SPEEDAM 28 International Symposium on Power Electronics, Electrical Drives, Automation and Motion Common and Differential Mode EMI Filters for Power Electronics V. Serrao, A. Lidozzi, L. Solero and A. Di Napoli University ROMA TRE DIMI, Via della Vasca Navale, 79 146 Roma, (Italy) Abstract--Electromagnetic interference (EMI) has become in recent years a great concern to deal with, in every kind of application and especially in power electronics. As a consequence of international and national standards requirements, the insertion of an EMI filter in power electronic converters assembly is today a common practise. In this paper the EMI characterization of the power conversion system that manages the power flows between the generation and storage units of a hybrid vehicle is presented. With the purpose of reducing conducted emissions and designing a proper EMI filter, the overall noise has been separated into its common mode (CM) and differential mode (DM) components. Experimental results underline the importance of considering CM and DM noises separately in order to minimize the filter components count and size. Index Terms--Common mode (CM) and Differential mode (DM) conducted emissions, Electromagnetic interference (EMI), EMI filter design, power electronics. I. INTRODUCTION In every kind of application making use of electric or electronic device, electromagnetic interference is a great concern to deal with. Today in fact, international and national standards requirements have been rising in order to guarantee that every apparatus could operate without introducing significant electromagnetic disturbances to the environment and without significant degradation of operational performance in the presence of external electromagnetic disturbances that can be expected in normal use. Power electronic converters are nowadays widely employed in industrial, commercial and residential areas, so that great effort has been spent in order to obtain high power density together with a size reduction of the overall power conversion unit. To this purpose new topologies and control strategies have been developed with the aim of increasing the switching frequencies of the active devices. Whereas on the one hand these procedures can reduce power converters size, on the other hand they contribute to worse electromagnetic interference phenomena. The main source of EMI in power electronics comes in fact from a rapid change in voltages and currents, following a certain PWM (Pulse Width Modulation) technique. In automotive applications these issues are even more considerable (and the relative regulations more severe) as today s vehicles are equipped with a large number of sensitive electronic communication and control systems together with several switching power converters. The first are responsible for radiated emissions (high frequency: 3 MHz 1 GHz) while converters are the 978-1-4244-1664-6/8/$25. 28 IEEE 918 main cause of conducted emissions (low frequency: 9 khz 3 MHz). Drives in electric (EV) and hybrid (HEV) vehicles are usually constituted of the cascade of ac motor and inverter fed by the dc-link, being the dc-link supplied by either battery unit (BU), or electric generator (EG), or fuel cell (FC), or a combination of the mentioned components together with an ultracapacitor tank (UC). The simultaneous presence of several power converters makes an EMI analysis more difficult, as noise propagation and coupling paths cannot be exactly predicted. To reduce the current noise from power cables, an EMI filter, a lumped or distributed inductive-capacitive circuit, must be employed. The size of the filter is related to the desired degree of attenuation of harmonics of the current noise. In order to meet standards specifications, filtering techniques designed with purposes of reducing EMI effects in adjustable speed drive are nowadays commonly used in electric traction systems. In this paper the EMI characterization of the power conversion system that manages the power flows between the generation and storage units of a hybrid vehicle is presented. An extensive experimental test campaign is carried out in a semi-anechoic chamber in order to evaluate the overall EMI performance of the single power converter and, in particular, to estimate which is the contribution to the system emission levels due to auxiliaries as the control platform and the related electronic boards (measures acquisition, filtering, driving circuits and power supply distribution network). Then, starting from the experimental results analysis, it is possible to identify the main EMI sources and coupling paths so that several adjustments can be performed to minimize conducted emissions. With the purpose of reducing conducted emissions and designing a proper EMI filter, the overall noise has been separated into its CM and DM components. Experimental results underline the importance of considering CM and DM noises separately in order to minimize the filter components count and size. II. SYSTEM OVERVIEW The power conversion system we are going to analyze from EMI point of view, has been designed and realized on purpose for the HOST (Human Oriented Sustainable Transport mean) project. This project, sponsored by the European Community, aims at developing an innovative modular transport mean suitable for the urban transport of persons and goods. In order to fulfill simultaneously the objectives of extremely low CO2, gaseous and particulate Authorized licensed use limited to: BIBLIOTECA D'AREA SCIENTIFICO TECNOLOGICA ROMA 3. Downloaded on May 17,21 at 1:3:39 UTC from IEEE Xplore. Restrictions apply.
pollutants reduction, within a medium term, the HOST consortium will produce multipurpose modular vehicle platforms (HOST) capable to integrate, in an optimized cost effective way, the most promising alternative fuel set, and the newest combustion modes technologies. HOST powertrain has a thermal-electric series hybrid configuration and the chassis houses all the powertrain components in shaped boxes. A full drive by wire solution will be adopted and the only mechanical connections between the cabin and the platform will be a specific designed mechanical anchorages system to secure them together. The HOST hybrid vehicle system scheme is depicted in Fig. 1 while in Fig. 2 a photo of the first release of the vehicle assembly is presented. The on-board energy storage system is made up of an ultracapacitors tank (UC) and a battery unit (BU) which are interfaced with the traction electric drives by means of power electronics converters through the common dclink. The on-board energy generation unit (GU) includes a permanent magnet synchronous generator and a power electronic converter, to be connected to the dc-link. The electric generator is direct driven by an internal combustion engine (ICE). The GU, for emissions and fuel consumption issues, is designed to supply power in the range 4.5 kw-13.5 kw (25% - 75% of ICE rated power). Generation Unit is asked to provide power as a function of the BU state of charge (SOC), thus the GU operating points will be consequently adjusted to the different services. base module of storage unit is rated on the basis of the maximum values of both energy and power requirements which are common to all the services the HOST vehicle should accomplish. Fig. 2. HOST vehicle picture. The power electronic converters control boards are the physical interface between several system sections as the power converter hardware (IGBT modules), the main controller units (DSP evaluation boards) and the external world. Several PCBs (Printed Circuit Board) have been designed and realized on purpose with the main task of managing the power converters driving, the measures acquisition and conditioning and the communication interface. In Fig. 3 a simplified functional block scheme (the same for each power conversion unit: BU, UC and GU) of the HOST power converters PCBs (Printed Circuit Board) is shown and in Fig. 4 the relative power supply distribution network is illustrated. Fig. 1. HOST vehicle system overview. The concept of modularity is taken into account for designing the HOST vehicle storage unit; as a consequence a base module of storage unit is supposed to be located in the vehicle chassis, whereas additional storage units are selected to be added to meet the rating values of either energy or power requirement to satisfy the specifications for each vehicle single service. The Fig. 3. HOST power converters interface PCBs. The main power supply source for all the electronic circuitry is the battery unit (1V 375V); by means of a proper isolated dc/dc converter a 24V dc bus is furnished for delivering all the necessary supply voltages by way of other (isolated or not) dc/dc switching converters and/or linear regulators. This is an important concern to deal with because in previous EMC experimental activities we 919 Authorized licensed use limited to: BIBLIOTECA D'AREA SCIENTIFICO TECNOLOGICA ROMA 3. Downloaded on May 17,21 at 1:3:39 UTC from IEEE Xplore. Restrictions apply.
have already observed the main emission source comes mainly from the power supply circuitry of the electronic boards related to power converters and controllers. So, even in the present EMI investigation measurements we will focus our attention on this issue. Fig. 4. Power supply distribution network of the HOST power converters. III. EMI FILTER DESIGN Conducted EMI related to power supplies can be divided into two types of phenomena: CM and DM noises, each one with its own sources and propagation paths. CM noise occurs on all power supply lines with respect to the reference ground plane and it is essentially caused by insulation leakage, electromagnetic coupling and secondary effects due to parasitic components. DM noise component is always present between the two power supply lines and it is mainly caused by pulsating currents and devices turn-on and turn-off transients. In order to design a good EMI filter, DM and CM noise components have to be considered separately because they come from different sources and they solve different effects. Besides, it is possible that only one noise component is predominant in overall conducted emission, so that only the related filter section can be realized minimizing components count and size. A simple and practical procedure for designing EMI filters is implemented, as proposed in literature [3]. It is based on the knowledge of the EMI noise level in its CM and DM components, while it is unrelated to the noise source (EUT: Equipment Under Test) impedance. For all frequencies in the measuring range (15 khz - 3 MHz), conducted emission testing is performed; the filter attenuation requirement A REQ is obtained as the difference between the actual measurement V MIS and the limit stated by the reference standard V LIM, as in (1). into account. In fact, because of difficulties in predicting high frequency performance (it involves parasitic effects of filter components, radiated coupling and resonance effects) during filter design process, the main EMI filter design target is to meet low frequency specification. High frequency effects can be considered when the filter has been designed and realized, adjusting EMI filter performance for example working on filter layout (cables and component shielding, minimizing connections length, components positioning). In Fig. 5 the EMI filter topology to be implemented is shown. It is a typical configuration constituted by a CM section (L C and C Y ) and by a DM section (L D, C X1 and C X2 ). The EUT is the EMI noise source and the two 5 Ω impedances represent the two LISNs necessary for conducted emission measuring and through which the EUT is connected to the power supply. Even if CM and DM filter components affect each other, it is possible to study separately the two stages (obtaining a CM and a DM equivalent circuit), so simplifying the design step. Fig. 5. EMI filter topology. Through proper approximations about impedance conditions (normally met in many applications) it is possible to reduce both filter sections in a simple LC network (independent from the noise impedance) so that the related filter attenuation is given only by the voltages measured on the LISN with and without EMI filter. Filter corner frequency can be determined by drawing a 4 db/dec slope line (that represents the EMI filter attenuation) that is tangent to the first peak (low frequency specification) of the requested attenuation curve obtained in the previous step. The horizontal intercept of the line (as depicted in Fig. 6) settles the filter corner frequency. A REQ = V MIS - V LIM + A S (1) A S is a safe margin added to the computed attenuation. Equation (1) refers to both CM and DM noise components, achieved from a proper power splitter/combiner set interposed between the LISN (Line Impedance Stabilization Network) outputs and the EMI receiver input. For determining the filter corner frequency and then for dimensioning the filter elements, the lower frequency components of the disturbance noise are mainly taken Fig. 6. Attenuation curves. Analytically, the EMI filter (CM or DM) corner frequency f R can be computed by means of (2): 92 Authorized licensed use limited to: BIBLIOTECA D'AREA SCIENTIFICO TECNOLOGICA ROMA 3. Downloaded on May 17,21 at 1:3:39 UTC from IEEE Xplore. Restrictions apply.
R (A req ) db - I 4 Pk f = f 1 (2) where f Pk I is the frequency related to the first peak of the requested attenuation curve. Once defined the filter corner frequency it is possible to determine the components values by means of (3) for the CM filter section and (4) for the DM filter section: 1 f R,CM= 2π L C 2C (3) Y 1 f R,DM = (4) 2π 2L +L C ( ) D l X where L l is the leakage inductance of the common mode inductor L C. For the components selection (L C, L D, C X and C Y ) there is a degree of freedom; it is possible to set C or L based on available commercial values and then compute the other one. IV. EXPERIMENTAL RESULTS The main EMC tests to be performed, together with the relative test site and instrumentation, have been selected accordingly to the following reference standards: CEI EN 5511 (equivalent to the CISPR 11) Industrial, scientific and medical (ISM) RF equipment. Electromagnetic disturbance characteristics Limits and methods of measurement and CEI EN 5522 Limits and methods of measurement of radio disturbance characteristics of information technology equipment (ITE). In particular, conducted emission testing has been carried out on different HOST power conversion system subparts in the frequency range between 15 khz and 3 MHz. This experimental activity has been accomplished in the semi-anechoic chamber and in the shielded room both sites being located in the PED (Power Electronics & Drives) Laboratory of the University of Rome Roma Tre. The HOST EMC testing purpose is the evaluation of the EMI performance of the single HOST power converter; in particular, it is interesting to estimate which is the contribution to the system emission levels due to auxiliaries as the control platform and the electronic boards (measures acquisition, filtering, driving circuits and power supply distribution network). First of all we have analyzed conducted emissions related to the common 24V DC bus in several operating conditions. When only the interface PCBs are supplied, conducted emission levels are quite low (under the reference limits), while enabling the control system (DSP device) too we obtain the EMI spectrum shown in Fig. 7. As it can be seen, conducted emissions from the EUT made up of control and interface PCBs, overcome the stated limits; this occurs at high frequency (about 15 MHz) and is probably due to control and communication peripherals characterized by clock frequency of about tens of MHz. At low frequency instead, conducted emissions are very low. 8 3 Fig. 7. Conducted emission from control and interface PCBs. When even the power converter is supplied, emission spectrum changes as illustrated in Fig. 8. As it can be expected, higher emission levels are now at low frequency (about 3 khz) and they are mainly caused by the switching power converter hardware. The high frequency peaks are instead lower with respect to the previous case, so confirming that EMI noise phenomena are difficult to predict as EMI sources impedance changes and coupling paths are not very well known. 8 3 Fig. 8. Total conducted emission (from power converter, control and interface PCBs). In order to design a proper EMI filter, by means of a power splitter/combiner set, total conducted emission of Fig. 8 has been separated into its common mode (CM) and differential mode (DM) components, shown respectively in Fig. 9 and Fig. 1. 8 3 Fig. 9. CM component of total conducted emission. While DM noise is below the reference standard limits in all the measuring frequency range, CM component 921 Authorized licensed use limited to: BIBLIOTECA D'AREA SCIENTIFICO TECNOLOGICA ROMA 3. Downloaded on May 17,21 at 1:3:39 UTC from IEEE Xplore. Restrictions apply.
overcomes stated limits both at low frequency (3 khz) and high frequency (2 MHz and 15 MHz). 8 3 Fig. 1. DM component of total conducted emission. Concerning the described EMI filter design procedure, the commonly applied strategy is to measure CM and DM noises separately and then to design related filter sections independently to respond to the specific requirements. Finally the two filters are combined together to fulfil reference standards overall specifications. For the particular case here analyzed, in the first design step it is possible to consider only CM noise and related filter section. By means of a quasi-peak detector, exact values of frequency and amplitude for the first noise peak over passing the limits have been measured. Then, the requested attenuation and the filter corner frequency have been computed. For the components selection, a common mode inductor L C =,73 mh and ceramic capacitors C Y = 1 nf have been chosen. Conducted emission after the insertion of the CM filter is showed in Fig. 11. 8 3 Fig. 11. Conducted emission with the realized CM filter. As it can be seen, overall conducted emission after the EMI filter (CM section) insertion are below the stated limits but with a low safe margin. The predominant noise component is now the differential mode one; in order to further reduce conducted emission levels it is possible to implement the related DM filter section. In this case however, the leakage inductance of the common mode inductor of the realized CM filter can be used as differential mode inductor instead of a dedicated inductor. Considering that the leakage inductance is about 5% (about 37 µh) of the total inductance, DM filter can be implemented by inserting only the DM capacitors, computed according to (3). In this way a good EMI performance can be achieved with no considerable increase in EMI filter size (total size is mainly due to inductors). The overall conducted emission after connecting even the DM filter (made up of only C X capacitors) is shown in Fig. 12. 8 3 Fig. 12. Conducted emission with CM and DM filters. As it can be seen, conducted emission is now very low in all the measuring frequency range. V. CONCLUSIONS In this paper the EMI characterization of the power conversion system that manages the power flows between the generation and storage units of a hybrid vehicle has been presented. An extensive experimental test campaign has been carried out in a semi-anechoic chamber in order to evaluate the overall EMI performance of the single power converter and, in particular, to estimate which is the contribution to the system emission levels due to auxiliaries as the control platform and the related electronic boards (measures acquisition, filtering, driving circuits and power supply distribution network). Then, starting from the experimental results analysis, it is possible to identify the main EMI sources and coupling paths so that several adjustments can be performed to minimize conducted emissions With the purpose of reducing conducted emissions and designing a proper EMI filter, the overall noise has been separated into its CM and DM components. Experimental results underline the importance of considering CM and DM noises separately in order to minimize the filter components count and size and to improve EMI filter performance. In fact, when only one noise component (DM or CM) exceeds the stated limits, it can be suitable implementing only the corresponding filter section. In the described case, in order to further reduce emission levels, even if the EUT passed limit test with only a CM filter, DM capacitors have been added; EMI performance has enhanced with no filter size increase. REFERENCES [1] Serrao, V.; Lidozzi, A.; Solero, L.; Di Napoli, A.: EMI characterization and communication aspects for power electronics in hybrid vehicles. Proc. of EPE 7 Conference - Aalborg (Denmark), September 27. [2] Serrao, V.; Conti, L.; Di Napoli, A.; Solero, L.: Emission testing for the EMC performance evaluation of an electric 922 Authorized licensed use limited to: BIBLIOTECA D'AREA SCIENTIFICO TECNOLOGICA ROMA 3. Downloaded on May 17,21 at 1:3:39 UTC from IEEE Xplore. Restrictions apply.
wheelchair, Proc. of SPEEDAM 6 Symposium Taormina (Italy), May 26, pp. 597-62. [3] Cadirci, I.; Saka, B.; Eristiren, Y.: Practical EMI-filterdesign procedure for high-power high-frequency SMPS according to MIL-STD 461. IEE Proc. Electr. Power Appl., Vol. 152, No. 4, July 25, pp. 775-782. [4] Guo, T.; Chen, D. Y.; Lee F. C.: Separation of the common-mode and differential-mode conducted EMI noise. IEEE Transactions on Power Electronics, Vol. 11, No. 3, May 1996, pp. 48-488. [5] Shih, F. Y.; Chen D. Y.; Wu, Y. P.; Chen, Y. T.: A procedure for designing EMI filters for ac line applications. IEEE Transactions on Power Electronics, Vol. 11, No. 1, January 1996, pp. 17-181. [6] Ye, S.; Eberle, W.; Liu, Y. F.: A novel EMI filter design method for switching power supplies. IEEE Transactions on Power Electronics, Vol. 19, No. 6, November 24, pp. 1668-1678. [7] CEI EN 5511 (and relative Amendments A1, A2): Industrial, scientific and medical (ISM) radio-frequency equipment Radio disturbance characteristics Limits and methods of measurement, 2 nd ed, 23. [8] CEI EN 5522: Limits and methods of measurement of radio disturbance characteristics of information technology equipment (ITE). 923 Authorized licensed use limited to: BIBLIOTECA D'AREA SCIENTIFICO TECNOLOGICA ROMA 3. Downloaded on May 17,21 at 1:3:39 UTC from IEEE Xplore. Restrictions apply.