Modulation Strategies for Three-Phase AC-DC

Matrix Converters: a Comparison

M. Mengoni, L. Zarri, A. Tani, G. Rizzoli, G. Serra, D. Casadei

Dept. of Electric, Electronic and Information Engineering "G. Marconi"
University of Bologna
Bologna, Italy
michele.mengoni@unibo.it

Abstract— The matrix converter rectifier is an AC-to-DC assumptions. In [6]-[7] an indirect AC-AC converter,
converter that can be considered for emerging applications, composed of two stages without intermediate energy storage
related to smart grids and renewable energy production. The elements, was proposed. The initial stage of this indirect ac-ac
most attractive characteristics of this converter are the matrix converter can be regarded as an ac-dc converter,
simultaneous control of the input power factor and the output although it is not analyzed as a stand-alone converter [8]-[9].
voltage, and the small volume of the input capacitors. In this
paper, after a small review of the modulation techniques, several In [10] a battery charger based on an ultrasparse AC-DC
modulation strategies are experimentally compared in terms of matrix converter is described, which cannot revert the power
power losses and THD of input and output currents. flow due to the reduced number of power switches.

Keywords—matrix converters, modulation strategies, rectifiers, In [11]-[12] an ac-dc matrix converter, which is obtained
power factor correction by simply removing one leg from a three-phase direct AC-AC
matrix converter, is presented and a space vector modulation is
proposed. This topology has the advantage that the modulation
I. INTRODUCTION techniques already developed for AC-AC matrix converters
Low costs and robustness are among the reasons why three- can be easily adapted. In fact, the modulation strategy showing
phase passive rectifiers have spread in all fields of industrial the minimum switching losses is identified in [13], and an
applications for AC-to-DC energy conversion. However, analysis of the converter losses is presented in [14]. Finally in
despite their innumerable advantages, the harmonic distortion [15] some closed-loop control schemes have been proposed to
of the output voltage and the input currents can be control the input power factor and improve the quality of the
unacceptably high without the filtering effect of bulky reactive input/output currents.
elements, such as LC networks or clamping capacitors. In
addition, diode rectifiers do not allow regulating the output The aim of the present paper is an experimental comparison
voltage, while the input power factor of thyristor-based of the performance of several modulation strategies in terms of
rectifiers decreases as long as the output voltage becomes THD of input and output currents, and efficiency.
smaller. In both cases, the power flow cannot be reversed
simply by changing the output current direction, so passive II. MODULATION STRATEGIES FOR AC-DC MATRIX
rectifiers cannot be used in applications that require CONVERTERS
bidirectional power flow to and from the grid [1]-[2]. In their The analysis of the modulation strategies for AC-DC
turn, a solution that can be adopted is the use of pulse-width- matrix converters is developed in detail in [13] and is
modulated voltage- and current-source rectifiers, which can summarized hereafter for the sake of clarity.
control the input power factor and produce low distortion input
currents [3]-[4]. However, the DC-link voltage of these
A. Input-Output Equations of AC-DC Matrix Converters
converters is necessarily greater than the peak value of the
phase-to-phase input voltage, so an additional step-down With reference to Fig. 1, let us define the duty-cycles mh,k
power stage may be necessary to feed low-voltage loads. (h=1, 2 and k=1, 2, 3) of the switches Sh,k over a switching
period Ts. Then, the mean value over a switching period of the
The three-phase AC-DC matrix converter can potentially input currents ii,k (k=1, 2, 3) and the output pole voltages vo,h
overcome this restriction while keeping many features of PWM (h=1, 2) can be written as follows:
active rectifiers [5]. In fact, in AC-DC matrix converters, while
the output voltage can be decreased to zero, the input power ii , k m 1, k  m2, k  io (k=1, 2, 3) (1)
factor can be adjusted, the power flow can be reversed and the
3
harmonic content of input and output currents is low.
vo , h ¦m v
h,k i , k (h=1, 2) (2)
The performance of the AC-DC matrix converter has k 1

already been analyzed in some papers under particular where io is the output current, and vi,k (k=1, 2, 3) are the input

978-1-5090-0737-0/16/$31.00 ©2016 IEEE

 ac

ac

ac

vgrid Rs Ls Lf S1,1 S1,2 S1,3 S2,1 S2,2 S2,3

Non ideal voltage supply

Cf
vo
Input filter
io El Rl Ll

Load

Fig. 1 – Basic scheme of AC/DC conversion system.

pole voltages. 3
vo vo ,1  vo , 2 m d ˜ vi . (10)
To represent the state of the converter in a compact way, it 2
is possible to combine the duty-cycles to form complex The modulation problem consists in controlling the
quantities, m1 and m2 , defined as follows: inverter switches so that the mean value of the output voltage
over a switching period is equal to the desired value, vo,ref, and
2 3 the input current vector has the desired direction, represented
mh ¦ mh,kD k (h=1,2) (3)
3k1 by the unity vector \ ref .
being D k (k=1, 2, 3) a unity vector defined as vo vo ,ref (11)
2S
j ( k 1)
Dk e 3
. (4) ii ˜ j\ ref 0. (12)

Since the sum of mh,k over index k has to be equal to 1 to It is worth noting that (12) states that \ ref and ii have the
avoid short circuits and overvoltages, the duty-cycles mh,k can same direction, not necessarily the same verse, which is a
be calculated from the space vectors m1 and m2 by using the consequence of the direction of the power flow.
following inversion equations:
1 B. Pulse-Width Modulation
mh , k  mh ˜ D k (h=1, 2, k=1, 2 ,3) (5) Substituting (7) and (10) in (11)-(12) and solving for md
3
leads to the following expression.
where "˜" is the dot product operator, defined as the real part
of the product between the first operand and the complex 2 vo , ref \ ref
conjugate of the second. md , ref . (13)
3 vi ˜ \ ref
The definition of the space vector of the input currents is
The knowledge of md is not sufficient to find the
2 3 quantities m1 and m2 , so it is necessary to introduce a new
ii ¦ ii ,kD k .
3k1
(6)
variable, m0 , defined as follows:
By combining (1), (3) and (6), one comes to the following
m1  m2
expression: m0 . (14)
2
ii m d io (7)
The variable m0 , usually referred to as "zero-sequence
where the quantity md is component", is a complex number, equivalent to two degrees
of freedom, that can be used to modify the properties of the
md m1  m2 . (8) modulation strategy.
In a similar manner, if the definition of the input voltage Solving (8) and (14) for m1 and m2 , and considering the
vector is explicit expression of md in (13), it is possible to calculate the
2 3 duty-cycles mh,k (h=1,2 and k=1,2,3):
vi ¦ vi , k D k (9)
3k1 1 (1)
h 1
\ref ˜ Dk
mh , k  vo , ref  m0 ˜ Dk (h=1,2, k=1,2,3). (15)
one finds that the output voltage can be expressed as a 3 3 \ref ˜ vi
function of md by combining (2), (5), (8) and (9):
In [13] it is shown that the upper bound of the voltage
transfer ratio is as follows:
 TABLE I - CONFIGURATIONS OF THE AC-DC MATRIX CONVERTER
vo , ref 3
d cos M i . (16)
vi 2 Conf. Switches
m1 m2 md mo
#
If (16) is satisfied, the ac-dc matrix converter can operate
S1,1 on 2 2 2 1
in the linear modulation range and the modulation problem 1
S2,2 on
D1 D2 D1  D 2  D1  D 2 
3 3 3 3
has admissible solutions. This means that there exits at least
one value of the zero sequence component m0 that makes all S1,1 on 2 2 2 1
2
S2,3 on
D1 D3 D1  D 3  D1  D 3 
duty-cycles remain in the interval [0,1]. Different values of 3 3 3 3
m0 lead to different values of the switch duty-cycles, i.e., S1,2 on 2 2 2 1
different modulation strategies. 3
S2,3 on
D2 D3 D 2  D 3  D 2  D 3 
3 3 3 3

III. SPACE VECTOR MODULATION S1,2 on 2 2 2 1

4
S2,1 on
D2 D1 D 2  D1  D1  D 2 
3 3 3 3
According to the principle of space vector modulation, the
mean value of the output voltage and the input current vector S1,3 on 2 2 2 1
over a switching period has to be approximated by a sequence 5
S2,1 on
D3 D1 D 3  D1  D1  D 3 
3 3 3 3
of different converter configurations.
Tab. I shows the possible states of the AC-DC matrix S1,3 on 2 2 2 1
6
S2,2 on D3 D2 D 3  D 2  D 2  D 3 
converter. The first column contains the identification number 3 3 3 3
of each configuration, the second column shows the switches
that are turned on, the third and fourth columns show the S1,1 on 2 2 2
01
S2,1 on
D1 D1 0 D1
instantaneous values of m1 and m2 . Finally, the remaining 3 3 3

columns are the values of md and m0 , calculated by means of S1,2 on 2 2 2

02
S2,2 on
D2 D2 0 D2
(8) and (14). As can be seen, there are six active 3 3 3
configurations, which produce a non-zero output voltage, and 2 2 2
S1,3 on
three zero configurations. 03
S2,3 on
D3 D3 0 D3
3 3 3
Fig. 2 shows the admissible values of md , which divide
the plane in six sectors.
freedom are introduced. In fact, the three duty-cycles G0,1, G0,2
Since the value of md , ref is known by means of (13), it is and G0,3 of the zero configurations 01, 02 and 03 have to be
possible to identify the two vectors, mdL and mdR , that delimit chosen under the constraint that the sum of all duty-cycles is
equal to 1:
the sector where md , ref is located. For example, if md , ref is in
 G0,1 + G0,2 + G0,3 + G L + G R = 1. (20)
2 2
sector 1, m and m may be D1  D 2  and D1  D3 
d
L
d
R

3 3 Tab. II shows the sequence of the vectors that have to be

(the order is not important). used in SVM, as a function of the sector number. To build this
table it is sufficient to note that there are only two sequences
The mean value of md over a switching period can be of converter states, being one the inverse of the other, that
written as a linear combination of mdL and mdR : minimize the number of switch commutations. These two
sequences are identified in Tab. II by the configuration
md , ref G L mdL  G R mdR . (17) patterns c1-c2-c3-c4-c5 and c5-c4-c3-c2-c1. Both sequences are
composed of three zero vectors interspersed with two active
where G L and G R are numbers in the interval [0,1], which
have to be interpreted as duty-cycles. 3
2
Solving (17) for G L and G R leads to the following sector 3
D 1  D 3 

expressions: 4 2
e d
3 m d , ref
GL  md , ref ˜ jmdR (18) f c
2
g h
3
GR md , ref ˜ jmdL . (19) 5 1
2
2
D 1  D 2 
2 3
Once the application times of mdL and mdR are known, it is 3
3
6
necessary to fill the remaining part of the switching period
with the three zero vectors. In this way, two degrees of Fig. 2 – Admissible instantaneous values of md and repartition of the plane
into six sectors.
 TABLE II - SEQUENCE OF VECTORS FOR SVM When the duty-cycles of the zero vectors are equal, the
(THE REVERSE ORDER IS ALSO POSSIBLE) modulation strategy is referred to as SVM 3Z. Similarly, it is
possible to define other modulation strategies, such as SVM
Vector sequence
2Z and SVM 1Z, that use a lower number of zero
md 0 mdL 0 mdR 0 configurations, as shown in Table III. The letters "L", "C" and
Duty- "R" adopted in the names of the modulation strategies identify
G 0L GL G C0 GR G 0R the position of the zero vectors used in the switching pattern,
cycles
Configurations c1-c5 of the sequence i.e., "Left", "Center", and "Right".
Sector
c1 c2 c3 c4 c5 A. Equivalence of SVM and PWM
number
1 02 1 01 2 03 Whereas the principles of PWM lead to the calculation of
2 01 2 03 3 02 the duty-cycles of all converter switches, SVM selects a
3 03 3 02 4 01 sequence of vectors and calculates the duty-cycle of each
4 02 4 01 5 03 vector. Once this sequence of vectors has been chosen, it is
5 01 5 02 6 02 possible to find a relationship linking the switching patterns
6 03 6 03 1 01 resulting from the two modulation approaches.
The duty-cycles of PWM is univocally determined by (15)
Single-sided switching sequence
as soon as m0 is chosen. Similarly, the switching pattern of
SVM is univocally determined if the allocation of the zero
S1,2=1 S1,1=1 S1,3=1 vectors has been decided. Therefore, to prove the equivalence
Output
phases of the two modulation methods, it is sufficient to verify that
S2,2=1 S2,1=1 S2,3=1 there is a one-to-one relationship between m0 and the duty-
Config.
sequence      cycles G0,1, G0,2 and G0,3 of the zero configurations.
Duty
G 0L G 0, 2 GL G C0 G 0,1 GR G 0R G 0, 3
cycles
vi,1- vi,1- By analyzing the switching pattern in Fig. 3, it turns out
vo 0 vi,2 0 vi,3 0 that G0,k can be written as follows:
Fig. 3 – Possible single-sided switching pattern with 3 zero-voltage G0, k min^ m1, k , m2, k ` (k=1,2,3). (21)
configurations (SVM 3Z) when the vector md lies in sector 1.
Combining (13), (15) and (21), one comes to a set of
equations linking m0 and G0,1, G0,2 and G0,3:
configurations. It can be noted that configuration c3 is the
same vector in both patterns and is related to the sector 1 1
number of md , ref . Since md , ref has the same direction of the G 0, k  md , ref ˜ D k  m0 ˜ D k (k=1,2,3). (22)
3 2
input current vector, it turns out that c3 can be 01, 02 or 03
depending on which phase is conducting the maximum Under the constraint (20), only two duty-cycles among
instantaneous absolute value of the current. For example, if G0,1, G0,2 and G0,3 are linearly independent, so (22) can be
the current space vector lies in sector 1, among all phases, the solved for m0 , which has this following closed-form
highest current flows in phase 1, hence configuration c3 is 01 expression:
in Tab. II.
2 3 § 1 ·
For the sake of completeness, Fig. 3 shows the typical m0 ¦ ¨ G 0, k  md , ref ˜ D k ¸ D k . (23)
switching pattern of converter when md , ref is in sector 1. In 3 k 1© 2 ¹
this generic example all three zero configurations are used. By using (13), one finds the following explicit expression
for (22) and (23)
TABLE III - SPACE VECTOR MODULATION STRATEGIES
1 1 vo , ref D k ˜ \ ref
Modulation Zero configurations of G 0, k   m0 ˜ D k (k=1,2,3) (24)
Constraints 3 3 vi ˜ \ ref
Strategy the sequence
SVM 3Z G 0L G C0 G 0R c1, c3, c5 used
2 3 §¨ 1 vo , ref D k ˜ \ ref ·
¸D .
SVM 2Zlc G L C
G ,G R
0 c1, c3 used m0 ¦ G 0, k  3 v ˜ \ (25)
0 0 0
3 k 1 ¨© i ref
¸ k
¹
SVM 2Zlr G 0L G 0R , GC0 0 c1, c5 used
C R L
Equations (24) show that the duty-cycles G0,1, G0,2 and G0,3
SVM 2Zrc G G ,G 0 c3, c5 used
0 0 0
can be derived from m0 , whereas (25) shows that the zero-
SVM 1Zl GC0 G 0R 0 c1 used sequence component m0 can be univocally calculated from
SVM 1Zc G 0L G 0R 0 c3 used the duty-cycles of the zero configurations.
SVM 1Zr G L
0 G C
0 0 c5 used However, it is worth noting that two modulation strategies
having the same value of m0 can differ from each other in the
switching pattern, i.e., in the ordering of the turn-on and turn-
off sequence of the switches. Since two sequences of
configurations are equally admissible for SVM, i.e., c1-c2-c3-
c4-c5 or c5-c4-c3-c2-c1, it is necessary to eliminate this
ambiguity.
In the following, it is assumed that the vector sequence of
SVM is selected in such a way that at the beginning of the
period each output phase is connected to the input phase
having the intermediate absolute value among the input
currents, then the highest absolute value and finally the lowest
absolute value.
Fig. 4. Prototype of ac-ac matrix converter, adapted as ac-dc matrix
This choice allows finding an algorithm linking the duty- converter.
cycles (G0,1, G0,2, G0,3) to the duty-cycles ( G 0L , G C0 , G 0R ) that
TABLE IV - PARAMETERS OF THE EXPERIMENTAL SET-UP
characterize the modulation strategy. To solve this problem,
let us define the triplet (t, m, b) as a permutation of the indexes Supply Input Filter Load
(1, 2, 3) of the input phases so that the absolute values of the Vi = 220 VLL,rms, Zs = 2S50 rad/s, Lf = 1.1 mH Rl = 50 :
input currents are sorted in decreasing order, i.e., RS = 0.2 :LS = 0.80 mH Cf = 25 PF (Y) Ll = 0.010 H

it t im t ib . (26)

Then, G0,k (k=1,2,3) can be easily calculated as follows: characterized by the absence of a zero configuration in the
switching sequence. This aspect is visible in Fig. 6, since the
­G 0L if k m values of the duty-cycles m1,1 and m2,1 periodically become
° C
G 0, k ®G 0 if k t (k=1, 2, 3) . (27) zero.
° R This phenomenon is emphasized for modulations strategies
¯G 0 if k b
of type SVM 1Z, which use only one zero configuration. The
Equation (27) makes the use of (25) unambiguous for all duty-cycles shown in Fig. 7 periodically become equal to zero,
modulation strategies listed in Tab. II. or even to one. In this latter case, as long as a duty-cycle
remain equal to one, all switches of the same output phase do
IV. EXPERIMENTAL RESULTS not commutate.
Some experimental tests have been carried out to compare Although the converter switching losses are influenced by
the performance of the modulation strategies of Tab. II. The the number of zero vectors used in the modulation strategy, it
performance of these modulation strategies is different, due to is not possible to predict the efficiency performance of the
the different switching patterns. converter on this basis only, since the power losses depend on
the converter current and the switching pattern.
The basic scheme of the experimental setup is the same of
Fig. 1. The converter, shown in Fig. 5, is based on the IGBT The modulation strategies listed in Tab. III have been
module FM35E12KR3. The control algorithm is implemented compared in terms of efficiency, and THD of input and output
on a fixed-point digital signal processor produced by Texas currents. The results of the comparison are plotted as functions
Instruments (model TMSF2812). The switching frequency is of the voltage transfer ratio in Fig. 8.
10 kHz, and the four-step commutation is used. The
parameters of the experimental set-up are shown in Table IV.
Although for the sake of simplicity the theory of
m1,1
modulation has been developed in Section III with reference m2,1
to single-sided switching patterns, in practical
implementations the double-sided switching patterns are
preferable, because their inherent symmetry leads to better
performance.
io
vo
Figs. 5, 6 and 7 show the waveforms of the duty-cycles
m1,1, m2,1, the output current and voltage when the voltage
transfer ratio is 1.5 for strategies of type SVM 3Z, 2Z and 1Z.
SVM 3Z is appreciable for its symmetry, but it uses three
zero configurations, thus increasing the number of switch
commutations per period. Fig. 5. Behavior of the ac-dc matrix converter when the input power factor is
1, the voltage transfer ratio is 1.5 and the modulation strategy is SVM 3Z.
The modulation strategies of type SVM 2Z are Load current (1A/div). Output voltage (200 V/div).
 m1,1 m2,1 m1,1 m2,1

io io
vo vo

a) a)

m1,1 m2,1 m1,1 m1,2

io io
vo vo

b) b)

m1,1 m2,1 m1,1 m1,2

io io
vo vo

c) c)

Fig. 6. Behavior of the ac-dc matrix converter when the input power factor is Fig. 7. Behavior of the ac-dc matrix converter when the input power factor is
1, the voltage transfer ratio is 1.5 and the modulation strategy is SVM 2Z. 1, the voltage transfer ratio is 1.5 and the modulation strategy is SVM 1Z.
Load current (1A/div). Output voltage (200 V/div). a) SVM 2Zlc. b) SVM Load current (1A/div). Output voltage (200 V/div). a) SVM 1Zl. b) SVM 1Zc.
2Zlr. c) SVM 2Zrc. c) SVM 1Zr.

The THD of the currents tends to increase as long as the ac-dc matrix converters.
voltage transfer ratio decreases. As can be seen, strategy SVM
2Zlr turns out to be the one with the best performance in terms The experimental results show that some modulation
of THD of the output currents, whereas SVM1Zl is the worst. strategies are preferable to others depending on the voltage
transfer ratio and on the adopted quality index. In particular,
However SVM1Zl offers the highest efficiency, whereas strategy SVM2lc minimizes the THD of the output current,
the less efficient one is SVM3Z. whereas 1Zl maximizes the converter efficiency.
As regards the THD of the input currents, the differences
of the modulation strategies do not result appreciable in the REFERENCES
experimental tests. [1] H. Mao, C. Y. Lee, D. Boroyevich, S. Hiti, “Review of high
performance three-phase power-factor correction circuits, ” IEEE Trans.
on Ind. Electron., vol. 44, issue 4, Aug. 1997, pp. 437-446.
V. CONCLUSION [2] J. W. Kolar and T. Friedli, “The essence of three-phase PFC rectifier
This paper establishes a connection between SVM and systems,” in Proc. Telecommunications Energy Conference (INTELEC),
PWM for AC-DC matrix converters and compares the 2011, IEEE 33rd International, Amsterdam, Oct. 2011, pp. 1-27.
performance of seven modulation strategies for three-phase [3] M. Hartmann, H. Ertl, J. W. Kolar, “Current Control of Three-Phase
Rectifier Systems Using Three Independent Current Controllers, ” IEEE
 3 Trans. on Pow. Electr., Vol. 28, No. 8, Aug. 2013, pp. 3988.
THD Output Current [%]. 3Z
[4] B. Singh, B. N. Singh, A. Chandra, K. Al-Haddad, A. Pandey and D. P.
2.5 2Zrc Kothari, “A review of three-phase improved power quality AC–DC
2 2Zlc converters,” IEEE Trans. Ind. Electron., vol. 51, no. 3, pp. 641-660, June
2004.
1.5 2Zlr [5] D. G. Holmes, Thomas A. Lipo, “Implementation of a controlled
1Zl rectifier using AC-AC matrix converter theory,” IEEE Trans. on Pow.
1 Electron., Vol. 7, No. 1, Jan. 1992, pp. 240-250.
1Zc [6] J. Kolar, M. Baumann, F. Schafmeister, H. Ertl, “Novel three-phase ac-
0.5
1Zr dc-ac sparse matrix converter—Part I: Derivation, basic principle of
0 operation, space vector modulation, dimensioning,” in APEC 2002.
0.5 0.75 1 1.25 1.5 Conf. Expo., Dallas, TX, Mar. 2002, vol. 2, pp. 777–787.
[7] L.Wei, T. A. Lipo, H.Chan, “Matrix converter topologies with reduced
Voltage Transfer Ratio a) number of switches,” in Power Electron. Spec. Conf., Jun. 2002, vol. 1,
1.7 3Z pp. 57–63.
THD Input Current [%]
.

[8] R. Pena, R. Cardenas, E. Reyes, J. Clare, and P.Wheeler, “Control

1.5 2Zrc strategy of an indirect matrix converter with modifying DC voltage,” in
1.3 2Zlc Proc. Eur. Power Electron. Appl., 2009, pp. 1–8.
2Zlr [9] Y. Sun, M. Su, G. Weihua, H. Wang, and X. Li, “Indirect four-leg
1.1
matrix converter based on robust adaptive back-stepping control,” IEEE
0.9 1Zl Trans. on Industrial Electronics, vol. 58, no. 9, pp. 4288–4298, Sep.
2011.
0.7 1Zc
[10] R. Metidji, B. Metidji, B. Mendil, “Design and implementation of a
0.5 1Zr unity power factor fuzzy battery charger using an ultrasparse matrix
rectifier,” IEEE Trans. on Power Electronics, Vol. 28, No. 5, May 2013,
0.5 0.75 1 1.25 1.5 pp. 2269-2276.
Voltage Transfer Ratio b) [11] K. You, D. Xiao, M. F. Rahman, M. Nasir Uddin, “Applying reduced
c) general direct space vector modulation approach of AC-AC matrix
1 converter theory to achieve direct power factor controlled three-phase
3Z AC-DC matrix rectifier,” IEEE Trans. on Industry Application. Vol. 50,
0.95 No. 3, May-June 2014, pp. 2243-2257.
2Zrc
0.9 [12] P. Dost, K. Korotkiewicz, C. Sourkounis, “A modified modulation
2Zlc
Efficiency

0.85 strategy for an active rectifier stage structurally based on the topology of
2Zlr an indirect matrix converter,” IECON 2014, Dallas, TX, Oct. 29-Nov.
0.8 04, 2014, pp. 4110-4114.
0.75 1Zl [13] M. Mengoni, L. Zarri, A. Tani, G. Rini, G. Serra, D. Casadei,
1Zc “Modulation strategy with minimum switching losses for three-phase
0.7
AC-DC matrix converters,” 16th European Conference on Power
0.65 1Zr Electronics and Applications (EPE'14-ECCE Europe), 2014, pp. 1-10.
1.5
0.5 1.25
0.75 1 0.75
1.25 0.5
1.5 [14] K. You, D. Xiao, M. F. Rahman, “Measurement-based analytical model
of conduction and switching losses of three-phase AC-DC matrix
Voltage Transfer Ratio
rectifier,” ECCE 2012, Raleigh, NC, Sept. 15-20, 2012, pp. 302-309.
Fig .8 - Behavior of the modulation strategies of Tab. III as a function of the
voltage transfer ratio. a) THD of the output current. b) THD of the input [15] Z. Wang, Y. Xie, Y. Wang, “Feedback control strategy for matrix
current. c) Efficiency. rectifier,” IPEMC 2012, Harbin, China, June 2-5, 2012, pp. 2002-2006.