conducted electromagnetic interference issues in
TRANSCRIPT
255
10th International Symposium „Topical Problems in the Field of Electrical and Power Engineering“
Pärnu, Estonia, January 10-15, 2011
Conducted Electromagnetic Interference Issues in Distributed Systems
Robert Smolenski, Adam Kempski University of Zielona Gora
[email protected], [email protected]
Abstract This paper addresses the specific issues associated with electromagnetic compatibility that should be taken into account at the developmental stage of distributed systems. The main aim is to establish how far conducted interferences can penetrate the electric grid and how the group of converters, which individually meet EMC standards, influence the mains supply.
Presented in this paper are the measuring results of the spread of electromagnetic interference (EMI) current over a typical local electric grid and the low and medium voltage side of the power transformer. The four-quadrant converter that is commonly applied in distributed power systems has been used as a test interference source.
The limitations in the applicability of standardized methods for evaluation of aggregated conducted interferences have been presented in a system consisting of a group of two-quadrant frequency converter drives.
Keywords
Electromagnetic Compatibility; Conducted Interferences; Distributed System; Inverter; Frequency Converter
Introduction
The four-quadrant frequency converter is currently often applied in various configurations of modern distributed systems of electric energy generation, transmission and conversion [1]. This converter typically generates a high level of electromagnetic interferences, especially in the conducted interference frequency range [1, 2]. The high emission results from an energy conversion method utilising two three-phase bridges which can adversely influence both line and load side. The analysis of electromagnetic compatibility of a four-quadrant converter and propositions for effective filtration methods have been presented in previous papers [2–4].
This paper presents a continuation of the research concerning penetration of the interferences deep into the electric grid. Additionally, interferences introduced into the mains by the group of two-quadrant frequency converters are evaluated. It has
been shown that the standardized methods of the conducted EMI measurements are useless or give erroneous results for a system consisting of several converters, because of the phenomena accompanying the interference summation.
1 Conducted EMI generated by three-phase ASD
The common mode (CM) currents, which are generated by switching states of the inverter, have an impulse-like waveform with high frequency (HF) oscillation. The frequency of the oscillatory mode is determined by the values of residual, parasitic parameters of the CM current path, Fig. 1. Then, the CM currents split according to the proportion of HF impedance of a PE cable wire (or shield) and HF impedance of the grounding arrangement between the grounding points of the inverter and the motor. The main return path for the CM currents passes via the heatsink to DC link capacitance [2].
The CM current causes a CM voltage drop on heatsink to DC link capacitance. In a blocking state of the diodes of the rectifier, only a small HF part of this current flows through the parasitic capacitance of the diode and converter supply arrangement. In the conduction state, this voltage drop causes oscillation of small amplitude and relatively low frequency in a closed loop consisting of a DC-link-to-heatsink capacitance and the resultant inductance of the mains or Line Impedance Stabilization Network (LISN), the cable and the input filter.
The CM currents and phase current on the line side of the converter in conduction and blocking state of the rectifier diodes have been shown in Fig. 2.
The time of appearance of both CM and DM currents depends on inverter control algorithm. Additionally, current level and oscillation frequency is modulated by conduction state of the input rectifier. Fig. 3 shows the spectrum of CM current on the line side of the converter. There are sidebands harmonics equal to 300 Hz resulting from modulation of inverter carrier frequency by means of 6-pulse rectifier. This is only slightly visible in normalized EMI measurements because of the selectivity of the intermediate frequency bandwidth (IF BW) that is equal to 200 Hz for frequency band 9 kHz-150 kHz (CISPR A).
256
10us/div0.5A/div
a.)
0.2us/div0.2A/divb.)
10us/div20mA/divc.)
Fig.1. a) Common mode current in motor PE wire, b) HF component of CM current, c) LF oscillatory of CM current
0.2ms/div50mA/div
a.)
0.2ms/div0.5A/div
if
iCM
10us/div50mA/div
b.)
10us/div50mA/div
c.)
Fig. 2. a) CM current on the line side of the converter and rectifier phase current, b) CM current in conduction state of the rectifier, c) CM current in blocking state of the rectifier
-40dBm
LogMag
dBm-100
/divdB 6
D: Ch1 Spectrum -10 dBmRange:
1 kHzSpan: Center: 16 kHz382.0312 mSecTimeLen: RBW: 10 Hz
Fig. 3. Spectrum of line side CM current at carrier frequency with sideband harmonics caused by 6-pulse rectifier
2 Conducted EMI generated by group of two-quadrant frequency converter-fed drives
Measurements have been taken in system consist of three identical 1 kW induction motor drives supplied via LISN. The converter scheme with EMI filter arrangement is presented in Fig. 4.
2x800mH
2x22nF66nF
3x50nF
Fig. 4. Frequency converter scheme with EMI filter arrangement The results of measurements have taken in this systems are presented in Fig. 5. In order to assure unchangeable interference currents circuits all of the frequency converters were connected to LISN for whole measurement time. Only inverters were switched on/off. The measurements for one, two, and three drives have been carried out like for one piece of the equipment in full compliance with EN 61800-3.
All of the oscillatory mode can be distinguished as those presented in time domain current waveform, Fig. 1 and Fig. 2. It is possible to observe that conducted EMI introduced to electric grid by group of the converter drives can be much higher than the level of EMI generated by the single drive. It means that we might expect increased number of EMC related problems in systems containing large number of the converters. In order to establish how interferences are aggregated additional tests not required by standards are performed.
Fig. 6 shows additional test for CISPR A frequency band IF BW=200 Hz.
257
30
35
40
45
50
55
60
65
70
75
80
85
90
150k 300 500 1M 2M 3M 4M 5M 6 8 10M 20M 30Mf [Hz]a.)
[dB
uV]
b.)30
35
40
45
50
55
60
65
70
75
80
85
90
150k 300 500 1M 2M 3M 4M 5M 6 8 10M 20M 30Mf [Hz]
[dB
uV]
c.)
30
35
40
45
50
55
60
65
70
75
80
85
90
150k 300 500 1M 2M 3M 4M 5M 6 8 10M 20M 30Mf [Hz]
[dB
uV]
Fig. 5. Conducted EMI according to EN 61800-3. a) single drive, b) two drives, c) three drives
50
55
60
65
70
75
80
85
90
95
100
105
110
9k 20k 30k 40 50 60 80 100k 150kf [Hz]a.)
[dB
uV]
b.)
50
55
60
65
70
75
80
85
90
95
100
105
110
9k 20k 30k 40 50 60 80 100k 150kf [Hz]
[dB
uV]
c.)
50
55
60
65
70
75
80
85
90
95
100
105
110
9k 20k 30k 40 50 60 80 100k 150kf [Hz]
[dB
uV]
Fig. 6. Conducted EMI spectra in CISPR A range. a) single drive, b) two drives, c) three drives In CISPR A frequency band the level of measured interferences increases with rising number of operating converters as well. The level changes between single drive and three drives reached 8dB. However, differences in results obtained for two and three drives, in multiple tests for the same measuring conditions, indicated that signals had been modulated by very slowly changeable envelope.
3 Aggregation of CM interferences
The spectrograms presented in Fig. 7 show CM current level variation of 16 kHz harmonic with sidebands vs. time for a) one drive, b) two drives, c) three drives. Spectrograms illustrate the changes of the level of carrier harmonic during 31.79 s. The bigger number of the operated converters the higher maximum level of the measured current. However, modulation envelope causes the decreasing of the interference level below those measured in a case of the single drive.
40mArms
-10 dBmRange:
650 HzSpan: Center: 16 kHzRBW: 12 Hz
31.79 Sec
1mArms
a.) 40
mArms
-10 dBmRange:
650 HzSpan: Center: 16 kHzRBW: 12 Hz
31.79 Sec
1mArms
b.) 40
mArms
-10 dBmRange:
650 HzSpan: Center: 16 kHzRBW: 12 Hz
31.79 Sec
1mArms
c.) Fig. 7. Spectrograms of interference currents at carrier frequency. a) single drive operates, b) two drives operate, c) three drives operate This is important to state that semiconductors act just like the radio frequency (RF) detector in a ‘crystal radio set’ and will demodulate whatever RF signals. Immunity measurement practice has shown that electronic equipment is much more sensitive to modulated signals. In systems consist of many converters controlled using different algorithms we should expect wideband envelopes of interferences that may cause more problems with assurance of the equipment immunity.
Vector signal analysis is useful method that can be successfully applied to investigate modulation influence on interferences.
Fig. 8 shows vector diagram of maximum interference current levels for a) single drive, b) two drives, c) three drives. The rising of current level and modulation effects are easy to evaluate.
258
150mA
I-Q
mA-150
/divmA 30
D: Ch1 Main Time 1ARange:
200 mA-200 mA
a.) b.)
150mA
I-Q
mA-150
/divmA 30
D: Ch1 Main Time 1ARange:
200 mA-200 mA
c.)
150mA
I-Q
mA-150
/divmA 30
D: Ch1 Main Time 1ARange:
200 mA-200 mA
Fig.8. Vector diagrams of maximum interference current levels for a) single drive, b) two drives, c) three drives
The box-and-whisker plots presented in Fig. 9 show the distributions of thousand of final measurements taken using average detector in normalized time equal to 1 s for switching frequency. The distributions of measuring results, especially in a case of group work of the converters, differ significantly. The differences reached 17 dB that showed ineffectiveness of standard measuring procedures for the evaluation of the aggregated conducted interferences generated by the group of converters.
The highest level recorded for three drives operated together was 6 dB bigger than the highest level observed for the drive 2 in a case of the single drive operation.
25%75%50%Max
Min
25%
75%50%Max
Min
25%
75%50%
Max
Min
Drive 1 Drive 2 Drive 380
90
100
110
Ave
rage
Det
ecto
r Le
vel [
dBuV
]
Drive 1 (Level dBuV) Drive 2 (Level dBuV) Drive 3 (Level dBuV)
25%
75%
50%
Max
Min
25%
75%
50%
Max
Min
25%
75%50%
Max
Min
25%
75%
50%
Max
Min
Drive 1&2 Drive 2&3 Drive 1&3 Drive 1&2&380
90
100
110
Ave
rage
Det
ecto
r Le
vel [
dBuV
]
Drive 1&2 (Level dBuV) Drive 2&3 (Level dBuV) Drive 1&3 (Level dBuV) Drive 1&2&3 (Level dBuV)
Fig.9. Box-and-whisker plot of average detector measurements for: a) single drives, b) group of drives
4 CM interferences generated by four-quadrant frequency converter
The four-quadrant frequency converter is currently commonly used in novel asynchronous drives and in distributed power generation systems as converters for asynchronous and permanent magnet variable speed generators. The main circuit of this converter consists of two IGBT based three-phase bridges and an intermediate circuit allowing two-way energy flow and four-quadrant operation. The switching of IGBT devices with typically high dv/dt causes high level of electromagnetic interferences (EMI) especially in conducted electromagnetic emissions frequency range (9 kHz - 30 MHz).
We have tested a two-pole, 10 kW induction generator connected with mains by industrial four-quadrant frequency converter supplied via LISN. Fig. 10 shows the results of measurements which have been carried out on the system consisting of LISN and EMI receiver ESCS-30, in the frequency range 150 kHz – 30 MHz (CISPR B) specified in EN 61800-3 .
a.)0
20
40
60
80
100
120
130
Level
[dBµV]
150k 300k 500k 1M 2M 3M 4M 6M 10M 30MFrequency
[Hz]
+
+ ++++
x
xxxx
x
CISPR B
b.)40
60
80
100
120
140
Level
[dBµV]
9k 20k 30k 40k 50k 60k 80k 100k 150kFrequency
[Hz]
CISPR A
Fig.10. Conducted EMI spectra for ranges: a) CISPR B, b) CISPR A We can observe spectra envelopes that indicate oscillation modes of EMI current waveforms at frequencies 2.5 MHz, 3.8 MHz and 4.7 MHz. The limits are slightly exceeded at the frequency 2.5 MHz, and significantly at the beginning of the CISPR B band (150 kHz). The shape of the EMI envelope [1] implies that its source could be located in a lower frequency range.
Additional measurement in the lower frequency part of the spectrum (CISPR A), not required by standards, have shown repeatable changes at frequencies 40 kHz, 80 kHz, … We have identified them with the time of the synchronized impulse of transistors switching (25 µs) and its harmonics. The envelope of this spectrum is characteristic for a damped sine wave pulse at a frequency of about 70 kHz [2].
259
5 Penetration of EMI into LV electric grid
The main reason for presented investigations concerning deepness of interference penetration into mains were observed malfunctions of electronic equipment caused by 4-quadrant converter located in relatively distant circuits.
In order to evaluate the deepness of interference penetration into electric grid the measurements of CM currents in PE wires in different points of local mains had to be performed. The converter was connected straight to mains without LISN. The result of CM current measurement in CISPR A frequency band is shown in Fig. 11.
20
30
40
50
60
70
80
90
100
110
Level [dBµA]
9k 20k 30k 40k 50k 60k 80k 100k 150kFrequency [Hz]
Fig. 11. Spectrum of CM current in converter PE wire Fig. 12 shows spectrum of CM current in the PE wire of the power cable that supplies the laboratory. Measuring point was located in transformer station near common PE bus above 200 m away from interference source. Fig. 12,a shows spectrum of the background noises and Fig. 12,b shows spectrum measured during converter operation [1].
20
30
40
50
60
70
80
90
100
110
Level [dBµA]
9k 20k 30k 40k 50k 60k 80k 100k 150kFrequency [Hz]
20
30
40
50
60
70
80
90
100
110
Level [dBµA]
9k 20k 30k 40k 50k 60k 80k 100k 150kFrequency [Hz]
Fig.12. Spectra of current in PE wire of power cable at transformer terminal for: a) switched off converter, b) switched on converter
The level of the interference in Fig. 12,b increases significantly. At the frequency 60 kHz that constitutes main oscillatory mode of the current the level of interferences increased 100 times (40 dB) compared to background interferences. The observed level is only 20 dB lower in comparison with interferences measured in PE wire near converter in spite of the existence of many alternative paths for the interference flow in the laboratory hall.
6 Penetration of EMI into MV electric grid
Fig. 13 shows the results of magnetic field strength measurements in the power transformer station on both low and medium voltage sides.
The necesity of interference investigations in medium voltage (MV) grids forced the application of field measuring method for conducted electromagnetic interferences. The active loop antenna was used for measurements of interference penetration depth into MV grid. The investigations were carried out in an urban type transformer station. The generator and converter was connected to LV side of the 160 kVA transformer.
9k 40k 80k 120k20
40
60
80
100H
[dB
uA/m
]
Frequency [Hz]
9k 40k 80k 120k20
40
60
80
100
H [d
BuA
/m]
Frequency [Hz] Fig.13. Magnetic field strength on both sides of power transformer: a) LV side, b) MV side It is important to note that the LV and the MV side of the transformer were located in relatively distant points on opposite buildings’ sides. Presented experimental results show that interferences introduced by the converter, are transferred by
260
parasitic capacitive couplings onto MV side of the transformer (not according to transformer ratio). In this case the transformer cannot be treated as an attenuating device for high frequency interferences.
Further investigations were performed under overhead MV lines. The first measurement was taken 20 m away from transformer station. The second measuring point was located under overhead MV line 1300 m away from the transformer station. In both cases the loop antenna was oriented along the lines in order to assure maximum level of interferences measured in the near field.
Fig. 14 shows an increase of interferences caused by converter in comparison with background interferences under MV overhead lines 20 m and 1300 m away from transformer station.
9k 40k 80k 120k-20
0
20
40
60
80
dB
Fraquency [Hz]
9k 40k 80k 120k-20
0
20
40
60
80
dB
Frequency [Hz] Fig. 14. Increase of interferences caused by converter under MV overhead lies: a) 20 m away from station, b) 1300 m away from station Presented results show that the four-quadrant converters, generating high level of conducted EMI, connected to LV grid may cause 40...60 dB increase of interferences in distant points under MV lines. For characteristic, oscillatory mode frequencies, introduced by the converter, the observed attenuation came to 10...20 dB for those two points. It is important to note that due to travelling wave, especially standing wave phenomena, the measured attenuations should be treated as approximate levels and might change along the line.
Conclusion
The assurance of electromagnetic compatibility is an important factor conditioning the development of distributed power systems. The connection of susceptible control and communications electronic equipment to a high emission power electronics sector of a distributed system requires a special caution and in-depth EMC analysis to ensure system reliability.
The experimental results have shown that the aggregated interference components introduced into the electric grid by the system and consisting of the same power electronic converters that individually fulfill EMC requirements might act as an incompatible unit. These kinds of systems may cause problems with both internal and external electromagnetic compatibility.
Aggregated interference signals are modulated by slowly changeable envelopes that cause measuring difficulties. Normalized tests with their typical measuring times are not sensitive enough to properly evaluate interference levels. In order to investigate the aggregated interference components in systems containing a group of power electronic converters special measuring methods need to be developed.
The presented results of the investigations have shown that interference caused by a four quadrant converter, which is often used in distributed power systems, can reach a distant point of the local low voltage grid. Interference coupled by common impedance might cause immunity problems in the distorted mains. Additionally, magnetic field measurements have shown that such interference can be transferred by parasitic couplings on the medium voltage side of the transformer.
References
1. R. Smolenski, Selected conducted electromagnetic interference issues in distributed power systems, Bulletin of the Polish Academy of Sciences: Technical Sciences - 2009, Vol. 57, pp. 383-393.
2. Kempski A., Smolenski R., Aggregated Conducted EMI Generated by Group of Frequency Converter-Fed Drives, CPE’09, 2009
3. Akagi H., Doumoto T., A passive EMI filter for preventing high-frequency leakage current from flowing through the grounded inverter heat sink of an adjustable-speed motor drive system, IEEE Trans. on Ind. Appl. Vol.41, 2005, 1215-1223
4. Akagi H., Hasegawa H., Doumoto T., Design and performance of a passive EMI filter for use with voltage source PWM inverter having sinusoidal output voltage and zero common-mode voltage, IEEE Trans. on Power Electr. 19, 2004, 1069