Electromagnetic Interference Generated from Fast Switching Power Electronic Devices

Transcription

1 Electromagnetic Interference Generated from Fast Switching Power Electronic Devices K. M. Muttaqi Integral Energy Power Quality and Reliability Centre School of Electrical, Computer and Telecommunication Engineering University of Wollongong New South Wales, Australia M. E. Haque Centre for Renewable Energy and Power Systems School of Engineering University of Tasmania Hobart, Australia Abstract This paper investigates the negative effects of electromagnetic interference (EMI) due to fast switching power devices (high dv/dt and di/dt) used in power electronic converters and industrial equipment. Mitigation techniques have been explored to reduce EMI noise effectively. Remedial measures to reduce the risk of equipment malfunction and health risk due to EMI have been explored. In this paper, EMI generation and propagation mechanism, high rates of change of voltage and current in fast switching power devices (such as IGBT), modelling and identifying EMI noise sources and coupling paths have been discussed. A practical EMI measurement system has been suggested to extract more information from EMI noise through analysis in frequency-domain and time-domain, and to test equipment emitting EMI to comply with electromagnetic compatibility (EMC) standards. Different filter topologies have been investigated for minimizing EMI noise and effect of high dv/dt and di/dt due to high frequency switching. Index Terms Electromagnetic Interference, Power Devices, Fast Switching, EMI Noise, Measurement System, and Power Converters. I. INTRODUCTION Power electronic converters are widely used in many applications including renewable energy generation, industrial equipment/motor drives, electric vehicle/train, air-craft, household appliances, electronic ballasts, computer power supplies, power supplies for telecommunication equipment, etc. These power converters use the fast switching power semiconductor switches, such as MOSFET (Metal-oxide semiconductor field effect transistor), IGBT (Insulated Gate Bipolar Transistor) as the preferred switching devices as they have many properties, such as higher efficiency, smaller size, and lower overall cost, low losses associated with switching device. However, fast switching speed of new converter/inverter technologies has the potential to cause EMI and high dv/dt [1-4]. All power electronics equipments generate and emit unwanted electrical signals (EMI noise) that can lead to performance degradation of other electrical/electronic equipments. They generate highfrequency conducted and radiated EMI noise and draw distorted line currents due to the sharp edges of the switching waveforms with high dv/dt. The undesirable EMI effects are interference with wireless systems (e.g., radio, TV, mobile, data transmission), malfunction of biomedical equipments (e.g., cardiac pace maker), misbehaviour of security doors of banks, ABS brake systems of cars, and electronic control systems in airplanes and the increase in power system losses associated with improper performance or failure of a range of industrial power equipments. It is necessary to find a mitigation technique to overcome this problem and to avoid costly equipment failures in industry. To International Journal of Innovations in Energy Systems and Power, Vol. 3, no. 1 (April 008) Page 19 of 45

2 secure an interference-free environment, proper electromagnetic compatibility (EMC) solutions are needed to ensure that devices, equipments, or systems that generate EMI noise can coexist satisfactorily (or being compatible). High-frequency switching operations in power electronic devices has improved the dynamic performance of ac motor drives, but created unexpected problems, such as motor bearing damage, high levels of conducted EMI, breakdown of winding insulation and motor leakage currents. It has been proven that its time derivative, dv/dt, as well as common mode voltage generated in ac drives are responsible for most of these problems [5]. In most of the previous work, passive EMI filters have been employed to reduce the effect of EMI noise and high dv/dt in power converter [6]. However, in designing passive filters the compensating bandwidth is comparatively narrow and only a certain part of noise can be eliminated. The size, weight, temperature, and reliability issues are significant design constraints. Active EMI filters provide alternative approaches to the problem [7]. Further investigation into noise sources and coupling path is desirable, as well as more accurate identification of noise propagation mechanism in circuits with considering all parasitic elements is required. In this paper, we have investigated the negative effect of EMI noise generated from power electronic switches. Modelling of EMI noise, coupling path and propagation mechanism has been discussed. An EMI measurement system in time domain is suggested along with the conventional frequency domain measurement system to extract more information on EMI noise. Analytical techniques for EMI characterisation is discussed. II. PROBLEM STATEMENT The introduction of international regulations on electromagnetic compatibility (EMC) has prompted active research in the study of electromagnetic interference (EMI) emission from power electronic converters, which are now indispensable components in modern household/industrial equipment, such as industrial motor drives, power electronic converters for renewable energy generation, electric vehicle, bio-medical equipment, computers, TV, washing machines, etc. For power electronic circuits, the high dv/ dt involved in the switching operation of high speed power electronic devices is widely believed to be a major source of EMI emission. The use of power electronic converters is increasing very rapidly in application to clean energy power generation systems such as solar/wind power generation and electric vehicles, which are friendlier to the environment. According to the Electric Power Research Institute (EPRI), about 50% 60% of the electric power is flowing through some kind of power electronics equipment, and eventually 100% is likely in the near future. This trend is already ongoing in Australia, because of the increasing demand for electric power to cater to the needs of the expanding industrial, commercial and residential sectors. Australia is now seeking to explore and support research works in the area of renewable energy generation. As a result of the increasing proportion of electronically processed power, the EMI would increase in the coming years, too, unless well-thought-out EMC standards and proper mitigation techniques are introduced and enforced at an early design stage. Failure to consider EMI/high dv/dt during early phases of the design process may result in expensive modifications (possibly with many additional components), printed circuit board (PCB) re-layout, product introduction delays, and EMC consultant fees to conform to the required standards. It is necessary to find remedial measures for the problems associated with EMI and high dv/dt as they can cause many undesirable effects in power electronic converters/industrial equipments. Due to strict EMC regulations, the EMI issue in power converters has recently become a topical area of research. III. MODELLING OF EMI SOURCE, COUPLING PATH, AND PROPAGATION MECHANISM The modelling of EMI noise sources and coupling paths in power electronic equipment is helpful to analyse the EMI mechanism and for designers to improve its EMC performance to satisfy national and international EMC standards. Designers may wish to characterise the EMI noise sources and identify the noise coupling paths through EMI simulations. However, modelling of parasitic elements has been a very difficult task as they are difficult to identify and also they may be physically inaccessible inside the module package. International Journal of Innovations in Energy Systems and Power, Vol. 3, no. 1 (April 008) Page 0 of 45

3 Many methods have been proposed in literature for parasitic modelling, such as three-dimensional finite element analysis, time-domain reflectometry, and partial-element equivalent circuit method. These methods are all purely mathematical and developed based on computation and computer simulation, and thus are very time-consuming, because the circuit models are very complicated. Another fundamental limitation of these methods is that expensive instruments and sophisticated simulation tools are mandatory. For better prediction of EMI behaviours without complicated circuit models and extensive calculation, equivalent lumped circuit models (shown in Fig.1) are proposed to characterise EMI noise from converter systems [8, 9]. Although some EMI phenomena have been described and useful analyses have been reported, the fundamental mechanisms by which the EMI noise are excited and coupled have not been adequately investigated. Fig. shows a typical EMI source and coupling paths model. In Fig., v m is the voltage on the power switch (IGBT/MOSFET) and i D is the diode current in a chopper, and V LISN1 and V LISN are voltages of two resistors in LISNs. Z S Z N V + S - v m i D EMI coupling paths V LISN1 V LISN LISN Converter Fig.1: Simplified EMI noise lumped circuit model Fig.: EMI sources and coupling paths model IV. AN EMI MEASUREMENT SYSTEM A practical, low cost EMI measurement system is suggested to capture EMI noise for frequencydomain and time-domain analysis, and to test an equipment emitting EMI to comply with Australian EMC Standards. The EMI measurement set-up typically requires a LISN (line impedance stabilisation network), noise separator, spectrum analyser, and computer as shown in Fig. 3. A LISN is required for capturing conducted EMI emission. The noise separator separates common mode (CM) and differential-mode (DM) noise components. The output of the noise separator is fed into a spectrum analyser and the corresponding frequency spectrum can be obtained. Then the data is fed into a computer for analysis and design of filters. Power Supply LISN Equipment Under Test Power Supply LISN Equipment Under Test Noise Separator Spectrum Analyzer Digital Storage Oscilloscope/EMI Analyzer Computer Fig. 3: Frequency domain EMI measurement system. Fig. 4: Time domain EMI measurement system. Traditionally, electromagnetic interference (EMI) measurement is performed with conventional analogue EMI receivers operated in frequency domain. Measurement in the frequency domain takes a long time, of typically 30 minutes for a frequency band from 30 MHz to 1 GHz [10]. EMI receivers use a pre-selector to obtain the required dynamic range of 36 db according to the standard by the International Special Committee on Radio Interference, CISPR 16-1 [11]. A time domain EMI measurement system (shown in Fig. 4) is suggested for measurement of EMI with a reduced number of accessories and cost, to make the system more reliable and simple. This type of measurement system International Journal of Innovations in Energy Systems and Power, Vol. 3, no. 1 (April 008) Page 1 of 45

4 can provide both magnitude and phase information. Also, a number of other statistical virtual measurement systems can be used to simulate the conventional detection system (e.g. peak, average, RMS and quasi peak detector through digital signal processing) [10]. It should be noted that our aim is not to replace the conventional frequency-domain EMI measurement system, but only to have a simple and efficient method of EMI measurement system that provides more information. Information obtained from both measurement systems can be used for accurate design of EMI filters and performance testing. The EMI noise emitted from the equipment is captured from the line (L) and neutral (N) outputs of LISN. The EMI presence on the line and neutral phases has the following relations [10]: VDM V L = VCM + (1) V N VDM = VCM () Where, V L is the positive line EMI voltage and V N is the negative line EMI voltage. The two signals given by Eqs.(1) and () will be fed into the digital storage oscilloscope (DSO). Using the inbuilt features of oscilloscope such as sampling, add and subtract, the CM and DM of EMI noise can be separated. The two channels of the DSO are added and subtracted in real-time and on-line to separate the CM and DM noise components (without using a noise separator as in frequency domain measurement system) as follows [10]: V CM VL + VN = (3) V DM VL VN = (4) Digital Storage Oscilloscope Data Storage Fourier Transform Time-domain signal reconstruction at each spectral point Detectors Peak Quasi Peak Average RMS Fig. 5: Data acquisition process in time-domain measuring system The data acquisition process for the time-domain measurement starts with the sampling process of the oscilloscope. Then the spectra via the Fourier transform (FT) are digitally computed. The errors due to the frequency characteristics of LISN, transmission line, amplifier, and anti-aliasing filter are corrected by signal processing. Next, the analysis of peak, rms, average, and quasi-peak values of the EMI signal can be performed as shown in Fig. 5. For the measurement of EMI noise current, a current probe with a very wide frequency band-width can be used. V. ANALYTICAL TECHNIQUES FOR EMI CHARACTERISATION AND IDENTIFICATION The high speed switching action (high dv/dt and di/dt) in a power converter emits both CM and DM of EMI noise. The purpose of analysis of EMI noise is to investigate the fundamental mechanism of the conducted EMI noise generation from power device switching. The mechanism of EMI noise has been analysed in [1] through simplified time-domain models to predict the switching noise across the LISN of the measurement system. However, it has been developed based on several assumptions (such as, ideal EMI noise source, ideal switching waveforms of power devices, etc.), which impair a great deal of accuracy of the model and make the model unsuitable to apply in practice. The switching transient in a power converter has traditionally been analysed by modelling it as a single slope dv/dt and di/dt transients. Neither the diode reverse-recovery current s effect nor the internal interconnect parasitic has been addressed. In reality, the switching transient of an IGBT has multiple slopes and shows complex switching behaviour. The frequency domain model is also used to quickly predict the EMI spectrum [13]. Since it is based on the assumptions used for the simplified time domain model, the inherent drawbacks are apparent. The IGBT turn-on switching introduces a major change in device current i c, di c /dt, which can be expressed as [14]: International Journal of Innovations in Energy Systems and Power, Vol. 3, no. 1 (April 008) Page of 45

5 di dt C g m ( V g + V th ) (τ ON ) = (5) R g C ies + g m L s Where, g m is the trans-conductance of the IGBT, V g is the IGBT gate voltage, and V th is the IGBT threshold voltage. The i c rise during time τ on causes v ce to fall down because of the stray inductance L s. The change in v ce can be given as: di C Δ v ce = L s ( τ on ) (6) dt The change in device voltage, dv ce /dt during τ on can be written as: dv dt ce ( τ on ) ( di C / dt ) τ on = L (7) s τ Where τ is the time required by the collector current (i C ) to change from peak value to steady state value as shown in figure 6. i C I C, peak I C τ on τ Figure 6. Turn-on waveform of IGBT with inductive load. The di c /dt during the current rise has a direct impact on the reverse-recovery current (I rr ) of the freewheeling diode. The relation between di c /dt and I rr is given by [14]: I rr di dt ( τ ) c = τ LT I (8) L on Where, τ LT is the minority carrier lifetime of diode. It has been revealed that large reverse-recovery current increases the EMI level. A larger turn on di c /dt leads to a higher dv ce /dt. High dv/dt and di/dt during switching of power devices is related to switching frequency and conducted EMI level. VI. MITIGATION OF EMI NOISE GENERATED FROM POWER CONVERTERS Filters can be designed and used to reduce EMI emission from power converters. Fig. 7 shows the simulation model for the investigation of EMI on a pulse width modulated (PWM) IGBT-Inverter fed ac motor drive. Fig. 8(a) shows a typical PWM inverter output voltage. Fig. 8(b) shows over voltage at the motor end due to EMI and high dv/dt. It is observed in Figs. 9(a) and 9(b) that the over voltage is reduced significantly using RC and LC passive filters, respectively. International Journal of Innovations in Energy Systems and Power, Vol. 3, no. 1 (April 008) Page 3 of 45

6 Fig.7: EMI investigation on a pulse width modulated IGBT-Inverter fed ac motor drive. Traditionally, passive filters are employed to attenuate EMI emitted from power devices [6]. However, only a certain part of noise can be eliminated using passive filters as the compensating bandwidth is comparatively narrow. An alternative to the passive filters is the use of an active filter, in which an active electronic circuit is used to cancel or suppress ripple components at the filter output [7]. 600 Voltage (V) Voltage (V) Motor terminal voltage DC bus voltage Time (Sec) Time (µs) (a) Typical PWM Inverter output (b) Over voltage at motor terminal due to EMI and high dv/dt without filter Fig.8: Over voltage at the motor terminal due to EMI Voltage (V) Voltage (V) Time (Sec) Time (Sec) (a) Motor terminal voltage with RC filter (b) Motor terminal voltage with LC filter Fig. 9: EMI mitigation using filters VII. CONCLUSIONS In this paper, the impact of EMI noise generated from power electronic switches due to fast switching (high dv/dt and di/dt) has been investigated. EMI noise modelling, coupling path and propagation mechanism issues are also discussed. A time domain measurement technique is suggested together with the conventional frequency domain measurement system to collect more information for EMI noise characterisation. The suggested EMI measurement system will be able to capture EMI noise for frequency-domain and time-domain analysis, and to test equipment emitting EMI to comply with electromagnetic compatibility (EMC) standards. Mitigation techniques for the effect of EMI have been discussed. From simulation results, it is revealed that the effect of EMI can be reduced significantly with the help of RC or LC filters. International Journal of Innovations in Energy Systems and Power, Vol. 3, no. 1 (April 008) Page 4 of 45

7 REFERENCES [1] G. L. Skibinski, R. J. Kerkman and D. Schlegel, EMI emissions of modern PWM AC drives, In IEEE Industry Applications Magazine, vol. 5, Issue: 6, pp , [] M. E. Haque, A. A. Bokhari and A. I. Alolah Simulink modeling of the problem associated with fast switching PWM IGBT-inverter fed AC motor drive with long cable and its remedies IEEE Intl. conference on Systems, Signals & Devices, Sousse-Tunisia, March 1-4, 005. [3] M. E. Haque, M. F. Rahman. and T. R. Blackburn, A study of the over-voltage stress with IGBT inverter waveforms on motor and supply cabling and their remedial measures, Proc. of AUPEC/EECON Conference, Darwin, Australia, pp. 87-9, Sept. 6-9,1999. [4] T. Haider, M.E. Haque, M. F. Rahman, T.R. Blackburn and C. Grantham, Modeling and experimental studies of effect of steep fronted inverter waveform on motor and supply cabling and their remedies, IEEE International Conference on Power Electronics and Drives (PEDS'99), Hong Kong vol., pp , July 7-9, [5] L-H. Kim, N-K. Hahm, W-C. Lee, J-S. Yu, Y-C. Kim, C-Y. Won, Y-R. Kim, Analysis of a new PWM method for conducted EMI reduction in a field oriented controlled induction motor, in Proc. of IEEE Applied Power Electronics Conference and Exposition, 006 (APEC '06), 19-3 March 006 pp [6] C. Rengang, Integration of EMI filter for distributed power system (DPS) front-end converter, in Proc. IEEE Power Electronics Specialist Conf. 03, Jun , 003, pp [7] W. Chen, X. Yang and Z. Wang, An active EMI filtering technique for improving passive filter low-frequency performance, IEEE Trans. On Electromagnetic Compatibility, vol. 48, no.1, pp , Feb [8] D. Gonzalez, J. Gago, and J. Balcells, Analysis and simulation of conducted EMI generated by switched power converters: Application to a voltage source inverter, IEEE Trans. Ind. Electron., vol. 50, no. 6, pp , Dec [9] J. Meng, W. Ma, Q. Pan, Z. Zhao, and L. Zhang, Noise source lumped circuit modeling and identification for power converters, IEEE Trans. Industrial Electron., vol. 53, no.6, pp , Dec [10] F. Krug, and P. Russer, Quasi-peak detector model for a time-domain measurement system, IEEE Transactions On Electromagnetic Compatibility, vol. 47, no., May 005. [11] AS/NZS CISPR Standards (16.1.1:006), Specification for radio disturbance and immunity measuring apparatus and methods - radio disturbance and immunity measuring apparatus - measuring apparatus. [1] L. Ran and S. Gokani et al., Conducted electromagnetic emissions in induction motor drive systems Part I: Time domain analysis and identification of dominant modes, IEEE Trans. Power Electron., vol. 13, no. 4, pp , Jul [13] Q. Liu, F. Wang, and D. Boroyevich, Model conducted EMI emission of switching modules for converter system EMI characterization and prediction, in Proc. IEEE IASAnnu. Meeting Conf. Rec., Oct. 004, pp [14] M. Jin, M. Weiming, Power converter EMI analysis including IGBT nonlinear switching transient model, IEEE Trans. Industrial Electron., vol. 53, no.5, pp , Oct BIOGRAPHY Dr. Kashem M. Muttaqi completed B.Sc.Eng. from Bangladesh University of engineering and Technology in He received M.Eng.Sc. degree from the University of Malaya in 1997 and Ph.D. degree from Multimedia University, Malaysia, in 001. Currently, he is working as an Associate Professor at the School of Electrical, Computer and Telecommunication Engineering, University of Wollongong, Australia. He was the Deputy-Director for the Centre of Renewable energy and Power Systems and worked as a Senior Lecturer at the School of Engineering, University of Tasmania, Australia. He was associated with the Queensland University of Technology, Australia as a Postdoctoral Research Fellow from 000 to 00. Previously, he also worked for Multimedia University as a Lecturer for three years. His special fields of interests include distributed generation, renewable energy, distribution system automation, power system planning, and artificial intelligence. He is a Senior Member of IEEE. International Journal of Innovations in Energy Systems and Power, Vol. 3, no. 1 (April 008) Page 5 of 45

8 Dr. Md. Enamul Haque graduated in electrical and electronic engineering from Bangladesh Institute of Technology (BIT), Rajshahi, Bangladesh, in He received M.Eng.Sc. in electrical engineering from University Technology Malaysia in 1998, and Ph.D. in electrical engineering from The University of New South Wales, Sydney, Australia, in 00. He has worked as an Assistant Professor for King Saud University, KSA, and United Arab Emirates University, UAE for four years. Dr. Haque is currently working as a Research Fellow in the Centre for Renewable Power and Energy Systems, University of Tasmania, Australia. His research interests include power electronics and DSP based electric drives, Wind/Solar remote area power supply systems, EMC/EMI issues in power electronics, industrial equipments and Renewable energy applications. International Journal of Innovations in Energy Systems and Power, Vol. 3, no. 1 (April 008) Page 6 of 45