Improvement in low frequency emission test method by live impedance measurement

Improvement in Low Frequency Emission Test Method by Live Impedance Measurement Soydan Çakır Electromagnetic Laboratory TUBITAK UME Kocaeli, Türkiye 0000-0001-8210-2171 Osman Şen Electromagnetic Laboratory TUBITAK UME Kocaeli, Türkiye 0000-0002-5716-3316 Serdar Büyük Electromagnetic Laboratory TUBITAK UME Kocaeli, Türkiye 0000-0002-8773-3519 Marco A. Azpúrua EMC Electromagnetic BCN S.L. EMC Barcelona Barcelona, Spain 0000-0001-8078-5116 Engin Özdemir Faculty of Technology Kocaeli University (KOU) Kocaeli, Türkiye 0000-0003-0882-332X Abstract— The MIL-STD-461 CE101 test is one of the essential low frequency emission tests in the range of 30 Hz – 10 kHz for military and aerospace equipment, however, its application remains challenging. The primary concern is the effect of the power source or grid impedance on test results because the LISNs are not functional in such a low frequency range. Therefore, the emission levels depend on the source impedance and, consequently, the reproducibility of test results is generally poor. In this paper, we thoroughly analyse this problem and propose a solution based on the live impedance measurement of the CE101 test circuit. For the live impedance measurement, we used the CS101 military low frequency immunity test system along with the addition of FFT and current measurement capabilities. Keywords—aerospace, CE101, EMC, emission, low frequency, military, live impedance measurement I. INTRODUCTION The CE101 test defined in the standard MIL-STD461G [1] is applied to military and aerospace equipment. This test is performed to measure low frequency emissions from supply ports of the Equipment Under Test (EUT) in the 30 Hz - 10 kHz range. First, the CE101 test requires a calibration process performed on a known resistive load to verify the test bench. During this calibration, a simple circuit shown in Fig. 1 (a) is set to produce a known current through the reference resistor at the frequencies stated by the standard. Thereafter, the known current on the circuit is measured by the CE101 test system and it is checked whether the system yields the expected results. If the deviation between the known and measured current is greater than ± 3 dB, the system is out of tolerance and cannot be used. In the test stage, current emissions from the EUT are measured with a current probe and a frequency-selective receiver, as shown in Fig. 1 (b). Although the EUT is supplied through LISNs during the test, LISNs are ineffective in the low frequency range. Therefore, test results become source impedance dependent, which has implications in the measurement's reproducibility and repeatability. In the scope of this paper, the source impedance is comprised of the combination of LISNs, the line filters of the chamber and grid/power source impedance values. A study that investigates the influence of the source impedance on CE101 test results was performed in [2]. It was observed that source impedance was modified by using different types of LISNs and chamber line filters. The resultant discrepancies in the test results were thoroughly demonstrated. A similar issue and its relevant solution were investigated in the upper frequency range (150 kHz - 30 MHz) through impedance measurements via a Vector Network Analyser (VNA) in [3-4]. However, to the best of our knowledge, there is no reported work to overcome the aforementioned problem in the frequency range from 30 Hz to 10 kHz. (a) (b) Fig. 1. CE101, (a) calibration setup, (b) test setup. In this paper, to analyse the CE101 test and alleviate the uncertainty and the discrepancies arising from the test circuit impedance, we firstly measured the live loop impedance of the CE101 test circuit, which includes the EUT, LISNs, chamber filters, power supply and cables, in different source conditions. For the live impedance measurement, we utilised our CS101 test system whose calibration and test setups are depicted in Fig. 2. The CS101 test is a low frequency immunity test which is considered the counterpart of the CE101. During the CS101 test low frequency sinusoidal voltage ripples are injected to power ports of the EUT in the range from 30 Hz to 150 kHz, as investigated in detail in [510]. Here, we improved our CS101 test system with a complementary current measurement feature in addition to the intrinsic voltage measurement. As a result, the CS101 test system was adapted for impedance measurements. In addition, we integrated a post-processing Fast Fourier Transform (FFT) - based time domain solution to separate injected ripples from the AC power frequency of the EUT and to facilitate accurate live impedance measurements under adverse grid or EUT supply voltage conditions. The integration of the FFT analysis into CS101 testing and the details of FFT analysis can be found in detail in [9-13]. Alternatively, integrated FFT features of oscilloscopes can be used instead of the post-processing for expediting the process if it is available on the oscilloscope and a proper FFTactivated instrument driver exists on the used CS101 test software. Another option for live impedance measurements of low-current devices may be the VNA method employed in [34]. Still, the CS101-based method proposes more robust measurements in the low frequency range for high current or high voltage devices. In this regard, we must consider that the VNA is a sensitive and expensive instrument which generally starts efficiently operating from 10 kHz, consequently it is not optimal for the intended application. (a) (b) Fig. 2. CS101, (a) calibration setup, (b) test setup. Finally, we calculated the test circuit loop impedance deviations, by using the setup and equation given in Fig. 4 and in (5) respectively, considered as correction factors between different low frequency test environments and compared them with loop current deviations to verify the proposed method. II. METHOD As any EUT can be modelled as a combination of a constant voltage source and an internal impedance (Thevenin equivalent), we used this model to analyse the CE101 test and alleviate the uncertainty arising from the test circuit impedance. As common mode (CM) currents are not expected in this low frequency range, we limited the analysis only to the differential mode (DM), and the relevant circuit diagram is given in Fig. 3 for the differential mode. Fig. 3. DM circuit model for low frequency conducted emission. In this context, The CE101 test system, along with the live impedance measurement setup utilising the FFT-enabled CS101 test system, is depicted in Fig. 4. The first channel of the oscilloscope with 1 MΩ input impedance is dedicated to voltage measurements whereas the second channel of the oscilloscope with 50 Ω input impedance is assigned to current measurements. For current measurements, we used the 50 Ω input of the oscilloscope because the current probe factors are generally calibrated to be used along with 50 Ω matched instruments, such as a measuring receiver. Alternatively, current measurements may be accomplished with a separate frequency-selective instrument. In this measurement system, the total loop impedance of the CE101 test circuit, which involves the EUT, source and cable impedance values together (see Fig. 4), is targeted instead of separate EUT or source impedance values. The loop impedance is detected by the ratio of the voltage measured across the coupling transformer to the measured loop current as presented in (5) to be used in the calculation of the correction factor (K) as given in (1), which is expected to establish a link between different test environments. Here, ZCE101_reference_loop is the impedance of the test circuit selected as reference, and ZCE101_test_loop is the loop impedance measured in the targeted CE101 test setup under investigation. ZCE101_reference_loop and ZCE101_test_loop include the impedance of the EUT, power source, and other components, e.g., LISNs, filters, and cables. The correction factor is also known as impedance deviation. 𝐾 = 𝑍𝐶𝐸101_𝑟𝑒𝑓𝑒𝑟𝑒𝑛𝑐𝑒_𝑙𝑜𝑜𝑝 𝑍𝐶𝐸101_𝑡𝑒𝑠𝑡_𝑙𝑜𝑜𝑝 (1) Then, the emission level (ICE101_reference_loop) in the reference circuit can be linked to the emission level (ICE101_test_loop) in the targeted circuit by using the measured impedance deviation as given by (2), 𝐼𝐶𝐸101_𝑡𝑒𝑠𝑡_𝑙𝑜𝑜𝑝 = 𝐼𝐶𝐸101_𝑟𝑒𝑓𝑒𝑟𝑒𝑛𝑐𝑒_𝑙𝑜𝑜𝑝 × 𝐾 (2) Alternatively, emission results obtained in different test environments may be scaled to a predefined load such as 50 Ω by using the measured loop impedance values as given in (3) and (4). This can be convenient to harmonize emission results obtained from different test environments and make them comparable between each other. 𝐼𝑠𝑐𝑎𝑙𝑒𝑑_𝑐𝑢𝑟𝑟𝑒𝑛𝑡 = 𝐼𝐶𝐸101_𝑡𝑒𝑠𝑡_𝑙𝑜𝑜𝑝 × 𝑆 (3) In this case, the scaling factor 𝑆 can be defined as follows; 𝑆 = 𝑍𝐶𝐸101_𝑡𝑒𝑠𝑡_𝑙𝑜𝑜𝑝 50 Ω (4) Fig. 4. CE 101 test system along with FFT and current measurement enabled CS101 system used for impedance measurement. 𝑍𝑙𝑜𝑜𝑝 =𝑉 𝐼 (5) The current probe in the setup shown in Fig. 4 has two functions. One of them is to detect the current required for the live impedance measurement and the other is to perform current measurements required for CE101 testing. After the loop impedance measurement, the coupling transformer is removed from the circuit for CE101 testing. This is made because the measured loop impedance only includes the CE101 test circuit not the coupling transformer. The removal of the coupling transformer is simply attained by short circuiting the output of the coupling transformer just before the CE101 test after the loop impedance measurement. After the loop impedance measurement, the succeeding step is calculating correction factors. The verification of correction factors is performed by means of measuring loop currents, considered as CE101 emission results, in the same test environment and comparing loop current deviations with impedance deviations. The good consistency between the loop impedance and current deviations is expected to verify the proposed method. We also developed a piece of software by using LabWindows/CVI in order to perform impedance measurements based on the CS101 test system and carry out FFT-based time domain processing to measure voltage and current values under the adverse AC EUT power frequency. The same software is also able to perform CE101 testing after the impedance measurement. III. EXPERIMENTAL STUDY Initially, we verified the impedance measurement method by using lumped elements with known impedance as test subjects, specifically, 0.5 Ω, 50 Ω, 80 µF, 430 µH. Next, we employed three types of EUT to verify that the proposed CE101 measurement method improved with the impedance measurement. One of them was a signal generator RF output to verify the method in nearly ideal conditions. For the simulation of source impedance diversity, we employed an assortment of impedance values (50 Ω, 80 µF, 430 µH) behind the LISNs, and they were directly connected to the power source side of the LISNs (see Fig. 5). The other EUT was a homemade harmonic reference device which is designed as a square wave generator. As the harmonic reference device was a standalone device and it does not require direct supply voltage coming from LISNs to operate, a variety of resistors (50 Ω as the reference, 26.6 Ω, 13.3 Ω and 6.6 Ω) were directly connected to the output of the device as seen in Fig. 6 to create different dummy source impedance values with respect to 50 Ω. After these preliminary measurements with the predictable devices, finally, an uncontrolled device, an Uninterruptible Power Supply (UPS) as the EUT, was tested in an actual MIL-STD461 CE101 test setup (see Fig. 7). Nevertheless, we did not follow the metal-surface table and 2 m laid cable requirements of the standard in this setup as these rules are not relevant in this frequency range. A clean power supply (Schaffner, NSG 1007-45) providing the EUT with 220 VAC, 50 Hz, together with a variety of combinations of two in-house chamber filters [6] and a commercial chamber filter (ETS, Model: N5007), is used behind the LISNs at the power source side to supply the UPS and create different source impedance situations. Different source situations were created through the inclusion and removal of the chamber filters. As the chamber filters between the LISNs and the power supply were likely to influence the source impedance, they were very instrumental in changing the source impedance. The UPS was tested on 6 different scenarios in aggregate as given and depicted in Table I. As the UPS was predominantly emitting only in the range of 30 Hz - 2 kHz, its test and analysis were stopped at 2 kHz, and we did not proceed beyond 2 kHz whereas the signal generator RF output and harmonic reference device were tested and analysed in the entire CE101 frequency range (30 Hz – 10 kHz). In addition, for scenario 5 and scenario 6, which include multiple chamber filters, we had to confine the frequency range to around 1 kHz because the harmonics start to become very low and unusable beyond 1 kHz for analysis and calculation due to the inclusion of more than one chamber filter. In an attempt to stoke up the harmonics and increase the sensitivity beyond 1 kHz in the use of two chamber filters in series, we also supplied the UPS with 300 VAC instead of 220 VAC and repeated the measurement for the scenario 6. Lastly, in order not to clutter the paper with too many graphs, we give the corresponding absolute impedance and current values, which produce the deviations shown in the graphs, only for one case of the UPS, not for all the test cases. Fig. 5. CE 101 test setup installed with resistors, capacitors and inductors as source impedance simulation and the signal generator RF output as EUT. Fig. 6. CE 101 test setup installed with the harmonic reference device as EUT. Fig. 7. CE 101 test setup installed with the UPS as EUT. TABLE I. UPS MEASUREMENT SCENARIOS After the installation of the required setups, we completed the loop impedance measurements of each of them. The frequencies, at which the impedance measurement was performed, were decided through a preliminary and quick emission test just before the impedance measurement. The preliminary emission check gave us peak values and frequencies for the impedance measurement. In order not to get in conflict with the EUT emission frequencies during the impedance measurement, we slightly shifted each detected EUT emission frequency left and right and recorded these revised frequencies for the impedance measurement. For example, if a device emits at 150 Hz, 450 Hz, 850 Hz, 1.15 kHz, 1.55 kHz, 1.95 kHz, impedance measurements should be carried out at 120 Hz, 180 Hz, 420 Hz, 480 Hz, 820 Hz, 880 Hz, 1.120 kHz, 1.180 kHz, 1.52 kHz, 1.58 kHz, 1.92 kHz, 1.98 kHz in order not to collide with the EUT emission frequencies but also to be close to them for more accurate impedance measurements and correction factors calculated in (1). IV. RESULTS The verification results of the impedance measurement system are presented in Fig. 8. As observed in Fig. 8, the impedance measurement system yields acceptable agreement that proves its trustworthiness. In this simple verification, as the reference values, we directly used the rated values of 0.5 Ω and 50 Ω whereas we used the rated values and the theoretical impedance equations of the capacitor (80 µF) and the inductor (430 µH). (a) (b) (c) (d) Fig. 8. Impedance measurement verification with (a) the 0.5 Ω resistor, (b) the 50 Ω terminator, (c) the 80 µF capacitor, (d) the 430 µH inductor. (a) (b) (c) Fig. 9. Impedance and loop current deviations for the signal generator RF output used as EUT between the different source impedance conditions (a) 50 Ω versus 80 µF, (b) 50 Ω versus 430 µH, (c) 80 µF versus 430 µF. After the impedance verification, the loop current and impedance deviation results of the signal generator RF output used as a dummy EUT in different source impedance conditions are given in Fig. 9. As clearly observed in Fig. 9, although the impedance results are significantly distinct in different source environments, the loop current deviation acceptably follows the loop impedance deviation per graph, which verifies the proposed method in an ideal test environment. In each graph, the impedance deviation signifies the difference in loop impedance magnitudes in decibels between two different source impedance conditions while the current deviation shows the same for the loop current flowing in the test circuits. For example, in Fig. 9 (a), while a deviation of 5 dB in loop impedance occurs due to the change of the dummy source impedance condition from 50 Ω to 80 µF behind the LISNs, the loop current deviates in a similar manner as the loop impedance. The consistent change in both the loop impedance and current deviations may be regarded as a good start for the proposed method. It should be reiterated here that the impedance deviation is in fact the correction factor between the two cases per graph. Fig. 10. Loop impedance and current deviations with respect to 50 Ω for the source impedance conditions: 6.6 Ω, 13.3 Ω, 26.6 Ω for the harmonic reference device. In the same vein, the harmonic reference device results are presented in Fig. 10 in the range of 30 Hz – 10 kHz. Again, there is good agreement in the impedance and current deviation results with respect to 50 Ω. When we change the dummy source impedance from 6.6 Ω to 26.6 Ω with irregular steps and compare the results with the 50 Ω results, the loop impedance and current deviation curves change in harmony with each other. To exemplify this, when we change the dummy source impedance from 6.6 Ω to 50 Ω, it yields a deviation of around 14 dB in both the impedance and current values, as seen in Fig. 10. The good agreement in impedance and current deviations also exists in the other impedance transitions from 13.3 Ω and 26.6 Ω to 50 Ω. (a) (b) (c) (d) (e) Fig. 11. Impedance and loop current deviations with respect to the standard setup (scenario 1) for the UPS supplied by 220 VAC for the source conditions: (a) scenario 2, (b) scenario 3, (c) scenario 4, (d) scenario 5, (e) scenario 6. After the preliminary results obtained with the signal generator RF output and harmonic reference source that may be regarded as ideal or dummy EUTs, ultimately the results of the UPS selected as an actual piece of EUT are presented in Fig. 11 - 12. In this step, the standard CE101 test setup (scenario 1) that only comprises the EUT and the LISNs was selected as the reference setup. In this standard setup, the LISNs were directly connected to our clean AC power source without any chamber filters. In all the other conditions (scenarios 2 - 6), the chamber filters were placed between the LISNs and the power supply in order to change the source impedance and create different test conditions. In Fig. 11 (a), we observe that the commercial chamber filter inserted between the LISNs and the power supply in scenario 2 does not markedly change the source impedance and does not produce significant impedance deviation with respect to the standard CE101 test setup. Similarly, the current does not change in the two test conditions. When we study Fig. 11 (b), we see that inserting our in-house filter 1 between the LISNs and the power supply in scenario 3 changes the loop impedance of the CE101 test setup. It also causes the loop current to deviate similarly. That means that we can calculate the loop current, also called emission level, in a targeted test environment by just measuring the loop impedance values of the reference and the targeted test environment. In our case here, even if we did not measure the loop current value in the targeted environment (scenario 3), we could estimate it by using the loop current in the reference environment (scenario 1) and the impedance deviation considered as the correction factor calculated in (1). We obtain another good consistency in Fig. 11 (c). The in-house filter 2 inserted between the LISNs and the power supply in scenario 4 produces a deviation quite equivalent to the deviation produced by the in-house filter 1. (a) (b) Fig. 12. (a) Impedance and loop current deviations with respect to the standard setup (scenario 1) for the UPS supplied by 300 VAC for the source impedance condition: scenario 6, (b) absolute impedance and loop current values which are used for deviation calculation. When we place the commercial and in-house filters in series between the LISNs and the power supply in scenario 5, it produces an impedance deviation visually alike to scenarios 3 and 4 but with a higher deviation at higher frequencies. As clearly seen in Fig. 11 (d), even at 1.3 kHz, the impedance deviation in scenario 5 reaches -6 dB but we were not able to record the harmonic currents beyond 1.3 kHz as they become very low and useless with the insertion of the two chamber filters in series. In Fig. 11 (e), a similar graph is obtained when the in-house filter 1 and in-house filter 2 are placed in series between the LISNs and the power supply in scenario 6. Also, here we had to limit the frequency range to 1.1 kHz as the magnitudes of the harmonics sagged significantly beyond 1.1 kHz. Finally, the results of the UPS supplied by 300 VAC instead of 220 VAC in an attempt to strengthen the harmonics are shown in Fig. 12 (a) along with the absolute impedance and current values in Fig. 12 (b) just for information. When we increase the supply voltage from 220 VAC to 300 VAC, the strength of the harmonics seems to increase but the stability of the harmonics seems to worsen. As a result, as seen in Fig. 12 (a), when the supply voltage is increased to 300 VAC, acceptable harmonics starts to occur up to 2 kHz but with higher fluctuation. This final result shows the effectiveness of the proposed method as the two curves are reasonably following each other in Fig. 12 (a) up to 2 kHz despite the remarkable fluctuation in the harmonic levels. V. CONCLUSIONS AND FUTURE WORK The experimental results support the effectiveness of the proposed method, that is, low frequency emission testing can be significantly improved throughout live impedance measurements. When the loop impedance and current values are known for a certain test site, the measured emission levels, namely the loop current, can be estimated for other sites/conditions as long as the loop impedance is also known at that targeted site. Alternatively, emission results obtained in different test environments may be scaled to a defined impedance value, such as 50 Ω, to harmonise emission results obtained from different test environments and make them comparable. Even a standard limit corresponding to the defined impedance can be determined for standardisation and source impedance-independent test results. The promising results encountered provide confidence in the proposed approach; however, the methodology remains at the early stages. The research must be further replicated and extended since the experiments conducted so far only covered three EUT types. We will expand the scope to more EUT types and attempt to generalise this procedure. Additionally, we will recommend that emission test results in the low frequency range, e.g. CE101 test results, should be accompanied by and reported together with grid or power source impedance values seen by the EUT and taken at the time of the test. This would allow tracking, explaining and correcting discrepancies in emission test results between test laboratories. The impedance measurement of the grid or power source seen by the EUT can be easily performed by means of the CS101 test system, as introduced in this paper. Although we always focused on the overall loop impedance in this research, only the source side impedance or even only the EUT impedance, if requested, can be measured by just moving the voltage probe from the output of the coupling transformer to the LISN outputs or the EUT input. We expect this method will grow in maturity and eventually could be standardised. Moreover, the knowledge and experience acquired in this research may also be extended to commercial harmonic measurements carried out as per IEC 61000-3-2 [14] or IEC 61000-3-12 [15] in future applications. ACKNOWLEDGMENT This research is performed in the scope of the project “21NRM06 EMC-STD Metrology for emerging electromagnetic compatibility standards”. The project (21NRM06 EMC-STD) has received funding from the European Partnership on Metrology, co-financed by the European Union’s Horizon Europe Research and Innovation Programme and by the Participating States. REFERENCES [1] “Requirements for the Control of Electromagnetic Interference Characteristics of Subsystems And Equipment”, Department of Defence USA, MIL-STD-461G, 2015. [2] A. Karaali, E. Akpinar, O. O. Gursahbaz and B. Ozbey, "The Effects of Shielded Room Power Line Filters on CE101, CE102 and CS101 Test Results," 2022 International Symposium on Electromagnetic Compatibility – EMC Europe, Gothenburg, Sweden, 2022, pp. 691696, doi: 10.1109/EMCEurope51680.2022.9901139. [3] S. Çakir, O. Sen, M. Çinar and M. Çetintas, "Alternative conducted emission measurements for industry," 2014 International Symposium on Electromagnetic Compatibility, Gothenburg, Sweden, 2014, pp. 1037-1042, doi: 10.1109/EMCEurope.2014.6931055. [4] O. Sen, S. Cakır, M. Cınar, M. Pous, F. Silva and M. Cetintas, "Alternative conducted emission measurements on mains without LISNs," in IEEE Electromagnetic Compatibility Magazine, vol. 4, no. 4, pp. 58-65, Fourth Quarter 2015, doi: 10.1109/MEMC.2015.7407180. [5] S. Cakir, M. Ozturk, O. Sen, B. Tektas, S. Acak and M. Cetintas, "MIL STD 46IF CS101 testing and power frequency cancelation," 2016 Asia-Pacific International Symposium on Electromagnetic Compatibility (APEMC), Shenzhen, China, 2016, pp. 855-857, doi: 10.1109/APEMC.2016.7522889. [6] S. Cakir, O. Sen, M. Ozturk, M. Celep and T. Acarer, "Investigation of undesired ripple voltage across power source side and solutions in military low-frequency immunity tests," in IEEE Electromagnetic Compatibility Magazine, vol. 8, no. 3, pp. 65-73, 3rd Quarter 2019, doi: 10.1109/MEMC.2019.8878239. [7] S. Cakir, O. Sen, M. Ozturk and M. Celep, "Influence of Injection Point Selection on Military Low Frequency Immunity Test Results," 2019 ESA Workshop on Aerospace EMC (Aerospace EMC), Budapest, Hungary, 2019, pp. 1-5, doi: 10.23919/AeroEMC.2019.8788952. [8] P. T. Jensen and P. Davari, "Power Converter Impedance and Emission Characterization Below 150 kHz," 2021 IEEE International Joint EMC/SI/PI and EMC Europe Symposium, Raleigh, NC, USA, 2021, pp. 255-260, doi: 10.1109/EMC/SI/PI/EMCEurope52599.2021.9559177. [9] S. Cakir, M. Oztürk, B. Tektas, O. Şen, S. Acak and M. Pous, "FFTbased time domain solution to power frequency issue of CS101 testing for military and aerospace equipment," 2018 IEEE International Symposium on Electromagnetic Compatibility and 2018 IEEE AsiaPacific Symposium on Electromagnetic Compatibility (EMC/APEMC), Suntec City, Singapore, 2018, pp. 177-182, doi: 10.1109/ISEMC.2018.8393762. [10] S. Cakir, O. Sen and M. Ozturk, "Investigation of Ripple Voltage Across Capacitor in Military CS101Test by Using FFT -Based Time Domain Solution," 2018 IEEE Symposium on Electromagnetic Compatibility, Signal Integrity and Power Integrity (EMC, SI & PI), Long Beach, CA, USA, 2018, pp. 82-87, doi: 10.1109/EMCSI.2018.8495307. [11] “Time Domain EMC Emissions Measurement System”, Final Report, AY4489, May 2004. Online:https://www.ofcom.org.uk/__data/assets/pdf_file/0019/40348/ timedomainemc.pdf [12] Bartlett, M, “Smoothing Periodograms from Time Series with Continuous Spectra”, Nature (London), Vol. 161, pages 686-687, 1948. [13] P. Welch, "The use of fast Fourier transform for the estimation of power spectra: A method based on time averaging over short, modified periodograms," in IEEE Transactions on Audio and Electroacoustics, vol. 15, no. 2, pp. 70-73, June 1967, doi: 10.1109/TAU.1967.1161901. [14] IEC 61000-3-2:2014 Electromagnetic compatibility (EMC) Limits. Limits for harmonic current emissions (equipment input current ≤16 A per phase). [15] IEC 61000-3-12:2011 Electromagnetic Compatibility (EMC) - Limits For Harmonic Currents Produced By Equipment Connected To Public Low-Voltage Systems With Input Current >16 A And ≤ 75 A Per Phase