Professional Documents
Culture Documents
2013 - Common-Mode Voltage Reduction Pulsewidth Modulation Techniques For Three-Phase Grid-Connected Converters
2013 - Common-Mode Voltage Reduction Pulsewidth Modulation Techniques For Three-Phase Grid-Connected Converters
(010) V 3 V 2 (110) In the following, the important RCMV PWM methods will be
briefly discussed and their pulse patterns will be shown.
Sa 1 1 1 0 1 1 1 Sa 0 1 0 1 0 Sa 1 1 1 1 1
Sb 0 0 1 1 1 0 0 Sb 1 0 0 0 1 Sb 1 0 0 0 1
Sc 1 0 0 0 0 0 1 Sc 0 0 1 0 0 Sc 0 0 1 0 0
V6 V1 V2 V3 V2 V1 V6 V3 V1 V5 V1 V3 V2 V1 V6 V1 V2
V no V no V no
V dc /6 V dc /6
0 0 0
-V dc /6 -V dc /6 -V dc /6
Fig. 3. Illustrations of various RCMV-PWM schemes. (a) AZSPWM1 in region A1. (b) RSPWM3 in region B1. (c) NSPWM in region B1.
Smaller HDF implies smaller ac inverter flux ripple that re- For grid-connected PV applications, the converter usually
sults in smaller output filter size for the same current ripple operates with unity power factor and injects real power into
value, or smaller ripple with the same filter size. The dc-link cur- the grid. The dc bus voltage is maintained only at a sufficient
rent of the converter affects the dc-link capacitor sizing because level for effective current control without overmodulation and
the capacitor needs to absorb the PWM ripple current. Each to uphold the conversion efficiency. Thus, the converter usually
RCMV-PWM technique has its unique ripple current, which de- operates with high modulation index. The linear modulation
pends on the modulation index Mi and the displacement power range of RSPWM is limited, for example, RSPWM3 is linear up
factor angle φ. The ratio of the harmonic RMS value of the dc- to Mi = 0.604. Inevitably, it will require higher dc bus voltage
link current Idch -rm s and the converter ac output fundamental for grid connection [8], which likely leads to higher operational
current RMS value I1rm s is evaluated in [6] and [9] as the dc losses. Even at low modulation index range, RSPWM methods
link current quality factor may require larger filter due to its high HDF and Kdc .
RSPWM methods and AZSPWM2-3 methods require simul-
Idch -rm s
Kdc = . (5) taneous switching of two converter poles to reduce their CMV.
I1rm s However, in practice this is hardly realized because of the dead-
Smaller Kdc implies less capacitor ripple current and less stress. time, differences in gate-delay, and nonidentical semiconductor
The Kdc factor for various RCMV-PWM methods are given as devices. The CMV often becomes as large as conventional. In
follows [10]: addition, the high CMV caused by dead-time and mismatches
1) AZSPWM in these two methods happens as very narrow CMV pulses,
√ √ and results in high over-voltages which may damage the insula-
9 3 3 3 tion of the PV panels. The HDF and Kdc characteristics given
Kdc = 1 + Mi 2 − cos (2ϕ)
π 2π in Fig. 4 show that both AZSPWM1 and NSPWM have good
18 performance at high modulation index. In NSPWM, only two
− Mi2 cos2 (ϕ) (6) converter poles switch in every PWM cycle, and the other con-
π2
verter pole is locked to either the positive or the negative rail of
2) RSPWM the dc bus. This is a significant advantage as the PV converter
6 18 systems place the conversion efficiency as the top priority. Of
Kdc = 1 + Mi cos (2ϕ) − Mi2 2 cos2 (ϕ) (7) the RCMV PWM methods, only AZSPWM1 and NSPWM will
π2 π
be considered and compared with the conventional SVPWM
3) NSPWM and DPWM1 methods in this paper.
√
24 3 3 18
Kdc = 1 + Mi 2 − cos (2ϕ) − Mi2 2 cos2 (ϕ). IV. CARRIER-BASED AZSPWM1 AND NSPWM
π π π
Both AZSPWM1 and NSPWM can be easily implemented
(8) with their respective zero-sequence injection command and two
1974 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 28, NO. 4, APRIL 2013
V0
(a)
(a)
Frequency (Hz)
Frequency (Hz)
CMC (A)
CMC (A)
Frequency (Hz)
Frequency (Hz) (b)
(b)
Magnitude (V)
Magnitude (V)
Frequency (Hz)
(c)
Frequency (Hz)
(c) Fig. 7. AZSPWM1 at 10 kHz. (a) Phase current and the CMC; 2 A/div,
5 ms/div. (b) Spectra of phase current (top) and CMC (bottom). (c) Spectrum of
Fig. 6. SVPWM at 10 kHz. (a) Phase current and the CMC; 2 A/div, 5 ms/div. the CMV.
(b) Spectra of phase current (top) and CMC (bottom). (c) Spectrum of the CMV.
(a) (a)
Phase Current (A)
CMC (A)
Magnitude (V)
Frequency (Hz)
(c) Frequency (Hz)
(c)
Fig. 8. DPWM1 at 15 kHz. (a) Phase current and the CMC; 2 A/div, 5 ms/div.
(b) Spectra of phase current (top) and CMC (bottom). (c) Spectrum of the CMV. Fig. 9. NSPWM at 15 kHz. Phase current and the CMC; 2 A/div, 5 ms/div. (b)
Spectra of phase current (top) and CMC (bottom). (c) Spectrum of the CMV.
a base frequency of 30 kHz, and for 10 kHz of SVPWM- Most grid-tied inverters adopt LCL filters on the grid side. The
AZSPWM1 this corresponds to 30 kHz. As a result, the domi- filter capacitor terminal voltage is equivalent to the ideal grid
nant components of CMV for all methods are around 30 kHz. source voltage as in Fig. 1. The inverter side inductor, typically
However, as the waveforms do not have full symmetry, lower a few percent per unit in terms of its impedance, is set such that
frequency components down to the carrier frequency exist. The the inverter ripple current is sufficiently small. The grid side
CMV harmonic spectrums of the studied methods are shown inductor is typically smaller than the inverter side inductor. The
in Figs. 6(c), 7(c), 8(c), and 9(c) to confirm this result. With line current THD is then an issue of PWM methods. NSPWM
the CMV harmonic spectrum being wide, and the resonant fre- at high modulation index range has low ripple. DPWM1 and
quency of the zero sequence equivalent circuit being fairly low SVPWM also have low ripple. All these methods achieves sat-
for PV applications, there exists a risk of exciting the resonance. isfactory level of output current THD, whether on the inverter
In fact, this becomes the case for the conducted experiment. side or on the line side.
HOU et al.: COMMON-MODE VOLTAGE REDUCTION PULSEWIDTH MODULATION TECHNIQUES 1977
Impedance (Ω)
2 (15kHz, 161 ohm)
10
Stray capacitance Cstray = 0.1 µF
(10kHz, 54.25 ohm)
TABLE II 10
1
MEASUREMENT OF PWM METHODS UNDER TEST (WITHOUT SNY
COMMON-MODE FILTER)
0
SVPWM AZSPWM1 DPWM1 NSPWM 10
Carrier 10k 10k 15k 15k
frequency
−1
(Hz) 10
3 4 5
CMC 2.95 0.77 0.99 0.37 10 10 10
(A,RMS) Frequency (Hz)
Phase current 3.4 3.0 2.8 3.0
(A, RMS) Fig. 11. Impedance of the CMC equivalent circuit.
Phase current 4.2 4.6 4.1 4.2
THD (%)
DC bus ripple 1.01 0.37 0.27 0.21
(V, pk-pk) the frequency increases the circuit becomes more inductive and
the CMV with large magnitude cannot force large CMC due
V cm L cm to increasing impedance at increasing frequencies. Thus, the
o filter design and control of the grid-connected converter with
NSPWM become easier tasks.
The PV grid interface application involves a major differ-
C stray C stray ence from the conventional motor drive application in terms of
i cm common-mode behavior. In motor drive applications, the CMC
n is strongly dependent on dv/dt of the inverter, and partially on
the CMV magnitude and waveshape. The peak CMC is typically
determined by the dv/dt of the inverter and is weakly dependent
Fig. 10. Common-mode model of transformerless grid-connected PV system.
on the PWM method. This is because the parasitic capacitance
of the motor drives is very small (several nanofarad) and the
Fig. 10 shows the common-mode equivalent circuit model of motor common-mode inductance is approximately 1%–4% pe-
the converter shown in Fig. 1. The common-mode equivalent runit. As a result the resonant frequency is high (order of magni-
inductance can be expressed as tudes higher than the carrier frequency). Further, in motor drives
1 the carrier frequency is typically lower than the grid-connected
Lcm = (Lsa + Lsb + Lsc ). (11) converters; therefore, the CMV base frequency is far below the
3
common-mode equivalent circuit resonant frequency and the
The resulting resonant frequency of the common-mode circuit circuit can be considered open circuit for the carrier frequency
is calculated range. As a result, the CMV magnitude contribution is limited
1 to CMC. However the dv/dt transitions force a larger current
ω0,cm = . (12)
Lcm · 2Cstray (also including the effect of electrodischarge machining phe-
nomenon). Fig. 12(a) shows the CMC and CMV waveforms of
The impedance of the CMC equivalent circuit given in Fig. 11. NSPWM with 15-kHz carrier and Cstray = 1nF. In this test, the
With the three single-phase ac filter inductors of 5mH and the resonant frequency of the common-mode equivalent circuit is
ground capacitance of 0.2μF, the resulting resonance frequency 123.2 kHz, far above the NSPWM carrier frequency. The CMC
is at 8.7 kHz, which is a bit below the tested carrier frequen- contains current spikes closely associated with the dv/dt of the
cies (10 kHz for SVPWM-AZSPWM1, 15 kHz for DPWM1- CMV.
NSPWM). As can be seen from the CMC currents, the CMC of In PV applications due to the large capacitance, the resonant
SVPWM and AZSPWM1 is very high at the carrier frequency. frequency is low and close to the PWM frequency such that the
This is mainly due to the resonance. While the CMV of all the CMV magnitude plays a major role in determining the CMC
methods at the 30 kHz is most dominant, the CMC for this peak and RMS values. Thus, a PWM method with low CMV
frequency is less than the lower frequency current due to this magnitude and low RMS value becomes favorable. NSPWM
resonance. Based on this result, it becomes obvious NSPWM that carries these properties is, therefore, favorable for this ap-
which has its carrier at 15 kHz which is almost twice distant plication. The experimental waveforms confirm that the CMC
from the common-mode circuit resonant frequency (of 8.7 kHz) consists of PWM frequency and multiple frequency ripple that
has a significant advantage of avoiding the resonance. Thus, the is dominant to the ripple of switching instants due to dv/dt.
CMC of NSPWM is much lower. It is worth to mention that as Fig. 12(b) shows CMC and CMV of NSPWM with 15 kHz
1978 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 28, NO. 4, APRIL 2013
2.5
1.5
−0.5
−1
−1.5
−2
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Time (Sec.) x 10
Fig. 13. DC bus voltage ripples. From top to bottom: SVPWM, AZSPWM1,
DPWM1, and NSPWM.
VI. CONCLUSION
This paper presents a survey of common-mode reduction
Fig. 12. CMC (top: 2 A/div, 50 μs/div) and CMV (bottom: 200 V/div,
PWM techniques suitable for grid-connected PV converters.
50 μs/div) of NSPWM at 15 kHz with different C stray . (a) C stray = 1nF. Due to the substantial parasitic capacitance of the PV panels,
(b) C stray = 100 nF. the ground current produced to the converter PWM operation be-
comes a critical safety issue. The German standard VDE 0126-
1-1 very has specific limits for ground leakage and fault current.
carrier and Cstray = 100nF. In this case, the resonant frequency Since the conversion efficiency is the critical factor for grid-
is 8.7 kHz. The resulting CMC waveform is much smoother. connected PV systems, transformerless dc/ac converter systems
Fig. 13 shows the dc bus voltage ripple waveforms under the are often favored. The CMC becomes an even more pronounced
four modulation methods. The peak-to-peak values are 1.01, issue due to the lack of galvanic isolation.
0.37, 0.27, and 0.21 V for SVPWM, AZSPWM1, DPWM1, Among all the common-mode reduction PWM techniques,
and NSPWM, respectively. NSPWM is superior for this case this paper selects AZSPWM1 and NSPWM based on their
as predicted with the Kdc curve of Fig. 4(b) This feature of outstanding performances, in terms of ripples at the ac side
NSPWM is due to the fact that with unity power factor and (HDF) and at ac side (Kdc ), especially at high modulation in-
high modulation index, the dc bus current instantaneous and dex range. They are tested along side their “conventional” PWM
average voltage values over each PWM cycle are closer to each counterpart, SVPWM and DPWM1, respectively. The outcomes
other than other methods. This fact was illustrated in [13] in show that by nullifying the zero vector states, AZSPWM1 and
detail. This favorable result has a significant meaning in PV NSPWM exhibit superior performance in reducing the CMC.
applications. It has been reported that the dc bus voltage ripple Because NSPWM operates with discontinuous voltage refer-
has negative effect on the PV panels (performance and life). ences, only two converter poles switch within one PWM cy-
Thus, less ripple implies better performance and life of PV cle. Under the constraint of same number of switchings per
system. Further, small ripple favors decreasing the electrolytic unit time, NSPWM can operate with a carrier wave frequency
capacitor size and allowing reduced cost. Thus, NSPWM is the 1.5 times higher than what AZSPWM1 operates with. And this
most favorable method for PV grid interface in terms of dc bus explains why NSPWM has the best performance among the
side performance also. examined modulation techniques. NSPWM can meet the re-
The resonant frequency of the common mode is closely re- quirement of industry standards like VDE 0126-1-1 with the
lated to the equivalent common-mode inductance and the stray smallest common-mode filter.
capacitance. For the grid side filter inductor, some manufactur- Most RCMV-PWM techniques were first developed for mo-
ers use three-phase-integrated inductor, which is more econom- tor drives applications. In motor drives, the parasitic capaci-
ical, and some use three single-phase inductors, which are better tance is very small, and the resulting resonance frequency of the
cooled and more reliable. In either case the common-mode fil- common-mode equivalent circuit is much higher than the PWM
ter can be added if CMC needs to be further suppressed. The carrier frequency. Thus, the peak CMC is more dependent on
resulting resonant frequency for the common mode is likely to the dv/dt of the inverter, and less dependent on the magnitude
HOU et al.: COMMON-MODE VOLTAGE REDUCTION PULSEWIDTH MODULATION TECHNIQUES 1979
and wave shape of the CMV. However, for PV applications, the Chung-Chuan Hou (S’08–M’09) received the B.S.
parasitic capacitance is much more significant, and the resulting and M.S. degrees in 1987 and 1989, respectively, in
the Department of Power Mechanical Engineering,
resonance frequency may fall below the PWM carrier frequency and the Ph.D. degree in 2009 in the Department of
as the grid-converter tends to switch at high frequency to reduce Electrical Engineering all from the National Tsing
filter size and meet harmonic requirement. In this case, the CMV Hua University, Hsinchu, Taiwan.
He is currently an Associate Professor in the De-
magnitude and frequency are critical in determining the CMC partment of Electrical Engineering, Chung Hua Uni-
peak and RMS values. Again, NSPWM becomes favorable due versity, Hsinchu, Taiwan. His research interests in-
to its low CMV magnitude and low RMS value. For up to hun- clude active power filters, microgrid, and smart grid
issues.
dreds of kilowatt range of PV inverters, the dc bus voltage is
typically below 1000 V for two-level inverters. These PWM
methods are applicable, and the CMC and CMV discussions are
valid for this capacity range.
REFERENCES
Chih-Chung Shih received the B.S and M.S. degrees
[1] M. Calais and V. G. Agelidis, “Multilevel converters for single-phase grid in the Department of Electrical Engineering, National
connected photovoltaic systems an overview,” in Proc. IEEE Int. Symp. Tsing Hua University, Hsinchu, Taiwan, in 2006 and
Ind. Electron., Jul. 1998, vol. 1, pp. 224–229. 2010, respectively.
[2] M. Meinhardt and P. Mutschler, “Inverters without transformer in grid con- He is currently with the Delta Electronics
nected photovoltaic applications,” in Proc. 6th Eur. Conf. Power Electron. Inc., Taoyuan, Taiwan. His research interests in-
Appl., Sep. 1995, vol. 3, pp. 3086–3091. clude power converters and pulsewidth modulation
[3] Automatic Disconnection Device Between a Generator and the Public methods.
Low Voltage Grid, Paragraph 4.7.1 photovoltaik, dke deutsche kommission
elektrotechnik elektronik informationstechnik im din und vde, Standard din
vde 0126-1-1, Feb. 2006.
[4] Y.-S. Lai and F.-S. Shyu, “Optimal common-mode voltage reduction
PWM technique for inverter control with consideration of the dead-time
effects—Part I. Basic development,” IEEE Trans. Ind. Appl., vol. 40,
no. 6, pp. 1605–1612, Nov./Dec. 2004.
[5] R. M. Tallam, R. J. Kerkman, D. Leggate, and R. A. Lukaszewski, Po-Tai Cheng (S’96–M’99–SM’09) received the
“Common-mode voltage reduction PWM algorithm for ac drives,” IEEE Ph.D. degree from the University of Wisconsin,
Trans. Ind. Appl., vol. 46, no. 5, pp. 1959–1969, Sep./Oct. 2010. Madison, in 1999.
[6] A. M. Hava and E. Un, “Performance analysis of reduced common- He is currently a Professor in the Department of
mode voltage PWM methods and comparison with standard pwm methods Electrical Engineering, National Tsing Hua Univer-
for three-phase voltage source inverters,” IEEE Trans. Power Electron., sity, Hsinchu, Taiwan. His research interests include
vol. 24, no. 1, pp. 241–252, Jan. 2009. microgrid and smart grid issues, power electron-
[7] M. Cacciato, A. Consoli, G. Scarcella, and A. Testa, “Reduction of ics for distributed energy resources, and high-power
common-mode currents in PWM inverter motor drives,” IEEE Trans. converters.
Ind. Appl., vol. 35, no. 2, pp. 469–475, Mar./Apr. 1999.
[8] M. Cavalcanti, K. C. D. Oliveria, A. M. de Farias, F. A. S. Neves, G.
M. S. Azevedo, and F. C. Camboim, “Modulation techniques to eliminate
leakage currents in transformerless three-phase photovoltaic systems,”
IEEE Trans. Ind. Electron., vol. 57, no. 4, pp. 1360–1368, Apr. 2010.
[9] E. Un and A. M. Hava, “A near-state PWM method with reduced switching
losses and reduced common-mode voltage for three-phase voltage source
inverters,” IEEE Trans. Ind. Appl., vol. 45, no. 2, pp. 782–793, Mar./Apr. Ahmet M. Hava (S’91–M’98) was born in Mardin,
2009. Turkey, in 1965. He received the B.S. degree from
[10] A. M. Hava, R. J. Kerkman, and T. A. Lipo, “Simple analytical and Istanbul Technical University, Istanbul, Turkey, in
graphical methods for carrier-based PWM–VSI drives,” IEEE Trans. 1987, and the M.Sc. and Ph.D. degrees from the Uni-
Power Electron., vol. 14, no. 1, pp. 49–61, Jan. 1999. versity of Wisconsin, Madison, in 1991 and 1998,
[11] L. J. Garces, “Current control of field oriented ac induction motor drives,” respectively, all in electrical engineering.
IEEE Tut.: Microprocessor Control Motor Drives Power Converters, In 1995, he was with Rockwell Automation-
vol. 1, pp. 5.1–5.46, 1993. Allen Bradley Company, Mequon, WI. From 1997
[12] M. Depenbrock, “Pulse width control of a 3-phase inverter with non- to 2002, he was with Yaskawa Electric America,
sinusoidal phase voltages,” in Proc. IEEE ISPC Conf. Rec., 1977, vol. 1, Inc., Waukegan, IL. Since 2002, he has been with
pp. 399–403. the Electrical and Electronics Engineering Depart-
[13] E. Un and A. M. Hava, “Performance characteristics of the reduced com- ment, Middle East Technical University, Ankara, Turkey, where he is currently
mon mode voltage near state pwm method,” EPE J., vol. 19, no. 3, an Associate Professor. His research interests include power electronics, motor
pp. 41–49, Sep. 2009. drives, and power quality.