Multi-Channel Time-Domain EMI Evaluation of
Dominant Mode Interference for Optimized Filter
Design in Three-Phase Systems
Daria Nemashkalo
∗, Niek Moonen
∗, Frank Leferink
∗†∗University of Twente, Enschede, the Netherlands
†THALES Nederland B.V., Hengelo, the Netherlands
dasha.nemashkalo@utwente.nl
Abstract—This paper describes optimal three-phase filter de-sign based on the measured dominant mode of interference. The modes of interference are determined by simultaneously measuring currents and voltages in all phases in time domain, using a multi-channel digitizer. The measured results are eval-uated in frequency domain after fast Fourier transform. While usual electromagnetic compatibility measurement equipment is evaluating only one single channel, the described multi-channel technique allows rapid estimation of dominant modes, for sta-tionary loads, but also for cyclo-stasta-tionary, transient and non-linear loads. Using this method, the topology of power line filter for sufficient suppression of a specific mode of interference from commercial of the shelf three-phase power converter is estimated.
I. INTRODUCTION
One of the major sources of electromagnetic interference (EMI) are switched mode power supply (SMPS). Often, com-mercial of the shelf (COTS) power supplies are used in many industries, including power defense electronics. These COTS power supplies are specified and tested according to the basic civil electromagnetic compatibility (EMC) standards. which start at only 150 kHz. Although, the interest for the 10 kHz-150 kHz range is growing [1], most COTS equipment fail the requirements below 150 kHz [2], [3]. COTS SMPS with interference below 150 kHz which had to comply with the military standard for conducted emission were investigated in this paper. To comply with standards for professional applications, an external power line filter (PLF) should be implemented. The design and performance of a PLF strongly depend on the terminating impedances on both sides of the filter. Ignoring these parameters on the design stage can lead to oversized and costly filters. Moreover, under certain conditions filters with a poor design may even amplify the noise [4].
Components that define the common mode (CM) noise source impedance are the unintentional capacitance between the switching device and the heat sink, parasitic capacitance between the heat sink and the grounded chassis, and parasitic capacitance between other parts of the setup, which carry pulsating current, and the grounded chassis; components that define the differential mode (DM) noise source impedance is the turned-on resistance of the rectifying diodes, the equivalent series resistance and the equivalent series inductance of the bulk capacitor [5]. At lower frequencies (below 1 MHz), inter-ference is mainly DM while at higher frequencies it is mainly
CM [6] because parasitic effects create a low-impedance pass for interference. Ideally, each mode of noise is suppressed by the respective section of the filter [7]. A typical generic filter topology of a power supply filter is composed of a common mode choke (CMC), a line inductor (DM choke), X-capacitors (e.g. capacitors connected between lines for DM attenuation) and Y-capacitors (e.g. capacitors between lines and ground for CM attenuation) [8]. Despite the ubiquitous use of these components, there are some drawbacks that designers and users of the filters should be aware of. DM chokes or line inductors are usually bulky, so for systems that require
compact dimensions, those are not suitable. Furthermore, Cx
capacitors have a limited lifetime because of the reactive current. There are several typical topologies of PLF based on basic filtering circuits. With increasing filter order, the inser-tion loss (IL) increases with 20 dB/dec. This ideal IL rate can only be achieved if the terminated impedances on both ends of the filter are appropriate [9]. To give sufficient attenuation, an inductor that faces the input of the SMPS should have a much higher impedance than the noise source impedance (Fig. 1,a). If the capacitor faces the input of SMPS (Fig. 1,b), a capacitor’s impedance should be much lower than the noise source impedance [4]. COTS SMPS are designed to fulfil EMI requirements from 150 kHz, while many system integrators are faced with EMI requirements starting at 10 kHz. In short, this means more than 10 times bigger inductors and/or capacitors to achieve the required IL, which is a major challenge (and fight with mechanical engineers for the required volume). Thus, knowing the noise source and the load impedances, we can design a perfect filter for each particular case. Various
Fig. 1: Illustration of relation between filter component and noise source impedance: a) inductor faces the input of the noise source, b) capacitor faces the input of the noise source. approaches for in-situ impedance measurements were
devel-oped previously [5], [9]–[12]. However, all of the proposed approaches are time-consuming, due to the pre-measurement calibration process and require expensive equipment which is not always available. Furthermore, those approaches are not suitable for three-phase systems. For these reasons a trial and error method for designing and implementing a PLF is still the most used approach in the industry. As a result, overdesigned and expensive filters are implemented in systems and this becomes annoying when space and costs are an issue. These are the main reasons for developing a simplified approach for designing and implementing suitable filters for three-phase systems. It should be noted that the conventional method of measuring the interference in the industry uses a one or two-channel EMI receiver which requires a significant amount of time per test. Also, usually only the voltage is considered.
This paper proposes a method based on multi-channel time-domain measurements of the current, which allows us to determine the dominant mode of emission and find a suitable configuration of the filter. The aim of the paper is to create an approach convenient for industrial applications.
II. THEORETICAL APPROACH FOR A 3-PHASE SYSTEM
A. Proposed method
In order to optimize the filter design without measuring the impedances of the system, analysis based on the domi-nant mode of interference can be made. Whereas, the total interference is the sum of DM and CM interference, the most efficient way to reduce the total interference is to apply a filter for suppressing the dominant mode of interference. To determine the dominant mode of interference without using a noise separator the current measurement should be performed. DM and CM current for a single line can be obtained according to Eq. 1 and Eq. 2 by measuring the line currents in three-phase systems [13]. iCM = iu+ iv+ iw 3 = ignd 3 (1) iDM = iu− iCM (2)
Then, determining the dominant mode of emission is pos-sible by comparing DM and CM currents. Thus, the topol-ogy of the filter for the efficient suppression of interference can be chosen with the flow-chart shown on Fig. 2. Use of the multi-channel oscilloscope allows to measure each phase current (e.g. normal mode), all phase currents together (e.g. common mode) and line voltages simultaneously. That reduces measurement and post-processing time. The block diagram of the measurement setup representing the proposed approach can be seen in Fig. 3. Setup with three separate line impedance stabilisation network (LISN) is used to measure all line voltages simultaneously. With current clamps measuring current flowing in three separate lines simultaneously the CM current is obtained. By measuring current flowing in all lines together we obtain the normal mode (NM) current, that is used to calculate DM current. Then the analysis in terms of the dominant mode of the interference can be made.
Fig. 2: Approach flow chart.
Fig. 3: Block diagram of the measurement setup.
III. TEST CASE
The voltage measured with a conventional EMI receiver from the output of the LISN of a three-phase 6 kW
380/480 VAC to 52/48 VDC converter is shown in Fig. 4.
Even with an internal input filter, this converter does not comply with AECTP 500 NCE02-1 [14]. Hence, an additional PLF is required.
Fig. 4: Conducted emission from SMPS.
The schematic of the filter which was used initially can be seen in Fig. 6. Fig. 5 shows the IL for Common, Differential and Normal Modes provided in the datasheet from which we can see that the filter has an attenuation over 100 dB in a broad frequency range (approx. 200 kHz – 1 MHz), while the significant emission from the SMPS is present only at 100 kHz [15]. Furthermore, the information given in the datasheet is
relevant for a situation where the impedance is matched, i.e. when the noise source impedance and network impedance are equal to 50 Ω. As predicted, in the real system, when the noise source and the load impedances are not known and not equal, the filter shows significantly lower attenuation (Fig. 7). All measurements hereafter are performed using the time-domain measurements, and converted to frequency time-domain using fast Fourier transform (FFT). Also, it should be noted that in this set up the effective attenuation is only provided at the frequency range of 10 kHz – 200 kHz, since there is no interference to suppress at higher frequencies.
104 105 106 107 108 frequency, Hz 0 20 40 60 80 100 IL,dB Normal Mode Common Mode Differential Mode
Fig. 5: IL given in the datasheet.
Fig. 6: Schematic of the filter.
103 104 105 106 Frequency, Hz 20 40 60 80 100 120 Voltage, dBuV Filter inserted Nothing inserted X 100500 Y 111.1 X 100500 Y 68.4
Fig. 7: Voltage with PLF inserted.
IV. MEASUREMENTSETUP
To design a suitable filter for this particular case, measure-ments of the CM and NM currents were performed using the Picoscope TA189 current clamps and calculated according to Eq. 1 and Eq. 2. The measurement setup is shown in Fig. 8. Three one-phase LISN are used to provide a reliable
impedance and to allow simultaneous measurements of the three line voltages. The 8-channel Picoscope is used instead of an EMI receiver, which allows us to observe the three line voltages and CM and DM currents simultaneously. This significantly reduces the measurement time needed.
Fig. 8: Picture of the measurement setup.
V. RESULTS
Comparing the current spectrums shown in Fig. 9 the con-clusion is made that the DM is the dominant mode of emission of the device under test (DUT). This is, as expected for a low frequency, due to the nature of the DM noise propagation, as discussed in the introduction. Further measurements were
performed with inserted Cx capacitors or line inductor or a
combination of those (Fig. 10). First, a 1.5 mH inductor is inserted and then a 40 µF capacitor, which is a first order filter. 103 104 105 106 Frequency, Hz 20 30 40 50 60 70 80 90 100 Voltage, dBuV
Differential mode current Common mode current
Fig. 9: Measured current spectrums.
The inductor provided 5 dB higher attenuation than the capacitor on the frequency of interest, which allows us to make an assumption that the noise source impedance can be considered as ‘low’. To confirm this, a second-order topology filter was created using the previously selected components. Two configurations of filter circuits consisted of the same components were tested: LC and CL. The results can be seen in Fig. 11 and Fig. 12. The attenuation in total is improved when compared to inserting only one of these components, but a noticeable difference can be seen in performance by changing the LC configuration into CL. It shows that CL configuration gives a better attenuation than an LC configura-tion. As can be seen the CL configuration of the second-order filter topology is outperforming the initially used COTS PLF. According to [9] this confirms that the noise source impedance
can be considered as low and for the effective attenuation the inductor should face the input of the SMPS.
103 104 105 106 Frequency, Hz 0 20 40 60 80 100 120 Voltage, dBuV nothing inserted C inserted L inserted X 100500 Y 97.27 X 100500 Y 94.47 X 100500 Y 110.3
Fig. 10: Voltage with first order filter topology.
103 104 105 106 Frequency, Hz 0 20 40 60 80 100 120 Voltage, dBuV nothing inserted LC inserted X 100500 Y 110.3 X 100500 Y 79.72
Fig. 11: Voltage with second-order filter topology (LC com-bination). 103 104 105 106 Frequency, Hz 0 20 40 60 80 100 120 Voltage, dBuV nothing inserted CL inserted X 100500 Y 110.3 X 100500 Y 54.31
Fig. 12: Voltage with second-order filter topology (CL com-bination).
VI. CONCLUSIONS
A simple and time-efficient approach for filter design in three-phase systems is proposed using a low-cost multi-channel digital oscilloscope which allows performing more efficient testing compared to a conventional EMI receiver. By measuring the CM and DM currents, the dominant mode of interference of DUT can be assessed. Following the proposed approach this will result in the selection of most useful suppression components, much better than the conventional
method of applying COTS filters that often employ several useless but costly and bulky components. Based on the noise source impedance a suitable topology of the PLF can be determined, however as was shown in this paper the impedance can be estimated by inserting different component types with several values. The total interference from SMPS is suppressed sufficiently by implementing a second-order filter with a spe-cific orientation, as only in CL configuration the suppression was above the required 45 dB.
ACKNOWLEDGMENT
This research has received funding from the European Union’s SCENT (Smart City EMC Network for Training) project which are funded by the European Union’s Horizon 2020 research and innovation programme under the Marie Sklodowska-Curie grant agreement No. 812391.
REFERENCES
[1] F. Leferink, “Conducted interference, challenges and interference cases,” IEEE Electromagnetic Compatibility Magazine, vol. 4, no. 1, pp. 78–85, 2015.
[2] CLC/TR 50669, “Investigation Results on Electromagnetic Interference in the Frequency Range below 150 kHz,” 2017.
[3] CLC/TR 50627, “Study Report on Electromagnetic Interference between Electrical Equipment/Systems in the Frequency Range Below 150 kHz,” 2015.
[4] S. Ye, W. Eberle, and Y. F. Liu, “A novel EMI filter design method for switching power supplies,” IEEE Transactions on Power Electronics, vol. 19, no. 6, pp. 1668–1678, 2004.
[5] K. See and L. Yang, “Measurement of noise source impedance of SMPS using two current probes,” Electronics Letters, vol. 36, no. 21, p. 1774, 2002.
[6] C. Paul, Introduction to Electromagnetic Compatibility, 2009, vol. 38, no. 7-8.
[7] F.-y. Shih, D. Y. Chen, S. Member, Y.-p. Wu, and Y.-t. Chen, “A Procedure for Designing EMI Filters for AC Line Applications - Power Electronics, IEEE Transactions on,” vol. 11, no. 1, 1996.
[8] V. Tarateeraseth, B. Hu, K. Y. See, and F. G. Canavero, “Accurate extraction of noise source impedance of an SMPS under operating conditions,” IEEE Transactions on Power Electronics, vol. 25, no. 1, pp. 111–117, 2010.
[9] V. Tarateeraseth, “EMI filter design: Part III: Selection of filter topology for optimal performance,” IEEE Electromagnetic Compatibility Maga-zine, vol. 1, no. 2, pp. 60–73, 2012.
[10] J. Tan, D. Zhao, and B. Ferreira, “A method for in-situ measurement of grid impedance and load impedance at 2 k-150 kHz,” 9th International Conference on Power Electronics - ECCE Asia: ”Green World with Power Electronics”, ICPE 2015-ECCE Asia, pp. 443–448, 2015. [11] Z. Zhao, K. Y. See, E. K. Chua, A. S. Narayanan, W. Chen, and
A. Weerasinghe, “Time-Variant In-Circuit Impedance Monitoring Based on the Inductive Coupling Method,” IEEE Transactions on Instrumen-tation and Measurement, vol. 68, no. 1, pp. 169–176, 2019.
[12] Kang Rong Li, Kye Yak See, and Xing Ming Li, “Inductive Coupled In-Circuit Impedance Monitoring of Electrical System Using Two-Port ABCD Network Approach,” IEEE Transactions on Instrumentation and Measurement, vol. 64, no. 9, pp. 2489–2495, 2015.
[13] B. Wunsch, U. Drofenik, S. Skibin, and V. Forsstrom, “Impact of diode-rectifier on EMC-noise propagation and filter design in AC-fed motor drives,” IEEE International Symposium on Electromagnetic Compatibility, pp. 237–242, 2017.
[14] NATO International Staff - Defence Investment Division Allied Environ-mental Conditions and Tests Publication AECTP500 (edition 4), “Elec-tromagnetic Environmental Effects Test And Verification-Equipment and sub-system tests,” 2011.
