). Above this frequency the difference become smaller, but still there is a minimum deviation of 10 dB in a broad frequency range of 900 kHz – 7 MHz. At 11 MHz there is a resonance with 15 dB higher current amplitudes, but above the disturbance currents equal each other, because the input impedances become similar. REFERENCES [1] M. Reuter, S. Tenbohlen, W. Köhler, and A. Ludwig, "Impedance Analysis of Automotive High Voltage Networks for Measurements," in EMC Europe 2011, York, UK, 2011, p. 6. EMC [2] M. Reuter, S. Tenbohlen, and W. Köhler, "Influence of Termination Impedance on conducted Emissions in Automotive High Voltage Networks," in EMC Europe 2012, Roma, 2012, p. 6. [3] L. Thévenin, "Extension of Ohm’s law to complex electromotive circuits," Annales Télégraphiques, Comptes Rendus hebdomadaires des séances de l’Académie des Sciences, vol. 97, pp. 159–161., 1883. [4] Mohamed F. Moad, "On Thevenin's and Norton's Equivalent Circuits," The measurements show considerable differences in the EMC performance of the tested device, although its operating point is kept constant. The occurring deviations are explained by changed load impedances of the disturbance sources within the DUT. Therefore, the conducted emissions of HV components differ substantially, when they are inserted in a traction system. Thus, the same component at the same operating point may emit a different disturbance spectrum, as measured according to CISPR 25. Education, IEEE Transactions on, vol. 25, no. 3, pp. 99 - 102, Aug. 1982. [5] F. Tesche, M. Ianoz, and T. Karlsson, EMC Analysis Methods and Computational Models. New York: Wiley and Sons, Inc., 1997. [6] C. Paul, Introduction to Electromagnetic Compatibility. New York: Wiley & Sons., 1983. [7] CISPR 25. (2008) Radio disturbance characteristics for the protection of receivers used on board vehicles, boats and on devices – Limits and methods of measurement. IEC/CISPR 25:2002 / DIN-EN 55025:2003. [8] S. Yamamoto and O. Ozeki, "RF Conducted Noise Measurements of VIII. CONCLUSIONS AND OUTLOOK Automotive Electrical and Electronic Devices Using Artificial Network," in IEEE Trans. Veh. Technol., vol. VT-32, no. 4., 1983, pp. 247 -253. The EMC performance of HV components is recently tested according to CIPSR 25. This standard specifies the use of LISNs as supply cable termination. But the input impedances of HV batteries differ significantly from the input impedance of LISNs. This causes a changed EMC performance of the tested device, when it is inserted into the automotive system environment. [9] H. Rebholz, Modellierung leitungsgebundener Störgrößen in der Komponenten- und Fahrzeugmessung, 1st ed. Stuttgart: Sierke Verlag, 2010. [10] CISPR 16. (2005) Specification for radio disturbance and immunity measuring apparatus and methods. IEC/CISPR 16:2003 / DIN-EN 55016:2004. [11] P. Russer, Electromagnetics, microwave circuit and antenna design for To quantify the amount of disturbance deviation, a battery emulation has been developed, based on scattering parameter measurements of an automotive traction battery. The battery emulation replaces the LISNs within the component level EMC test setup and the differences of the conducted emissions in the HV harness of an electric traction system are recorded. communications engineering. Boston: Artech House, 2003. [12] Although the operating point of the traction system is kept constant there is a significant difference in the conducted 234 E. Hoene, S. Guttowski, R. Saikly, W. John, and H. Reichl, "RFProperties of Automotive Traction Batteries," in IEEE International Symposium on Electromagnetic Compatibility, Istanbul, 2003, pp. 425 428 Vol.1.