Thesis Marco Rivera
-
Upload
marco-rivera -
Category
Documents
-
view
73 -
download
3
Transcript of Thesis Marco Rivera
Universidad Tecnica Federico Santa MarıaDepartament of Electronics Engineering
Valparaıso, Chile
Doctoral Dissertation
Predictive Control in an Indirect Matrix Converter
Marco Esteban Rivera Abarca
Doctorate Program
Doctorate in Electronic Engineering
Thesis Supervisor
Dr. Jose Rodrıguez Perez - UTFSM
Evaluation Committee
Dr. Cesar Silva Jimenez - UTFSM
Dr. Jose Espinoza Castro - Universidad de Concepcion
November 2011
Dedicado a las personas que mas amo, ustedes le dan sentido a mi vida.
Contents
Contents i
List of Figures v
Nomenclature xiii
Abstract xvi
Resumen xviii
Acknowledgments xx
Agradecimientos xxi
1 Introduction 1
1.1 State of the art review . . . . . . . . . . . . . . . . . . . . . . 3
1.2 Hypothesis and contribution of this thesis . . . . . . . . . . . 5
1.3 Chapter review . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2 Review of three-phase ac/ac topologies 7
2.1 Classification of ac/ac power converters . . . . . . . . . . . . . 7
2.2 Topologies with dc-link . . . . . . . . . . . . . . . . . . . . . . 9
2.2.1 Back-to-back converter . . . . . . . . . . . . . . . 9
2.3 Topologies without dc-link . . . . . . . . . . . . . . . . . . . . 10
2.3.1 Cycloconverter . . . . . . . . . . . . . . . . . . . . 10
2.3.2 Direct matrix converter . . . . . . . . . . . . . . . 11
ii
Contents iii
2.3.3 Indirect matrix converter . . . . . . . . . . . . . . 13
2.3.4 Three-phase ac/ac buck converters . . . . . . . . . 14
2.3.5 Sparse indirect matrix converters . . . . . . . . . . 15
2.3.6 Indirect three-level matrix converter . . . . . . . . 17
2.4 Hybrid topologies . . . . . . . . . . . . . . . . . . . . . . . . . 18
3 A review of modulation and control methods for
matrix converters 20
3.1 Scalar techniques . . . . . . . . . . . . . . . . . . . . . . . . . 21
3.1.1 Direct method: Venturini . . . . . . . . . . . . . . 21
3.1.2 Roy’s method . . . . . . . . . . . . . . . . . . . . . 25
3.1.3 Current phase displacement control for Venturini
and Roy methods. . . . . . . . . . . . . . . . . . . 26
3.2 Pulse width modulation methods . . . . . . . . . . . . . . . . 27
3.2.1 Carrier-based modulation method . . . . . . . . . 27
3.2.2 Space vector modulation (SVM) method . . . . . 29
3.3 Direct torque control . . . . . . . . . . . . . . . . . . . . . . . 31
3.4 Predictive control . . . . . . . . . . . . . . . . . . . . . . . . . 32
3.4.1 Predictive current control . . . . . . . . . . . . . . 32
3.4.2 Predictive torque control . . . . . . . . . . . . . . 36
3.5 Assessment of the methods . . . . . . . . . . . . . . . . . . . . 42
3.6 Comments and conclusion . . . . . . . . . . . . . . . . . . . . 44
4 The indirect matrix converter 45
4.1 Description of the topology . . . . . . . . . . . . . . . . . . . . 45
4.2 Zero dc-link current commutation . . . . . . . . . . . . . . . . 47
4.3 PWM based control method . . . . . . . . . . . . . . . . . . . 48
4.3.1 Modulation of the rectifier stage . . . . . . . . . . 48
4.3.2 Modulation of the inverter stage . . . . . . . . . . 52
4.3.3 Simulations results . . . . . . . . . . . . . . . . . . 55
5 Model-based predictive control in an IMC 59
Contents iv
6 Predictive current control with reactive power minimization 64
6.1 Control scheme for the IMC . . . . . . . . . . . . . . . . . . . 64
6.2 Input filter and load discrete equations . . . . . . . . . . . . . 65
6.3 Cost function definition . . . . . . . . . . . . . . . . . . . . . . 66
6.4 Discrete time delay error compensation . . . . . . . . . . . . . 67
6.5 Simulation results . . . . . . . . . . . . . . . . . . . . . . . . . 68
6.6 Experimental results . . . . . . . . . . . . . . . . . . . . . . . 72
6.7 The problem in the source current with a weak ac-supply . . . 76
7 Current control for an IMC with input filter resonance mitiga-
tion 77
7.1 Current control scheme for the IMC with active damping ap-
proach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
7.1.1 Active damping approach and implementation . . 78
7.2 Experimental results . . . . . . . . . . . . . . . . . . . . . . . 79
7.3 Comments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
8 Imposed sinusoidal source and load currents for an IMC 87
8.1 Fundaments . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
8.2 The problem on the input side . . . . . . . . . . . . . . . . . . 88
8.3 Predictive current control for the IMC with imposed sinusoidal
source currents . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
8.3.1 Prediction model . . . . . . . . . . . . . . . . . . . 90
8.3.2 Cost function definition . . . . . . . . . . . . . . . 92
8.4 Generation of the source current reference is∗ . . . . . . . . . 92
8.5 Simulation and experimental results . . . . . . . . . . . . . . . 94
8.6 Comments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
9 Conclusions 101
Appendix 103
9.1 Publications in journals . . . . . . . . . . . . . . . . . . . . . . 103
9.2 Publications in conferences . . . . . . . . . . . . . . . . . . . . 104
Contents v
9.3 Projects related with the research . . . . . . . . . . . . . . . . 107
9.4 Experimental setup circuit diagram . . . . . . . . . . . . . . . 108
Bibliography 120
List of Figures
2.1 Classification of ac/ac power converters. . . . . . . . . . . . . . . 8
2.2 Three-phase ac/ac converter topologies with dc-link energy stora-
ge; (a) voltage dc-link back-to-back converter; (b) current dc-link
back-to-back converter. . . . . . . . . . . . . . . . . . . . . . . . 10
2.3 Three-phase ac/ac converter topology without dc-link energy sto-
rage: cycloconverter. . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.4 Three-phase ac/ac converter topology without dc-link energy sto-
rage: direct matrix converter (DMC). . . . . . . . . . . . . . . . 12
2.5 Three-phase ac/ac converter topology without dc-link energy sto-
rage: indirect matrix converter (IMC). . . . . . . . . . . . . . . . 13
2.6 Three-phase ac/ac converter topology without dc-link energy sto-
rage: three-phase buck converter. . . . . . . . . . . . . . . . . . . 14
2.7 Three-phase ac/ac converter topologies without dc-link energy
storage; (a) sparse matrix converter (SMC); (b) very sparse ma-
trix converter (VSMC); (c) ultra sparse matrix converter (USMC). 16
2.8 Indirect matrix converter with an additional bridge-leg that al-
lows the mains phase voltages to be switched directly or inverter
into the link. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.9 Circuit topology of the hybrid direct matrix converter and the
implementation of a single switching cell. A cascading of several
H-bridges in each connection of an input and output is also possible. 18
2.10 Circuit topology of the hybrid indirect matrix converter with a
series voltage source in the link. . . . . . . . . . . . . . . . . . . 19
vi
List of Figures vii
3.1 Summary of modulation and control methods for matrix converters. 21
3.2 Direct method: venturini, typical waveforms; a) output voltage
vaN [pu], its reference (bold line) and b) output current ia [pu]. . 24
3.3 Unipolar sinusoidal PWM method and desired output level voltage. 28
3.4 Carrier-based method, typical waveforms; a) line-to-line output
voltage vab [pu]; b) output current ia [pu]. . . . . . . . . . . . . . 29
3.5 Direct torque control scheme. . . . . . . . . . . . . . . . . . . . . 32
3.6 Predictive current control scheme. . . . . . . . . . . . . . . . . . 33
3.7 Predictive current control without power factor correction A =
0; a) output current [A]; (b) reactive power [kVAR]; (c) input
current [A] and input voltage [V/30]. . . . . . . . . . . . . . . . 35
3.8 Predictive current control with power factor correction A = 1; a)
output current [A]; (b) reactive power [kVAR]; (c) input current
[A] and input voltage [V/30]. . . . . . . . . . . . . . . . . . . . . 36
3.9 Predictive torque control scheme. . . . . . . . . . . . . . . . . . . 37
3.10 Predictive control torque without power factor correction λq = 0;
a) speed [rad/s]; b) electrical torque [Nm]; c) output current [A];
d) stator flux [Wb]; e) reactive power [kVAR]; f) input current
[A] and input voltage [V/30]; g) zoom of input current isA and
input voltage vsA. . . . . . . . . . . . . . . . . . . . . . . . . . . 40
3.11 Predictive control torque with power factor correction λq = 1; a)
speed [rad/s]; b) electrical torque [Nm]; c) output current [A]; d)
stator flux [Wb]; e) reactive power [kVAR]; f) input current [A]
and input voltage [V/30]; g) zoom of input current isA and input
voltage vsA. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
4.1 Indirect matrix converter topology. . . . . . . . . . . . . . . . . . 46
4.2 (a) Line-to-line input voltages vi and dc-link voltage vdc; (b) valid
current vectors in the α − β plane; (c) first-maximum, second-
maximum and average dc-link voltage vdc. . . . . . . . . . . . . . 49
4.3 Position of the reference current vector in sector II; θi is the angle
respect to α-axis; θsi is the angle in the sector. . . . . . . . . . . 50
List of Figures viii
4.4 Formation of the dc-link voltage vdc and average dc-link voltage
vdc. As it can be observed, the dc-link voltage is given by the
first and second maximun line-to-line input voltages. . . . . . . . 52
4.5 (a) Available vectors on the inverter side; (b) position of the
output reference vector in sector II. . . . . . . . . . . . . . . . . 54
4.6 Formation of the dc-link voltage vdc and dc-link current idc within
a pulse period. Switching state changes of the input stage do
occur at zero dc-link current. . . . . . . . . . . . . . . . . . . . . 55
4.7 Simulation results SVM technique in open-loop control; (a) source
voltage vsA [V/10] and source current isA [A]; (b) output current
ia [A]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
4.8 Schematic diagram of the current control strategy added to the
modulation technique; the PI controller is in d-q coordinates. . . 56
4.9 Simulation results SVM technique in closed-loop control; (a) source
voltage vsA [V/10] and source current isA [A]; (b) output current
reference i∗a[A] and measured output current ia [A]. . . . . . . . 57
4.10 Simulation results SVM technique in closed-loop control; (a) source
voltage vsA [V/10] and source current isA [A]; (b) output current
reference i∗a[A] and measured output current ia [A]. . . . . . . . 58
5.1 FS-MPC generic algorithm. . . . . . . . . . . . . . . . . . . . . . 63
6.1 Predictive current control scheme. . . . . . . . . . . . . . . . . . 65
6.2 Simulation results without instantaneous reactive power mini-
mization; (a) source voltage vsA/10 [V] and current isA [A]; (b)
output current reference i∗a and measured ia [A]; (c) reactive
power qs [VA]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
6.3 Simulation results; (a) spectrum of source voltage [pu]; (b) spec-
trum of source current [pu]; (c) spectrum of output current [pu]. 70
List of Figures ix
6.4 Simulation results with instantaneous reactive power minimiza-
tion; (a) source voltage vsA/10 [V] and current isA [A]; (b) output
current reference i∗a and measured ia [A]; (c) reactive power qs
[VA]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
6.5 Simulation results; (a) spectrum of source voltage [pu]; (b) spec-
trum of source current [pu]; (c) spectrum of output current [pu]. 71
6.6 Experimental results without instantaneous reactive power mini-
mization; (a) source voltage vsA [50V/div] and current isA [5A/div];
(b) output current reference i∗a and measured ia [5A/div]. . . . . 73
6.7 Experimental results; (a) spectrum of source voltage [pu]; (b)
spectrum of source current [pu]; (c) spectrum of output current
[pu]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
6.8 Experimental results with instantaneous reactive power mini-
mization; (a) source voltage vsA [50V/div] and current isA [5A/div];
(b) output current reference i∗a and measured ia [5A/div]. . . . . 75
6.9 Experimental results; (a) spectrum of source voltage [pu]; (b)
spectrum of source current [pu]; (c) spectrum of output current
[pu]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
7.1 Predictive current control with active damping scheme. . . . . . 80
7.2 Experimental results without active damping approach; (a) source
voltage vsA [50V/div] and current isA [5A/div]; (b) output cur-
rent reference i∗a and measured ia [5A/div]. . . . . . . . . . . . . 81
7.3 Experimental results without active damping approach; (a) spec-
trum of source voltage [pu]; (b) spectrum of source current [pu];
(c) spectrum of output current [pu]. . . . . . . . . . . . . . . . . 81
7.4 Experimental results current control with active damping ap-
proach; a) source voltage vsA [50V/div] and current isA [5A/div];
b) output current ia and reference i∗a [5A/div]. . . . . . . . . . . 83
7.5 Experimental results with active damping approach; (a) spec-
trum of source voltage [pu]; (b) spectrum of source current [pu];
(c) spectrum of output current [pu]. . . . . . . . . . . . . . . . . 83
List of Figures x
7.6 Experimental results current control without and with active
damping approach; a) zoom spectrum of source voltage vsA [pu];
b) zoom spectrum of source current isA [pu]; c) zoom spectrum
of output current ia [pu]. . . . . . . . . . . . . . . . . . . . . . . 84
7.7 Experimental results current control without active damping ap-
proach; a) source voltage vsA [50V/div] and current isA [5A/div];
b) output current ia and reference i∗a [5A/div] with output fre-
quency reference equal to 100Hz. . . . . . . . . . . . . . . . . . . 85
7.8 Experimental results current control with active damping ap-
proach; a) source voltage vsA [50V/div] and current isA [5A/div];
b) output current ia and reference i∗a [5A/div] with output fre-
quency reference equal to 100Hz. . . . . . . . . . . . . . . . . . . 86
8.1 Predictive source and output current control scheme with source
current reference. . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
8.2 Simulation results predictive control with imposed sinusoidal source
and load currents; a) source voltage [V/25] and current [A]; b)
output current and reference [A]. . . . . . . . . . . . . . . . . . . 95
8.3 Experimental results predictive control with imposed sinusoidal
source and load currents; a) source voltage [50V/div] and current
[5A/div]; b) output current and reference [5A/div]. . . . . . . . . 95
8.4 Simulation results predictive control with imposed sinusoidal source
and load currents; a) source voltage [V/25] and current [A]; b)
output current and reference [A] with an output frequency refer-
ence of 100Hz. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
8.5 Experimental results predictive control with imposed sinusoidal
source and load currents; a) source voltage [50V/div] and cur-
rent [5A/div]; b) output current and reference [5A/div] with an
output frequency reference of 100Hz. . . . . . . . . . . . . . . . . 97
8.6 Simulation results predictive control with imposed sinusoidal source
and load currents; a) source voltage [V/25] and current [A] 30o
displacement angle; b) output current and reference [A]. . . . . . 98
List of Figures xi
8.7 Experimental results predictive control with imposed sinusoidal
source and load currents; a) source voltage [50V/div] and current
[5A/div]; b) output current and reference [5A/div]. . . . . . . . . 98
8.8 Simulation results predictive control with imposed sinusoidal source
and load currents; a) source voltage [V/25] and current [A] −30o
displacement angle; b) output current and reference [A]. . . . . . 99
8.9 Experimental results predictive control with imposed sinusoidal
source and load currents; a) source voltage [50V/div] and current
[5A/div]; b) output current and reference [5A/div]. . . . . . . . . 99
9.1 Experimental setup in the laboratory. . . . . . . . . . . . . . . . 109
9.2 Schematic of the input filter used in the implementation. . . . . 110
9.3 Schematic of the adapter between the dSPACE and FPGA. . . . 111
9.4 Schematic of the FPGA with its different voltage levels. . . . . . 112
9.5 Schematics of voltage regulators included in the FPGA. . . . . . 113
9.6 Schematic of the PROM included in the FPGA card. . . . . . . 114
9.7 Schematic of the led and DIP-switches included in the FPGA card.114
9.8 Schematic of an analog-digital converter CAD included in the
FPGA. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
9.9 Schematic of a digital-analog converter CDA included in the FPGA.115
9.10 Schematics of input/output components included in the FPGA. 116
9.11 Schematics of the power circuit in the IMC. . . . . . . . . . . . . 117
9.12 Schematic of the driver implemented in the IMC for the switches. 118
9.13 Schematic of the implementation of voltage measurement in the
IMC. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
xii
List of Figures xiii
Nomenclature
Abbreviations
ac Alternating current
dc Direct current
AFE Active front end
CSC Current source converter
DMC Direct matrix converter
FPGA Field-programmable gate array
HDMC Hybrid direct matrix converter
HIMC Hybrid indirect matrix converter
IGBT Insulated gate bipolar transistor
IMC Indirect matrix converter
MPC Model-based predictive control
NPC Neutral point clamped
PCC Predictive current control
PI Proportional/integral controller
PWM Pulse-width modulation
SMC Sparse matrix converter
SVM Space vector modulation
THD Total harmonic distortion
USMC Ultra sparse matrix converter
VSC Voltage source converter
VSMC Very sparse matrix converter
ADC Analog digital controller
DAC Digital analog controller
List of Figures xiv
Variables
A,B,C Input-phase designators
a, b, c Output-phase designators
α, β ℜ (real) and ℑ (imaginary) components
s Laplace variable (not as subscript)
g Cost of quality function
k Discrete time
t Continuous time
Cf Filter capacitor
Lf Filter inductor
Rf Filter resistor
RL Load resistance
LL Load inductance
Ts Sampling time
Superscripts
p Superscript for predicted variables
∗ Superscript for reference variables
Vectors
is Source current [isA isB isC ]T
vs Source voltage [vsA vsB vsC ]T
ii Input current [iA iB iC ]T
vi Input voltage [vA vB vC ]T
io Load current [ia ib ic]T
vo Load voltage [va vb vc]T
i∗s Source current reference [i∗sA i∗sB i∗sC ]T
i∗o Output current reference [i∗a i∗b i
∗c ]T
List of Figures xv
Scalars
vdc Dc-link voltage
idc Dc-link current
qs Input reactive power
q∗s Input reactive power reference
Abstract
The indirect matrix converter has been the subject to investigation for some time.
One of the favorable features of this converter is the absence of a dc-link capacitor,
which allows for the construction of compact converters capable of operating at ad-
verse atmospheric conditions such as extreme temperatures and pressures. These
features have been explored extensively and are the main reasons why the matrix
converters family has been investigated for decades. The indirect matrix converter
features an easy to implement and secure commutation technique, the dc-link zero
current commutation. Moreover, the conventional indirect matrix converter has
bidirectional power flow capabilities and it can be designed to have small sized
reactive elements in its input filter. These characteristics make the matrix con-
verter a suitable technology for high efficiency converters for specific applications
such as military, aerospace, wind turbine generator system, external elevators for
building construction and skin pass mill, where these advantages make up for
the additional cost of an indirect matrix converter compared to conventional con-
verters. This converter uses complex pulse width modulation and space vector
modulation schemes to achieve the goal of unity displacement power factor and
sinusoidal output current. Thanks to technological advances, fast and powerful
microprocessors are used for the control and modulation of power converters.
To deal with the high processing power needed for these microprocessors, some
research has shown the positive potential of model-based predictive control tech-
niques in many power electronics applications. While there are a few challenges to
the predictive control method, it has been demonstrated as an appealing alterna-
tive to power converters control because its concepts are very intuitive and easy
xvi
List of Figures xvii
to understand, and it can be applied to a wide variety of systems. In addition,
it may involve multiple variable systems and non-linear constraints, making it
an easy controller to implement, especially since it is open to modifications and
extensions for specific applications. Predictive current control can be described
as a particular case of model-based predictive control which takes into account
the inherent discrete nature of the switching states of the power converter and
the digital implementation. Most of predictive current control methods applied
in matrix converters take into consideration the output current regulation and
the reactive power minimization on the input side, obtaining input currents in
phase with their respective phase voltages. However, this cannot ensure that they
present a sinusoidal waveform, especially when harmonic distortion in the source
voltage or strong resonances on the input filter are present. To enhance the qual-
ity of the source current, in this document is illustrate how the predictive current
control can be applied to an indirect matrix converter and how both source and
load currents waveforms can be controlled.
Resumen
El convertidor matricial indirecto ha sido objeto de investigacion durante algun
tiempo. Una de las caracterısticas favorables de este convertidor es la ausencia
de un condensador en el enlace dc, permitiendo la construccion de convertidores
compactos y capaces de operar en condiciones atmosfericas adversas, tales como
temperaturas y presiones extremas. Estas caracterısticas han sido estudiadas am-
pliamente y son las razones principales por las que la familia de convertidores
matriciales han sido investigados desde hace decadas. El convertidor matricial in-
directo cuenta con una tecnica de conmutacion facil de implementar y mas segura,
la conmutacion a corriente cero en el enlace dc. Ademas, el convertidor matricial
indirecto convencional tiene capacidad de flujo bidireccional de energıa y puede
ser disenado para tener pequenos elementos reactivos en su filtro de entrada.
Estas caracterısticas hacen que el convertidor matricial sea una tecnologıa ade-
cuada para convertidores de alta eficiencia en aplicaciones especıficas, tales como
aeroespacial, militar, sistemas de generadores eolicos, elevadores externos para
la construccion y el molinos de bolas, donde las ventajas compensan los costes
adicionales de una forma indirecta del convertidor matricial en comparacion a
los convertidores convencionales. Este convertidor usa esquemas de modulacion
por ancho de pulso y vectorial coplejos para asegurar el objetivo de factor de
potencia unitario y corrientes de carga sinusoidales. Gracias a los avances tec-
nologicos, rapidos y poderosos microprocesadores se utilizan para el control y
la modulacion de los convertidores de potencia. Para hacer frente al alto poder
de procesamiento necesarios por estos microprocesadores, algunas investigaciones
han demostrado el potencial positivo de las tecnicas de control predictivo basado
xviii
List of Figures xix
en modelos en muchas aplicaciones de electronica de potencia. Si bien aun existen
algunos desafıos en los metodos de control predictivo, se ha demostrado como una
alternativa atractiva para poder controlar los convertidores de potencia ya que
sus conceptos son muy intuitivos y faciles de entender, pudiendo aplicarse a una
amplia variedad de sistemas. Ademas, esta tecnica puede incluir varios aspectos,
la compensacion de tiempo muerto, y las restricciones no lineales, por lo que es un
controlador de facil aplicacion, sobre todo porque esta abierto a modificaciones y
extensiones para aplicaciones especıficas. El control predictivo de corriente puede
ser descrito como un caso particular de control predictivo basado en modelos, el
cual tiene en cuenta la propia naturaleza discreta de los estados de conmutacion
del convertidor y la implementacion digital. La mayorıa de los metodos de con-
trol predictivo de corriente aplicados en los convertidores matriciales consideran
la regulacion de la corriente de salida y la minimizacion de la potencia reactiva en
el lado de la entrada, obteniendo corrientes de entrada en fase con sus respectivos
voltajes de fase. Sin embargo, esto no puede asegurar que las corrientes de entrada
presentaran formas de onda sinusoidales, especialmente cuando esta presente dis-
torsion armonica en los voltajes de alimentacion y fuertes resonancias en el filtro
de entrada. Para mejorar la calidad de la corriente de entrada, a lo largo de este
documento se ilustrara como el control predictivo de corriente puede ser aplicado
a un convertidor matricial indirecto y como las formas de onda de las corrientes
de entrada y carga pueden ser controladas.
Acknowledgments
I really appreciate the patience, support, understanding, teaching, caring and love
of my family: Marianela, Marco, Consuelo, Caroll, Natalia, Constanza, Sebastian,
Sofıa. You give meaning to my life. Thank you very much.
My sincere thanks to financial support from the National Commission for Scien-
tific and Technological Research (CONICYT) and most especially, to Professor
Jose Rodrıguez of the Universidad Tecnica Federico Santa Marıa, professors Jose
Espinoza and Cesar Silva, without whose valuable assistance and moral support
would not have been possible to complete this work. Finally, I would like to thanks
to: Universidad Tecnica Federico Santa Marıa, Basal Project FB021 and FONDE-
CYT 1100404.
xx
Agradecimientos
Agradezco realmente la paciencia, ayuda, comprension, ensenanzas, carino y amor
de mi familia: Marianela, Marco, Consuelo, Caroll, Natalia, Constanza, Sebastian,
Sofıa. Ustedes le dan el sentido a mi vida. Muchas gracias.
Mi mas sincero agradecimiento al soporte financiero de la Comision Nacional de
Investigacion Cientıfica y Tecnologica (CONICYT) y de manera muy especial,
al profesor Jose Rodrıguez de la Universidad Tecnica Federico Santa Marıa, a
los profesores Jose Espinoza y Cesar Silva, sin cuya valiosa colaboracion y apoyo
moral no habrıa sido posible llevar a termino este trabajo. Finalmente, me gus-
tarıa agradecer a: la Universidad Tecnica Federico Santa Marıa, al Proyecto Basal
FB021 y FONDECYT 1100404.
xxi
Chapter 1
Introduction
The growing industrial development in Chile requires the availability of an ever
increasing amount of electrical energy. As the energy is nowadays limited, a en-
ergy efficiency of industrial processes is a highly relevant topic. Power converters
play an important role in energy efficient processes since they act as an interface
between the mains and the mechanical actuator, and as such are crucial to use
the energy efficiently. Ideal power converters should operate with reduced losses,
inject sinusoidal currents into the network and operate with unitary displacement
factor. The regeneration capability of a matrix converter is also a very important
issue in order to deliver energy back to the mains during braking of mechanical
loads. All these characteristics can be fulfilled by the matrix converters. There are
different kinds of matrix converters. The indirect matrix topology is similar to a
conventional back-to-back inverter but without dc-link capacitor and with bidirec-
tional switches in the front-end. Thanks to this feature, modulation schemes used
for the operation of conventional back-to-back converters can be more or less di-
rectly applied, without the necessity of complex transformations as in conventional
matrix topologies. Moreover, the capacitor-less dc-link permits to reduce the size
of the converter and increase its reliability, as this component has the shortest
lifetime compared to the other components. In addition, new topologies featur-
ing a reduced number of semiconductors have been introduced. These topologies
1
Chapter 1. Introduction 2
can be very competitive in those applications where regeneration is not required.
Despite the apparent simplicity in the topology, it should be considered that this
converter requires a complex switching strategy that avoids the interruption of
the load currents, in order to prevent over-voltages that could destroy the power
semiconductors.
These problems have been recently reported and solved in the literature. However,
there are still some open issues for the operation of this topology:
- The modulation methods that have been presented to date are complex and
need a high computation effort of the controller, making this topology less
attractive in comparison to well known standard solutions.
- Matrix converters operate with high switching frequency; therefore the effi-
ciency of the converter is reduced due to the high losses during each switch-
ing transition. This aspect is critical in high power applications. On the
other hand, the operation at low switching frequency adds the extra issue
of resonances in the input filter.
- Almost all modulation schemes are based on the assumption that the input
voltages are sinusoidal and balanced. This condition is not always met in
the industry, where the voltages can be unbalanced and also distorted.
Despite these issues, after almost three decades of intensive research, the develop-
ment of the matrix converter is reaching industrial application. In effect, at least
one big manufacturer of power converters (Yaskawa) is now offering a complete
line of standard units for up to several megawatts and medium voltage using cas-
cade connection. These units have rated power (and voltages) of 9-114kVA (200V
and 400V) for low voltage MC, and 200-6.000kVA (3.3kV, 6.6kV) for medium
voltage [1].
Chapter 1. Introduction 3
1.1 State of the art review
Matrix converters are forced-commutated converters which use an array of con-
trolled bidirectional switches to synthesize a variable output voltage with unre-
stricted frequency [2]. This topology has recently attracted a great interest as it
fulfills the most desirable characteristics in a converter: it is compact and allows
the generation of a load voltage with arbitrary amplitude and frequency. More-
over, the operation with sinusoidal input currents and power factor equal to one
is also possible. The capacity of regeneration makes this topology especially at-
tractive to drive loads that need to be braked, increasing the overall efficiency of
a system. The development of matrix converters starts with the works presented
in [3, 4], where the topology was described using bidirectional power switches.
These authors also present a method to generate the output voltages by using a
transfer function applied to the input voltages.
A rather different concept was presented in [5]. Here, the idea of a fictitious dc-link
was introduced and the switching states are arranged in such a way that the in-
put phase with most positive value and the input phase with most negative value
are always selected. This concept is known as indirect transfer function [6], as
an intermediate stage is used for the synthesis of the output voltages. The physi-
cal implementation of this mathematical concept gives rise to the indirect matrix
topology, where a six-switch bidirectional bridge corresponds to the rectifier, fol-
lowed by a capacitor-less dc-link stage connected with a conventional six-switch
unidirectional inverter [7]. This topology has the same performance as the conven-
tional matrix converter in terms of voltage transfer ratio, four quadrant operation,
unity displacement power factor and sinusoidal waveforms. However, due to the
topological similarity between this converter and standard inverters, well known
space vector modulation (SVM) methods can be applied, simplifying the imple-
mentation of control schemes. In addition, it allows the reduction of the number of
switches under certain conditions and the clamp capacitor can be greatly simpli-
fied to one diode and one capacitor [8]. Due to these advantages, indirect matrix
topologies have received considerable attention in the last few years and several
Chapter 1. Introduction 4
contributions have been made in the form of new modulation methods, with the
control of reactive power in the input currents, the analysis of the operation with
distorted power supply as well as new topologies with reduced number of switches,
also known as sparse matrix converters [9, 10]. In [11, 12] a SVM scheme is pre-
sented that achieves the theoretical maximum output voltage of this topology.
In [13], a modulation method is presented that uses the zero space phasors of the
inverter stage in order to achieve a zero dc-link current commutation of the rec-
tifier, thereby obtaining a soft-switching operation. Thanks to this characteristic,
the switching losses on the side of the rectifier are significantly reduced. In [10],
the efficiency of a sparse matrix converter is improved by employing the lowest
and the second largest phase voltage for the generation of the dc-link voltage. All
these schemes feature a high switching frequency, therefore high semiconductor
losses can be expected on the side of the inverter despite of the aforementioned
improvements. Even though the recent advances in semiconductor technology per-
mit a reduction of the losses by using reverse blocking IGBTs [14], the operation
at a lower frequency is still an interesting issue. In this case, the resonances of the
input filter play an important role that has to be taken into account. It should
be pointed out that most modulation strategies do not consider the filter in the
input side for the generation of sinusoidal currents [9]- [15]. Almost all the existing
modulation schemes assume symmetrical sinusoidal voltages for the operation the
converter. It has been suggested in [16] that the current controllers are unable
to eliminate the distortion in output currents, in this case those strategies that
measure the input voltages generate output voltages of higher quality. In [17] it
has been proposed a hybrid matrix topology to make the converter immune to
a voltage unbalance in the inputs, however, this approach includes the addition
of extra switches and a small dc-link, adding extra complexity to the topology.
The problem has been more or less extensively studied for conventional matrix
topologies, however there are almost no papers reporting the problem in indirect
matrix topologies. The importance of this issue is clear, as it can directly affect
the quality of the process in which the converter is involved.
Chapter 1. Introduction 5
1.2 Hypothesis and contribution of this thesis
The main contribution of our research is to propose a simple and effective pre-
dictive control scheme for the indirect matrix converter that accomplishes the
standard requirements of other techniques such as unitary power factor and the
operation under abnormal input conditions. This aims to improve the process
quality by reducing the effect in the output process of an abnormal condition in
the mains and also improve the performance of the source currents in despite of
the distorted ac-supply and resonances of the input filter. The hypotheses are:
- It has been reported that predictive control schemes permit a simple way to
control of standard matrix and back-to-back converter topologies featuring
near sinusoidal input currents, unitary displacement power factor and near
sinusoidal output currents; comprising all usual units as modulators and
PI-controllers in only one control block [18]. Therefore it should be easy to
extend this idea to the control of indirect matrix converters.
- It has been reported that an active damping method can be employed to
mitigate the potential resonance of the input filter [19]. This idea can be com-
plemented with predictive control to operate the indirect matrix converter
at a potential resonance frequency in the input filter. The active damping
method does not involve additional measurements or any modification to
the predictive algorithm and thus it is easy to implement.
- The control scheme works under the assumption of a good model of input
filter. Since the proposed scheme always chooses the best option out of the
measured values of input current and voltages, an abnormal supply can be
also considered in the calculation of the optimal switching state.
Chapter 1. Introduction 6
1.3 Chapter review
This thesis is divided in nine chapters. Chapter 1 is an introduction to the sub-
ject of the research. It contains a comprehensive review of research reported in
the literature to date, establishes the contributions of the thesis and a chapter
preview. Chapter 2 presents a summary of the most popular and important ac/ac
topologies with their relevant characteristics. In Chapter 3, a review of the most
important modulation and control techniques for matrix converters is detailed. In
Chapter 4 a brief description of the indirect matrix converter where its mathemat-
ical model and the most common control and modulation method is described.
An introduction to the model-based predictive control in power electronics is in-
troduced in Chapter 5. In Chapter 6, the first predictive approach is presented:
Predictive current control. The strategy is tested throughout simulation and prac-
tical results. Previously, this strategy was implemented by Dr. Pablo Correa, but
considering a programmable ac-supply. In contrast, we implemented this strategy
using a three -phase variac as the ac-source, which behaves like a weak ac-source
for the system, due to the associated inductance with the autotransformer connec-
tion. The effect of a distorted voltage and filter resonance on the source currents
is analyzed and discussed in order to provide an introduction to the Chapter 7,
where the predictive method was enhanced with an active damping implemen-
tation. In this chapter, the objective was to reduce the harmonic distortion of
the source currents. An improvement on the source current’s performance was
obtained but the ac-source distortion is still a major issue, because the harmonic
distortion of the source voltage is reflected on the source current. This problem
has been mitigated by imposing a sinusoidal source current on the input side,
which improved the performance of the converter, as reported in Chapter 8. Fi-
nally, Chapter 9 includes the conclusions and comments from this research. The
Appendix contains a description of the experimental setup and the list of publi-
cations in ISI Journals and international conferences derived from this research
and international cooperations.
Chapter 2
Review of three-phase ac/ac
topologies
2.1 Classification of ac/ac power converters
A classification of the main ac/ac converters presented in literature to date is
shown in Fig. 2.1. Three subcategories can be identified: converters with dc-link
energy storage, converters without dc-link energy storage, and an intermediate
category of hybrid matrix converters. In the first group are the current and voltage
source topologies, with which it is possible to obtain ac/ac conversion taking
into consideration the presence of a capacitive or inductive dc-link, respectively
[20, 21]. These structures have been widely studied and they are the converters
used in the industry today. In the group of ac/ac circuits without dc-link, different
topologies have been reported in the literature, which are classified into four main
groups: the cycloconverter in a wide power range, the direct matrix converter
(DMC), the indirect matrix converter (IMC) and the three-phase buck converter,
in medium and low power range [1]. The limited voltage control range of the
basic matrix converters is a significant disadvantage compared to converters with
dc-link storage. Therefore, as represented in the third group, combinations of the
basic matrix converter topologies and the voltage source topologies with dc-link
7
Chapter 2. Review of three-phase ac/ac topologies 8
Figure 2.1. Classification of ac/ac power converters.
storage (hybrid matrix converters) were suggested to overcome this limitation.
Table 2.1 indicates a summary with the number of transistors, diodes and power
flux for each ac/ac converter which are described in the next sections.
Chapter 2. Review of three-phase ac/ac topologies 9
Table 2.1. Summary of ac/ac topologies.
Converter Transistors Diodes Power Flux
Voltage dc-link 12 12 Bi-directional
Current dc-link 12 12 Bi-directional
Cycloconverter 18∗ 0 Bi-directional
DMC 18 18 Bi-directional
IMC 18 18 Bi-directional
Sparse 15 18 Bi-directional
Very sparse 12 30 Bi-directional
Ultra sparse 9 18 Uni-direccional
HDMC 36 36 Bi-directional
HIMC 22 22 Bi-directional
* thyristor as power switches
2.2 Topologies with dc-link
2.2.1 Back-to-back converter
The back-to-back converter, Fig. 2.2(a), is the coupling of two inverters via a
capacitive dc-link which allows the decoupling of the control tasks of the input
and output side. The input side could be alternatively realized as simple diode
bridge, but the input current would contain significant low-frequency harmonics.
The diode bridge could not feedback braking energy into the mains. Therefore, a
braking resistor in the dc-link would be needed. Alternatively a thyristor bridge at
the input side could feedback braking energy, but would still suffer from significant
low-frequency input current harmonics, especially during the inverter operation.
Instead of defining a voltage in the dc-link, one could also define the current in the
dc-ink via dc-link inductor, Fig. 2.2(b). A disadvantage of such a converter system
is the large physical volume of the dc-link storage element. Furthermore, this
bulky passive component would reduce the system lifetime because its reliability
is relatively low in comparison with the other components in the power circuit.
Chapter 2. Review of three-phase ac/ac topologies 10
Figure 2.2. Three-phase ac/ac converter topologies with dc-link energy
storage; (a) voltage dc-link back-to-back converter; (b) current dc-link back-to-back
converter.
2.3 Topologies without dc-link
2.3.1 Cycloconverter
The cycloconverter, Fig. 2.3, is very common in high power applications such as
ball mills in mineral processing and cement kilns. This converter uses thyristors
which are capable of working with high voltage and power over 10 megawatts,
but the main limitation is that the output frequency depends on the natural
commutation of these switches.
Chapter 2. Review of three-phase ac/ac topologies 11
U1
V1
W1 U2
V2
W2
ia
ib icva
CBA
Figure 2.3. Three-phase ac/ac converter topology without dc-link energy
storage: cycloconverter.
2.3.2 Direct matrix converter
The direct matrix converter (DMC), Fig. 2.4, consists of an array of bidirectional
switches, which directly connects the power supply to the load without using any
dc-link or large energy storage elements [1]. One of the biggest difficulties in the
operation of this converter was the commutation of the bidirectional switches [22].
This problem has been solved by introducing intelligent and soft commutation
techniques. As mentioned before, after almost three decades of intensive research,
the development of this converter is reaching industrial application. In effect,
at least one big manufacturer of power converters (Yaskawa) is now offering a
complete line of standard units for up to several megawatts and medium voltage
using a cascade connection. These units have rated power (and voltages) of 9-
114kVA (200V and 400V) for low voltage MC, and 200-6.000kVA (3.3kV, 6.6kV)
for medium voltage [1]. Years of continuous effort have been dedicated to the
development of different modulation and control strategies that can be applied to
matrix converters [22,23].
Chapter 2. Review of three-phase ac/ac topologies 12
The DMC can be implemented with a half-bridge or with a full-bridge topology.
Aiming for a minimal component count, the half-bridge topology is often em-
phasized in publications. An exception is found in [24], where for the full-bridge
topology of the DMC a control procedure is presented that enables a low stress on
the insulation of the motor windings by minimizing the common mode voltage at
the output with a unity power factor. Consequently, the topology is of particular
interest for high power variable speed drives.
SAa
AC Motor
Bidirectional Switch
vsA
Input Filter
vsB
vsC
isA
isB
isC
vA
vB
vC
iA
iB
iC
fC
fL fR
N SBa
SCa
SAb
SBb
SCb
SAc
SBc
SCc
va vb vc
ia ib ic
Figure 2.4. Three-phase ac/ac converter topology without dc-link energy
storage: direct matrix converter (DMC).
Chapter 2. Review of three-phase ac/ac topologies 13
2.3.3 Indirect matrix converter
The indirect matrix converter (IMC) is indicated in Fig. 2.5. One of the favorable
features of an IMC is the absence of a dc-link capacitor, which allows for the
construction of compact converters capable of operating at adverse atmospheric
conditions such as extreme temperatures and pressures [25]. IMC offers the same
performance that the DMC, such as four-quadrant operation, unit power factor,
sinusoidal waveforms with variable frequency and amplitude during motoring and
regeneration operation. But the IMC features an easy to implement and more
secure commutation technique than the former, the dc-link zero current com-
mutation [26]. These characteristics make the IMC suitable technology for high
efficiency converters for specific applications such as military, aerospace, wind tur-
bine generator system, external elevator for building construction and skin pass
mill as reported in [1, 27].
Figure 2.5. Three-phase ac/ac converter topology without dc-link energy
storage: indirect matrix converter (IMC).
Chapter 2. Review of three-phase ac/ac topologies 14
2.3.4 Three-phase ac/ac buck converters
In contrast to matrix converters, three-phase ac/ac buck converters (Fig. 2.6)
only enable control of the output voltage, whereas the stationary output fre-
quency equals to the mains frequency at the input. Three-phase buck converters
are traditionally applied as soft-starters for induction motors or for saving energy
by controlling the motor flux. Due to their capacity of instantaneous control of
the output voltage, they also allow for compensation of unbalanced mains volt-
ages and thus can be utilized for power conditioning in power distribution systems.
The topology requires three series switches, providing a bidirectional current path
between the input and output phases (correspond to S1), and three inter-phase
switches (correspond to S2), providing a bidirectional current path between the
output phases. In order to enable a safe commutation, as described in [28] the
series switch and the inter-phase switch that are connected to the input phase
with the smallest voltage are switched on, whereas the other four switches are
modulated at a given duty cycle. A simpler commutation is enabled if the star
point of the input capacitors is connected with the star point of the inter-phase
switches, as suggested in [29].
A
B
C
a
b
c
1S2S
Figure 2.6. Three-phase ac/ac converter topology without dc-link energy
storage: three-phase buck converter.
Chapter 2. Review of three-phase ac/ac topologies 15
2.3.5 Sparse indirect matrix converters
As depicted in Fig. 2.5, the input stage of the IMC is implemented with six
four-quadrant switches and could therefore also be operated with a negative link
voltage vdc < 0. On the other hand, for the PWM output stage it is mandatory to
maintain vdc > 0 due to the diodes. Hence, it is possible to consider a reduction in
the number of switches by limiting the operating range of the PWM input stage
to a unipolar link voltage that retains the option of bidirectional current flow.
This circuit variant is referred to as sparse matrix converter (SMC), Fig. 2.7(a).
Compared to the DMC, this topology provides identical functionality, but with a
reduced number of power switches and the option of employing an improved zero
dc-link current commutation scheme, which provides lower control complexity and
higher safety and reliability [30,31]. The very sparse matrix converter (VSMC) is
another fully bidirectional variant of the IMC, Fig. 2.7(b), which shows identical
functionality [30]. Compared to the SMC, there is a smaller number of transistors,
and higher conduction losses due to the increased number of diodes in the con-
duction paths. A more comprehensive simplification of the IMC circuit topology
is possible by limiting the converter to unidirectional power flow. The resulting
topology is the ultra sparse matrix converter (USMC), Fig. 2.7(c). The signifi-
cant limitation of this converter topology is the restriction of its maximal phase
displacement between load-side voltage and input current to ±π/6. Possible ap-
plications would be permanent magnet synchronous machine (PMSM) with no
energy-feedback into the mains.
Chapter 2. Review of three-phase ac/ac topologies 16
Figure 2.7. Three-phase ac/ac converter topologies without dc-link en-
ergy storage; (a) sparse matrix converter (SMC); (b) very sparse matrix converter
(VSMC); (c) ultra sparse matrix converter (USMC).
Chapter 2. Review of three-phase ac/ac topologies 17
2.3.6 Indirect three-level matrix converter
As mentioned before, the output stage of the IMC is a two-level PWM inverter
(Fig. 2.5). It is thus possible to employ an inverter stage with a three-level charac-
teristic to reduce the switching frequency harmonics of the output voltage. Such
topology was proposed in [30], in which the center point for the output stage is
provided by the star point of the input filter capacitors. The same functionality
with a reduced number of switches was proposed in [32] (Fig. 2.8) and the corre-
sponding space vector modulation was described in [33]. By restricting the system
to unidirectional power flow, a considerable simplification of the indirect three-
level matrix converter circuits is possible. The input stage of this new topology
exhibits the structure of a vienna rectifier [34] and its complexity of the input
stage is comparable to an USMC. It should be emphasized that on the input side,
the transistors are only switched with twice the mains frequency, which results in
very low switching losses.
p
AN
pN
Nn
Nn
pN
n
Figure 2.8. Indirect matrix converter with an additional bridge-leg that
allows the mains phase voltages to be switched directly or inverter into the link.
Chapter 2. Review of three-phase ac/ac topologies 18
2.4 Hybrid topologies
The limited voltage control range of the basic matrix converters is a significant
disadvantage compared to converters with dc-link storage. To overcome this lim-
itation, combinations of the basic matrix converter topologies and the converters
with voltage dc-link, so-called hybrid matrix converters, were presented. If the
four-quadrant switches in a DMC are replaced by cascaded H-bridge circuits with
interlink capacitors, then the hybrid DMC topology (HDMC, [35]) results, as
shown in Fig. 2.9. In contrast to all previously discussed topologies, both the in-
put and the output currents for the HDMC are impressed and can be controlled
according to [35] by the use of at least five half-bridges. Transferring the concept
of the HDMC to the hybrid IMC (HIMC) requires in the simplest case only one
H-bridge in the link, as described in [36] (Fig. 2.10).
Figure 2.9. Circuit topology of the hybrid direct matrix converter and the
implementation of a single switching cell. A cascading of several H-bridges in each
connection of an input and output is also possible.
Chapter 2. Review of three-phase ac/ac topologies 19
Figure 2.10. Circuit topology of the hybrid indirect matrix converter with
a series voltage source in the link.
Chapter 3
A review of modulation and
control methods for
matrix converters
The most relevant modulation and control methods developed up to now, for the
matrix converter are presented in Fig. 3.1. The first and highly relevant method is
called the direct transfer function approach also known as the Venturini method.
Here, the output voltage is obtained by the product of the input voltage and the
transfer matrix representing the converter. Another strategy is the scalar method
developed by Roy, which consists of using the instantaneous voltage ratio of spe-
cific input phase voltages to generate the active and zero states of the converter’s
switches. A very important solution for the control of matrix converters comes
from the use of pulse width modulation (PWM) techniques previously developed
for voltage source inverters. The simplest approach is to use carrier based PWM
techniques. A very elegant and powerful solution currently in use is to apply space
vector modulation (SVM) in matrix converters. An alternative solution is direct
torque and flux control which has also been proposed for the speed control of
an ac-machine driven by this converter. More modern techniques, such as pre-
dictive control have recently been proposed for the current and torque control of
20
Chapter 3. A review of modulation and control methods for matrix converters 21
Figure 3.1. Summary of modulation and control methods for matrix converters.
ac-machines using matrix converters. In the following pages, a description and a
comparison of these technologies will be presented.
3.1 Scalar techniques
3.1.1 Direct method: Venturini
Modulation is the procedure used to generate the appropriate firing pulses to
each bidirectional switches (Sij). This method was proposed by Venturini in [2,3]
and has been used since, as reported in [22, 37–41]. In this case, the objective of
the modulation is to generate variable frequency and variable amplitude sinusoidal
output voltages (vjN ) from the fixed frequency and fixed amplitude input voltages
(Vi). Here, the instantaneous input voltages are used to synthesize a signal whose
low-frequency component is the desired output voltage. If tij is defined as the time
during which switch Sij is on and Ts as the sampling interval, we can express the
aforementioned synthesis principle as:
vjN =tAjvA+tBjvB+tCjvC
Ts, (3.1)
where vjN is the low-frequency component (mean value calculated over one sam-
pling interval) of the jth output phase and changes in each sampling interval.
With this strategy, a high frequency switched output voltage is generated, but a
fundamental component has the desired waveform.
Chapter 3. A review of modulation and control methods for matrix converters 22
Obviously, Ts = tAj + tBj + tCj, ∀j, with j = a, b, c and therefore, the following
duty cycles can be defined,
mAj(t) =tAj
Ts, mBj(t) =
tBj
Ts, mBj(t) =
tBj
Ts. (3.2)
Extending eq. (3.1) to each output phase, and using eq. (3.2), the following equa-
tion can be derived,
vo(t) = M(t)vi(t), (3.3)
where vo(t) is the low-frequency output voltage vector, vi(t) is the instantaneous
input voltage vector and M(t) is the low-frequency transfer matrix of the con-
verter, defined as:
M(t) =
mAa(t) mBa(t) mCa(t)
mAb(t) mBb(t) mCb(t)
mAc(t) mBc(t) mCc(t)
. (3.4)
Following an analogous procedure for the input current it can be shown that,
ii(t) = MT (t)io(t), (3.5)
where ii(t) is the low-frequency component input current vector, io(t) the instan-
taneous output current vector and MT (t) the transpose of M(t). Eq. (3.3) and
eq. (3.5) are the basis of the Venturini modulation method, leading to the con-
clusion that the low-frequency components of the output voltages are synthesized
with the instantaneous values of the input voltages and that the low-frequency
components of the input currents are synthesized with the instantaneous values
of the output currents. Suppose that the input voltages vi are given by:
vi(t) =
Vicos(wit)
Vicos(wit− 2π/3)
Vicos(wit+ 2π/3)
, (3.6)
and that due to the low-pass characteristic of the load the output currents io are
sinusoidal and can be expressed as:
io(t) =
Iocos(wot+ φo)
Iocos(wot− 2π/3 + φo)
Iocos(wot+ 2π/3 + φo)
, (3.7)
Chapter 3. A review of modulation and control methods for matrix converters 23
with wi = 2πfi and wo = 2πfo, where fi and fo correspond to the source and
load frequencies, respectively. Vi corresponds to the input voltage amplitude and
Io the output current amplitude. Furthermore assuming that the desired input
current vector ii is given by:
ii(t) =
Iicos(wit+ φi)
Iicos(wit− 2π/3 + φi)
Iicos(wit+ 2π/3 + φi)
, (3.8)
with Ii as the input current amplitude. Also, assuming that the desired output
voltage vo can be expressed as follows,
vo(t) =
qVicos(wot)
qVicos(wot− 2π/3)
qVicos(wot+ 2π/3)
, (3.9)
and that the following active power balance equation must be satisfied,
Po =3qViIocos(φo)
2=
3ViIicos(φi)
2= Pi, (3.10)
where Po and Pi are the respective output and input active powers, φi is the input
displacement angle and q is the voltage gain of the matrix converter. With the
previous definitions, the modulation problem is reduced to finding a low-frequency
transfer matrix M(t) such that eq. (3.3) and (3.5) are satisfied. The explicit form
of the matrix M(t) can be obtained from [2, 3], and it can be reduced to the
following expression,
mij(t) =1
3
(
1 + 2viN (t)vjN/V2i
)
, (3.11)
where i = A,B,C and j = a, b, c. An important aspect of the solution presented
is that the voltage gain of the converter cannot exceed q = 0.5, due to the working
principle (mean value) and the input voltage waveforms. To increase the gain volt-
age to q =√
3/2 = 0.866, Venturini proposed the injection of the third harmonic,
resulting in the following expression,
mij(t) =1
3
(
1 +2viN (t)vjN
V 2
i
+ 4q
3√
3ζ
)
, (3.12)
Chapter 3. A review of modulation and control methods for matrix converters 24
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04-0.05
0
0.05
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04-1.5
-1
-0.5
0
0.5
1
1.5
Figure 3.2. Direct method: venturini, typical waveforms; a) output voltage
vaN [pu], its reference (bold line) and b) output current ia [pu].
with ζ = sin(wet + βi)sin(3wet) for i = A,B,C, j = a, b, c and βi = 0, 2π3 ,
4π3 .
The same gain voltage q =√
3/2 can be obtained by using the line-to-line volt-
ages in the modulation. Typical waveforms of the output voltage and current are
presented in Fig. 3.2.
Chapter 3. A review of modulation and control methods for matrix converters 25
3.1.2 Roy’s method
The scalar method which was proposed in 1987 by G. Roy and G.E. April in [42],
consists of using the instantaneous voltage ratio of specific input phase voltages
to generate the active and zero states of the converter’s switches. The value of
any instantaneous output phase voltage (j = a, b, c) is expressed as follows,
vjN = 1Ts
(
tKvK + tLvL + tMvM
)
, (3.13)
tK + tL + tM = Ts, (3.14)
where the subscript M is assigned to the input voltage which has a different
polarity to the other two inputs. The subscript L is assigned to the smallest of
the other two input voltage magnitudes, subscript K being assigned to the third
input voltage. Equations (3.13) and (3.14) are similar to the ones proposed by
Venturini as mentioned in the previous subsection. In this case, the switching
patterns depend only on the scalar comparison of input phase voltages and the
instantaneous value of the desired output voltage. So, the duty cycles are given
as indicated in eq. (3.15),
mLj =(vjN−vM )vL
1.5V 2
i
mKj =(vjN−vM )vK
1.5V 2
i
mMj = 1 − (mLj +mKj)
, (3.15)
for j = a, b, c, respectively. As with the previously presented basic solution for the
modulation problem, the voltage transfer ratio is limited to q ≤ 0.5, in order to
yield positive values for times tK , tL and tM . By modifying the switching time of
the basic scalar control law, it is possible to add the third harmonic to obtain an
overall voltage transfer ratio of q =√
3/2. So, the modulation duty cycles for the
scalar method can be represented by [42]:
mij =1
3
(
1 +2vivj
V 2
i
+ 23ζ
)
, (3.16)
for i = A,B,C, and j = a, b, c. Eq. (3.12) and eq. (3.16) are equal when the
output voltage is maximum (q =√
3/2).
Chapter 3. A review of modulation and control methods for matrix converters 26
The difference between the methods is that the term q is used in the Venturini
method and is fixed at its maximum value in the scalar method. The effect on
the output voltage is negligible, except at low switching frequencies, where the
Venturini method is superior.
3.1.3 Current phase displacement control for Venturini and Roy
methods.
According to [39,42], by intentionally shifting or delaying the timing sequence with
respect to the zero crossing of the associated input phase voltage, it is possible
to shift current ii relative to vi (i = A,B,C), therefore altering the input power
displacement factor. Let’s define the following fictitious phase voltages at the input
part of the matrix converter as:
v′A = Visin(wit+ ∆φ),
v′B = Visin(wit+ ∆φ− 2π3 ),
v′C = Visin(wit+ ∆φ+ 2π3 ),
, (3.17)
where, ∆φ is the displacement factor angle between the measured input voltage
vector vi and the input current vector ii.
For the Venturini’s method, the solution of the new m′ij is given by:
m′ij(t) =
1
3
(
1 +2v′
iN(t)vjN
V 2
i
+ 4q
3√
3ζ
)
, (3.18)
for i = A,B,C.
For the Roy’s method, let us now assign M , K and L to A′, B′ and C ′ according
to the rules mentioned before. Then,
m′Lj =
(vjN−v′M
)v′L
1.5V 2
i
m′Kj =
(vjN−v′M )v′K1.5V 2
i
m′Mj = 1 − (m′
Lj +m′Kj)
, (3.19)
for j = a, b, c, respectively. Of course, desired output voltage vjN is still expressed
by eq. (3.9) or eq. (3.13), for Venturini or Roy methods, respectively.
Chapter 3. A review of modulation and control methods for matrix converters 27
It follows that the input currents ii will be in phase with their respective ficti-
tious voltages. However they will be displaced by an angle ∆φ according to the
real voltage vi. So, the input power displacement factor is totally controllable by
proper adjustment of the timing sequence, regardless of the load characteristic. In
both methods, a reduction of the voltage transfer ratio is observed as the power
displacement factor is reduced as indicated in [42].
3.2 Pulse width modulation methods
3.2.1 Carrier-based modulation method
Many control strategies based on PWM methods which allow for output voltage
regulation while maintaining unity displacement power factor on the input side
have been applied to different kinds of matrix converters, as has been reported in
[43–59]. For simplicity, we will discuss a carrier-based modulation method applied
to a three-phase to single-phase matrix converter, which can be easily extended
to a three-phase to three-phase or multilevel converter. The technique is based on
a sinusoidal pulse width modulation (SPWM), a well know shaping technique in
power electronics where a high frequency triangular carrier signal vtri, is compared
with a sinusoidal reference signal vo as shown in Fig. 3.3, [43,46]. In this method
the switching pulses are generated by using a logical table as a function of the
input voltages and the desired levels on the output side. The different input voltage
states are identified by considering variables xA, xB and xC which are generated
according Table 3.1. If the conditions given in Table 3.1 are not satisfied, the
logic variable take the value 0. The gate pulse pattern generation of the matrix
converter is given according to a switching state selector generated by the following
equation,
N = 16xA + 8xB + 4xC + 2L1 + L0, (3.20)
where L0 and L1 are the output voltage levels (L0 is selected if the level of the
output voltage reference is less than or equal to zero. L1 is selected if the output
voltage level is above zero).
Chapter 3. A review of modulation and control methods for matrix converters 28
vo*
-1
vtri
+
-L
L1 L0
LoadDMCInput Filtervs
is Rf Lf
vi
ii
vo
io RL LL
n
CfVoltage Source
3 3
3
3
S1 S99
Switching State Selector
Level Selector
Comparator
≥
≥
xA xB xC
Figure 3.3. Unipolar sinusoidal PWM method and desired output level voltage.
Generally, PWM methods can work with variable input power factor, as demon-
strated in [59], where it is possible to synthesize the sinusoidal input currents
with a desired power factor by changing the slope of the carrier and using the off-
set voltages. However, the PWM method presented in this thesis is restricted in
its operation with unity displacement power factor, due to its simplicity. Typical
waveforms of output voltage and current are presented in Fig. 3.4. More details
about this method can be found in [43,46].
Table 3.1. PWM method: input voltage states
Condition Value
vA > vB xA = 1
vB > vC xB = 1
vC > vA xC = 1
Chapter 3. A review of modulation and control methods for matrix converters 29
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04
-1
0
1
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04
-0.2
-0.1
0
0.1
0.2
Figure 3.4. Carrier-based method, typical waveforms; a) line-to-line out-
put voltage vab [pu]; b) output current ia [pu].
3.2.2 Space vector modulation (SVM) method
This method has been proposed in [60–63]. The space-vector approach is based on
the instantaneous space-vector representation of input and output voltages and
currents. Among the 27 possible switching configurations available in three-phase
matrix converters, only 21 are usefull in the SVM algorithm. The first 18 switching
configurations determine an output voltage vector and an input current vector,
having fixed directions. The magnitude of these vectors depends upon the instan-
taneous values of the input voltages and output line currents, respectively. The
last three switching configurations determine zero input current and output volt-
age vectors. The SVM algorithm for matrix converters has the inherent capability
to achieve full control of both the output voltage vector and the instantaneous
input current displacement angle [62–79]. The two-stages SVM method [80] is a
variation of the classic SVM technique which has some important features such
as over modulation, but this method is no longer used. At any given sampling
instant, the output voltage vector and the input current displacement angle are
Chapter 3. A review of modulation and control methods for matrix converters 30
known as reference quantities. The input line-to-neutral voltage vector is imposed
by the source voltages and is recognized by its measurements. Then, the control
of the input side can be achieved, controlling the phase angle of the input current
vector. Both input current and output voltage vectors are synthesized by consid-
ering the duty cycles. The duty cycles are calculated based on the phase of output
voltage and input current vector references such as is indicated as follows [63],
δ1 = −1Kv+Ki+1 2m√3
cos(φ′o − π2 ) cos(φ′i − π
2 )
cos(∆φ), (3.21)
δ2 = −1Kv+Ki2m√
3
cos(φ′o − π2 ) cos(φ′i +
π6 )
cos(∆φ), (3.22)
δ3 = −1Kv+Ki2m√
3
cos(φ′o + π6 ) cos(φ′i − π
2 )
cos(∆φ), (3.23)
δ4 = −1Kv+Ki+1 2m√3
cos(φ′o + π6 ) cos(φ′i +
π6 )
cos(∆φ), (3.24)
where m is the modulation index; ∆φ is the displacement angle between the
measured input voltage vector vi and the input current reference vector ii∗; and
Kv and Ki are the voltage and current sectors respectively, and where,
φ′o = φo − (Kv − 1)π
6, φ′i = φi − (Ki − 1)
π
6. (3.25)
If the sign of any duty cycle is negative, then the name of the switching state to
apply must have a negative sign. The duty cycle δ0 of the zero vector is such that
the total duty cycle must be equivalent to the unit at a fixed sampling frequency,
i.e.
δ0 = 1 − δ1 − δ2 − δ3 − δ4. (3.26)
Assuming a displacement power factor of 1 on the input side of the DMC, i.e.
∆φ = 0, the maximum modulation index is m =√
3/2. SVM technique for the
IMC will be presented in the next chapter.
Chapter 3. A review of modulation and control methods for matrix converters 31
3.3 Direct torque control
Today, Direct Torque Control (DTC) is established as a high performance torque
and flux control method for ac-machines fed by voltage source inverters [81, 82].
This method is based on the torque equation of the machine which is expressed as
a function of the angle between the stator and rotor flux vectors in the following
way,
Te = kT (ψrαψsβ − ψrβψsα), (3.27)
with kT = 32p
Lm
LrLs−L2m
, (where Lr, Ls and Lm are the self and mutual inductances,
respectively). The method is also based on the fact that changes in the voltage
delivered by the inverter directly affect the behavior of the machine’s stator flux,
as shown by:
ψs(k + 1) = ψs(k) + Tsvs(k + 1) −RsTsis(k). (3.28)
In [83], the authors proposed that DTC controls an ac-machine by using a matrix
converter. As shown in Fig. 3.5, this method uses a nonlinear hysteresis compara-
tor to control the torque, which delivers the control variable cT . An additional
hysteresis controller is used to create another closed loop to control the flux, which
delivers the variable cψ. A third loop is used to control the power factor of the
input current by controlling the displacement factor ϕ with another nonlinear
controller, which delivers variable cϕ. These variables cT , cψ and cϕ, in conjunc-
tion with the position of the stator flux, reveal which direction to select the gate
drive pulses in the look up table for the matrix converter’s bidirectional switches.
Although look up tables for DTC using a voltage source inverter are well known
and published in several papers and textbooks, the application of this method in
a matrix converter has additional complexity. In effect, the selection of a single
switching state for the matrix converter is not based exclusively on the informa-
tion of torque error and flux error. Rather, the designer must know a priori what
additional effect this switching state will have on the behavior of the input power
factor. To obtain this information is complex [84]. The results of this method
generally show very good performance dynamics in the control of the machine.
However, the input filter of the matrix converter presents higher resonances.
Chapter 3. A review of modulation and control methods for matrix converters 32
T
Figure 3.5. Direct torque control scheme.
To improve the general drive performance, use of DTC in matrix converters is
a subject of intensive study today. Some works are focused on improving the
behavior of the input filter [84–88].
3.4 Predictive control
3.4.1 Predictive current control
Thanks to advances in processors, predictive control schemes have recently emerged
as feasible approaches [18]. Predictive current control (PCC) scheme is illustrated
in Fig. 3.6. It shows converter’s switching state selection that leads the controlled
variables closest to their respective references at the end of the sampling period.
This strategy uses the converter and load models to predict the future behavior of
load currents and reactive power. A simple but functional time-continuous model
of a passive load side can be expressed as:
diodt
=1
LLvo − RL
LLio. (3.29)
Chapter 3. A review of modulation and control methods for matrix converters 33
LoadDMCInput Filtervs
is Rf Lf
vi
ii
vo
io RL LL
n
CfVoltage Source
3 3
3
3
27
iok+1qs
k+1
io*
Load Current
Reference
S1 S99
Cost
Function
Optimization
Reactive
Power
Prediction
isvs
3 3 3
Output
Current
Prediction
iovi
3 3
vi
27
Figure 3.6. Predictive current control scheme.
The state variable model of the ac-input side is given by eq. (3.30) and eq. (3.31)
from Fig. 2.5 as follows,
disdt
=1
Lf(vs − vi −Rf is), (3.30)
dvi
dt=
1
Cf(is − ii). (3.31)
Given the first order nature of the load model, a first order discrete approximation
allows the future load current to be predicted as:
io(k + 1) =Tsvo(k + 1) + LLio(k)
LL +RLTs, (3.32)
where Ts corresponds to the sampling time. On the input side, the equations
represent a second order model. As such, an exact discrete state model is best
used to obtain the supply current in the sampling instant k + 1, in order to
predict the future reactive power. So, the general expression to predict the line
input current is:
is(k + 1) = c1vs(k) + c2vi(k) + c3is(k) + c4ii(k), (3.33)
where ci (i = 1, 2, 3, 4) values are calculated such that the discrete model provides
the exact values of the continuous system.
Chapter 3. A review of modulation and control methods for matrix converters 34
Two main conditions must be fulfilled for the converter to operate properly: first,
the line side of the converter must minimize the instantaneous reactive power and
secondly, the load current must follow the reference with good accuracy. Both
requirements can be merged into a single quality function g as follows,
g = io(k + 1) +Aqs(k + 1), (3.34)
where,
io(k + 1) = |i∗α − iα(k + 1)| + |i∗β − iβ(k + 1)|, (3.35)
qs(k + 1) = |vsα(k + 1)isβ(k + 1) − vsβ(k + 1)isα(k + 1)|. (3.36)
The first term considers the comparison between the reference load currents and
the predicted ones. The second term corresponds to the predicted input reactive
power. Both are expressed in α-β components. The control method operates as
follows: at each sampling time, all possible switching states are used to calculate
the predicted values of the load and input current, allowing the evaluation of
function g in eq. (3.34). After that, the valid switching state that produces the
minimum value of g is selected for the next modulation period. Fig. 3.7 shows the
behavior of the matrix converter when the quality function has a value of A = 0,
for the weighting factor. The load current ia is almost sinusoidal and the reactive
power has high values. In this case the input current presents very high distortion
which is originated by a strong resonance of the input filter. Fig. 3.8 shows the
behavior of the matrix converter when a control of the input power factor is being
considered. This is achieved by increasing the value of the weighting factor to
A = 1. It can be observed that the load current ia is almost sinusoidal and that
the input reactive power is near to zero. This new control strategy practically
eliminates the resonance of the input filter. The improvement in the quality of the
input current is remarkable. Different techniques for matrix converters have been
proposed under the name of predictive current control, as reported in [89–100].
In [91], a predictive current control for an induction machine fed by a matrix
converter is developed by considering a classic stage that handles speed, flux
and torque control. It does this by means of field-oriented control (FOC), which
generates the reference currents for the predictive stage.
Chapter 3. A review of modulation and control methods for matrix converters 35
0.02 0.03 0.04 0.05 0.06 0.07 0.08-20
0
20
0.02 0.03 0.04 0.05 0.06 0.07 0.08-20
0
20
0.02 0.03 0.04 0.05 0.06 0.07 0.08
-20
0
20
Figure 3.7. Predictive current control without power factor correction
A = 0; a) output current [A]; (b) reactive power [kVAR]; (c) input current [A]
and input voltage [V/30].
A similar idea is presented in [94] to control a permanent magnet synchronous
machine (PMSM). Using this scheme it is also feasible to control other variables
within an electrical system; for example, minimizing common-mode voltage, re-
ported in [92], or increasing efficiency and reducing switching losses, as presented
in [93]. Recently, this idea has been extended to indirect matrix converters, as
reported in [26] and [98].
Chapter 3. A review of modulation and control methods for matrix converters 36
0.02 0.03 0.04 0.05 0.06 0.07 0.08-20
0
20
0.02 0.03 0.04 0.05 0.06 0.07 0.08-20
0
20
0.02 0.03 0.04 0.05 0.06 0.07 0.08
-20
0
20
Figure 3.8. Predictive current control with power factor correction A = 1;
a) output current [A]; (b) reactive power [kVAR]; (c) input current [A] and input
voltage [V/30].
3.4.2 Predictive torque control
A diagram of the predictive torque control (PTC) strategy is shown in Fig. 3.9.
This control method has been introduced in [23, 101–107]. Similar to the pre-
viously explained method, predictive torque control (PTC) consists of choosing,
at fixed sampling intervals, one of the 27 feasible switching states of the DMC. The
selection of the switching state for the following time interval is performed using
a quality function minimization technique. This quality function g represents the
evaluation criteria in order to select the best switching state for the next sampling
interval. For the computation of g, the input current vector is, the electric torque
Te, and the stator flux ψs in the next sampling interval are predicted, assuming the
application of each valid switching state, by means of a mathematical model of the
input filter and the induction machine (IM). A PI controller is used to generate
the reference torque T ∗e to the predictive algorithm. A mathematical discrete-time
model is derived to predict the behavior of the system under a given switching
state, based on well known dynamic equations for an IM [23,101–103].
Chapter 3. A review of modulation and control methods for matrix converters 37
Figure 3.9. Predictive torque control scheme.
The stator and rotor voltage equations in fixed stator coordinates for a squirrel-
cage induction machine can be presented as:
vo = Rsio +dψs
dt, (3.37)
vr = Rrir +dψr
dt− jpωψr = 0, (3.38)
where Rs and Rr are the stator and rotor resistances, ψs and ψr are the stator
and rotor fluxes, ω is the mechanical rotor speed and p is the number of pole pairs
of the IM. The stator and rotor fluxes are related to the stator and rotor currents
by:
ψs = Lsio + Lmir, (3.39)
ψr = Lmio + Lrir, (3.40)
where Ls, Lr and Lm are the self and mutual inductances respectively.
Finally, the electric torque produced by the machine can be obtained by:
Te =3
2pξImψrψs =
3
2pξ(ψrαψsβ − ψrβψsα), (3.41)
Chapter 3. A review of modulation and control methods for matrix converters 38
where ξ = Lm
LrLs−L2m
and ψr is the complex conjugate of vector ψr and the sub-
scripts α and β represent real and imaginary components of the associated vector.
Eq. (3.37) and eq. (3.38) can be rewritten, solving the stator and rotor currents
in terms of the stator and rotor fluxes from eq. (3.39) and eq. (3.40), as:
dψs
dt=
−RsLrLrLs − L2
m
ψs +RsLm
LrLs − L2m
ψr + vo, (3.42)
dψr
dt=
RrLmLrLs − L2
m
ψs −RrLs
LrLs − L2m
ψr − jpωψr. (3.43)
The next step is to define a discrete-time model based on these continuous-time
equations. Using a forward Euler approximation [23], the following discrete equa-
tions are computed from (3.42) and (3.43),
ψs(k + 1) = (1 − χLr)ψs(k) + χLmψr(k) + vo(k), (3.44)
ψr(k + 1) = λLmψs(k) + (1 − λLs)ψr(k) − jpω(k)ψr(k), (3.45)
where χ = TsRs
LrLs−L2m
, λ = TsRr
LrLs−L2m
, and Ts is the sampling period.
If a certain voltage vector vo(k) is applied from the matrix converter, then equa-
tions (3.41), (3.44) and (3.45) are used by the proposed method to predict the
stator flux and the electric torque produced by the IM during the next sampling
interval. The quality function represents the evaluation criteria to decide which
switching state is the best to apply. The function is composed of the absolute error
of the predicted torque, the absolute error of the predicted stator flux magnitude
and the absolute error of the predicted reactive input power, resulting in:
g = Te(k + 1) + λψψs(k + 1) + λqqs(k + 1), (3.46)
where λQ and λψ are weighting factors that handle the relation between reactive
input power, torque and flux conditions. This quality function must be calculated
for each of the 27 feasible switching states. The state that generates the optimum
value, in this case a minimum, will be chosen and applied during the next sampling
period. In that sense, the technique assigns costs to the objectives reflected in g,
weighted by λT , λψ and λq, and then chooses the switching state that presents the
Chapter 3. A review of modulation and control methods for matrix converters 39
lowest cost. Typical simulation waveforms without and with input factor correc-
tion are presented in Fig. 3.10 and Fig. 3.11, respectively. Both cases present good
behavior on the output side. Input currents, on the other hand, present significant
differences. Implementing the method with λq = 0, the input current shows high
distortion and phase shift with its phase voltage. Using the added term to control
the input factor and considering λq > 0, the input current is close to sinusoidal,
as shown in Fig. 3.11. Such as reported in [107] the same idea has been extended
for an indirect matrix converter.
Chapter 3. A review of modulation and control methods for matrix converters 40
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
-100
0
100
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
-50
0
50
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8-20
0
20
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.80
0.5
1
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8-20
0
20
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
-20
0
20
0.36 0.37 0.38 0.39 0.4 0.41 0.42 0.43 0.44-20
-10
0
10
20
Figure 3.10. Predictive control torque without power factor correction
λq = 0; a) speed [rad/s]; b) electrical torque [Nm]; c) output current [A]; d) stator
flux [Wb]; e) reactive power [kVAR]; f) input current [A] and input voltage [V/30];
g) zoom of input current isA and input voltage vsA.
Chapter 3. A review of modulation and control methods for matrix converters 41
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
-100
0
100
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
-50
0
50
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8-20
0
20
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.80
0.5
1
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8-20
0
20
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
-20
0
20
0.36 0.37 0.38 0.39 0.4 0.41 0.42 0.43 0.44-20
-10
0
10
20
Figure 3.11. Predictive control torque with power factor correction λq =
1; a) speed [rad/s]; b) electrical torque [Nm]; c) output current [A]; d) stator flux
[Wb]; e) reactive power [kVAR]; f) input current [A] and input voltage [V/30]; g)
zoom of input current isA and input voltage vsA.
Chapter 3. A review of modulation and control methods for matrix converters 42
3.5 Assessment of the methods
The performance of all methods can be compared considering the following figures
of merit:
i) theoretical complexity,
ii) quality of load current,
iii) dynamic response,
iv) sampling frequency,
v) switching frequency,
vi) resonance of input filter.
Table 3.2 presents a comparison of these methods. In terms of complexity, al-
though carrier based methods involve many equations, with respect to the other
techniques they are very simple to implement for generating gate drive pulses for
bidirectional power switches [53]. Predictive technique [18, 89] is very simple in
comparison to SVM [60, 63] and direct torque control methods, which are com-
plex. In DTC, the engineer must know the effect of any switching state on the
behavior of torque, flux and the input power factor of the matrix converter, which
is a complex task [83, 84]. All the methods deliver a high quality current to the
load. The main difference is that some methods work with fixed switching fre-
quency and other strategies, such as DTC and PTC, work with variable switching
frequency. It can also be observed that some methods operate with low sampling
and switching frequency while others require higher frequencies. All methods have
good dynamic behavior, which is acceptable for all main practical applications.
The resonance of the input filter is a key issue in the operation of matrix convert-
ers. An important observation, usually overlooked is that a control or modulation
method has a very significant influence on the behavior of the input filter [108]. In
effect, methods working with fixed switching frequency, like Venturini, Roy and
SVM have a reduced resonance in the input filter.
Chap
ter3.
Areview
ofm
odulation
and
control
meth
ods
form
atrixcon
verters43
Table 3.2. Comparison between control and modulation methods for matrix converters
Venturini Scalar Carrier
Based
PWM
Space
Vector -
Modulation
Direct
Torque
Control
Predictive
Current
Control
Predictive
Torque
Control
Complexity low low very low very high high low low
Sampling
frequency
very low very low low low very
high
high high
Switching
frequency
very low very low low low high high high
Dynamic
response
good good good good fast very fast very fast
Resonance
input fil-
ter
low low medium low very
high
from very
high to
low
from very
high to
low
Chapter 3. A review of modulation and control methods for matrix converters 44
Carrier based methods that do not take care of the quality of the input current
originate strong resonances in the input filter. This behavior can be drastically
improved taking into consideration the input current. DTC has very strong res-
onances in the input filter, while predictive techniques have mixed results. The
introduction of the control of the reactive power in the quality function introduces
an important reduction of the resonance in predictive methods [26,89,91,98,100,
102,106]. Other techniques that have been applied to matrix converters are fuzzy
control [109,110], neural networks [111,112], genetic algorithms [113,114], etc.
3.6 Comments and conclusion
The area of matrix converters has shown continuous development in recent years
in terms of new topologies, new control methods and applications. This chapter
has presented a number of control methods highly investigated today which, in
principle, exhibit good performance [39,51,80]. These methods have different the-
oretic principles and different degrees of complexity. With the results reported in
this chapter, predictive control appears as the most promising alternative due to
its simplicity and flexibility to include additional aspects in the control. However,
with the results reported to date in the literature it is not possible to establish
which method is the best. A deeper research must be done in the future to clarify
the advantages of each method or to select the best alternative. This comparison
must include more advanced aspects such as detailed evaluation of losses, system
integration, electromagnetic compatibility, etc.
Chapter 4
The indirect matrix converter
4.1 Description of the topology
The most important characteristics of matrix converters are [3]:
• A simple and compact power circuit;
• Generation of load voltage with arbitrary amplitude and frequency;
• Sinusoidal input and output currents;
• Operation with unity displacement power factor;
• Regeneration capability.
These highly attractive characteristics are the reason for the tremendous interest
in this topology. The indirect matrix converter (IMC) topology is shown in Fig.
4.1 and it consists of a rectifier connected to the inverter through a dc-link without
energy storage element. The converter synthesizes a positive voltage in the dc-link
by selecting a switching state in the rectifier that connects one phase to point p
and the other phase to point n. On the rectifier side, dc-link voltage vdc is obtained
as a function of the rectifier switches and the input voltages vi as follows,
vdc =[
Sr1 − Sr4 Sr3 − Sr6 Sr5 − Sr2
]
vi. (4.1)
45
Chapter 4. The indirect matrix converter 46
r1S r3S r5S
r4S r6S r2S
i1S i3S i5S
i4S i6S i2S
o
Figure 4.1. Indirect matrix converter topology.
Input currents ii are defined as a function of the rectifier switches and the dc-link
current idc as:
ii =
Sr1 − Sr4
Sr3 − Sr6
Sr5 − Sr2
idc. (4.2)
On the inverter side, dc-link current idc is determined as a function of the inverter
switches and the output currents io as:
idc =[
Si1 Si3 Si5
]
io, (4.3)
and finally, output voltages are synthesized as a function of the inverter switches
and the dc-link voltage vdc as:
vo =
Si1 − Si4
Si3 − Si6
Si5 − Si2
vdc. (4.4)
These equations correspond to the nine and eight valid switching states for the
rectifier and the inverter stage, respectively. Following the restrictions of no short
circuits in the input and no open lines in the output, the whole converter presents
72 possible switches combinations. But, another operational condition for the IMC
is that the dc-link voltage must always be positive vdc > 0. As indicated in eq.
(4.1), the dc-link voltage is synthesized by the rectifier stage switches and the
Chapter 4. The indirect matrix converter 47
input voltages vi. At any instant, only three of the nine valid switching states
that can be applied to the rectifier stage to produce a positive dc-link. For this
reason, at every sampling time, only three of the nine valid switching states are
considered [26].
Finally, the number of valid switching states is reduced to 24. It should be noted
that the IMC topology includes an extra freedom degree that alleviates the com-
plexity of the commutation sequence, the zero dc-link current commutation [7,31].
In addition, the rectifier includes an LfCf filter on the input side which is needed
to prevent over-voltages and to provide filtering of the high frequency components
of the input currents produced by the commutations and the inductive nature of
the load. As indicated previously, the filter consists of a second order system
described by:disdt
=1
Lf(vs − vi) −
RfLf
is, (4.5)
dvi
dt=
1
Cf(is − ii). (4.6)
The load model is obtained similarly. Assuming an inductive-resistive load as
shown in Fig. 4.1, the following equation describes the behavior of the load,
diodt
=1
LLvo − RL
LLio. (4.7)
4.2 Zero dc-link current commutation
One approach to current commutation in indirect matrix converters is given in [30]
by combining a multistep commutation strategy of the input stage and the dead-
time commutation of a conventional dc-link inverter for the output stage, in such
a way as to avoid the possibility that more than one bidirectional switch is on at
the same time at each half bridge, so as to avoid short circuit of the input lines.
IMC provides a degree of control freedom that is not available for the conventional
direct matrix converter. This can be uses to simplify the complex commutation
problem. A simplified approach to solve the current commutation problem in
an IMC was presented in [7, 30], where a zero dc-link current commutation is
Chapter 4. The indirect matrix converter 48
Table 4.1. Valid switching state on the rectifier side.
State Sr1 Sr2 Sr3 Sr4 Sr5 Sr6 iA iB iC vdc
1 1 1 0 0 0 0 idc 0 −idc vAC
2 0 1 1 0 0 0 0 idc −idc vBC
3 0 0 1 1 0 0 −idc idc 0 −vAB4 0 0 0 1 1 0 −idc 0 idc −vAC5 0 0 0 0 1 1 0 −idc idc −vBC6 1 0 0 0 0 1 idc −idc 0 vAB
7 1 0 0 1 0 0 0 0 0 0
8 0 0 1 0 0 1 0 0 0 0
9 0 1 0 0 1 0 0 0 0 0
proposed. If the three-phase to two-phase matrix converter commutates when the
inverter stage is in freewheeling operation (either all the upper devices or all the
lower devices in the inverter circuit are gated), it is only important to make sure
that short circuit of the input lines is avoided. The open circuit of the load in this
case would not cause any problem as the dc-link current would be zero. The zero
dc-link current commutation therefore would allow a dead-time commutation of
two bidirectional switches on the input side during the period of time while the
inverter stage is in a freewheel state. At first sight the main advantage of this
commutation technique is the reduction in the switching losses of the input stage
which ideally would be negligible as it can be arranged to commutate at zero
current every time.
4.3 PWM based control method
4.3.1 Modulation of the rectifier stage
This technique is applicable to conventional IMC, SMC, VSMC and USMC, al-
lowing the switching of the rectifier stage at zero dc-link current. Valid switching
states on the rectifier side are given as indicated in Table 4.1. In this case, zero
Chapter 4. The indirect matrix converter 49
0.022 0.024 0.026 0.028 0.03 0.032 0.034 0.036 0.038 0.040
200
400
600
0.022 0.024 0.026 0.028 0.03 0.032 0.034 0.036 0.038 0.04200
400
600 st
nd
st
nd
Figure 4.2. (a) Line-to-line input voltages vi and dc-link voltage vdc; (b)
valid current vectors in the α−β plane; (c) first-maximum, second-maximum and
average dc-link voltage vdc.
states are not considered in order to obtain consistently positive dc-link voltage.
For instance, if state 1 is applied, that is, if Sr1 and Sr2 are on, phase A of the
input voltage will be connected to the positive rail of the dc-link and phase C
of the input voltage to the negative rail. Therefore, the line-to-line voltage vAC
will be reflected in the dc-link. At the same time, idc current will flow throughout
Chapter 4. The indirect matrix converter 50
si
i
Figure 4.3. Position of the reference current vector in sector II; θi is the
angle respect to α-axis; θsi is the angle in the sector.
phase A, and throughout phase C will flow −idc. With this, each current vector
can be represented in α − β axis, as indicated in Fig. 4.2(b). In this figure, it is
possible to identify six sectors of π/3.
As shown in Fig. 4.2(c), in this method the dc-link voltage is defined by segments
of the input line-to-line voltages, according to the rectifier state. Therefore, the
voltage employed by the inverter for output voltage formation shows two different
levels. For example, in sector II, the dc-link voltage is given by the line-to-line
voltages vAC and vBC . In addition, it is possible to appreciate that in the first π/6
of sector II, the maximum voltage corresponds to vAC and the second maximum
to vBC . But in the next π/6 of sector II, the maximum voltage corresponds to vBC
and the second maximum to vAC . If a unity displacement power factor operation
is desired, a current vector in phase with its respective input voltage must be
defined, which will be given by eq. (4.8) as follows,
i∗A
i∗B
i∗C
=
Iicos(wit)
Iicos(wit− 2π/3)
Iicos(wit+ 2π/3)
, (4.8)
with wi = 2πfi where fi is the input frequency and Ii corresponds to the input
current amplitude given by the load current throughout eq. (4.2) and eq. (4.3).
As this current vector is in phase with the input voltage vector, the position of
Chapter 4. The indirect matrix converter 51
the current vector in α−β axis is determined considering the measurement of the
input voltages vi, given by:
θi = arctan(vβs /vαs ), (4.9)
where vαs and vβs are the α − β components of vi. Comparing θi with the angles
that limit each sector, it is possible to know in which of the six sectors the current
reference vector is located. To know the position inside of the sector, the angle θi
is compared with the inferior angle of the previously identified sector ∠inf−sectorr ,
θsi = θi − ∠inf−sectorr . (4.10)
With all this information it is possible to determine the working cycles for the
rectifier side as follows,
δ1 = sin(π/3 − θsi), δ2 = sin(θsi). (4.11)
Finally, the time that is connected to the dc-link the voltage which generates the
first and second maximum are given by:
δr1 = δ1/(δ1 + δ2), δr2 = δ2/(δ1 + δ2). (4.12)
Thus, the average dc-link voltage vdc, which is considered for the inverter modu-
lation, is defined as:
vdc = δr1vfmax + δr2vsmax, (4.13)
where vfmax corresponds to the first maximum and vsmax, the second one. Fig. 4.4
shows the dc-link voltage vdc and average dc-link voltage vdc, where it is possible
to appreciate that the dc-link voltage is given by the first and second maximum
line-to-line input voltages.
Chapter 4. The indirect matrix converter 52
0.02 0.025 0.03 0.035 0.04 0.045 0.05 0.055 0.060
200
400
600
0.02 0.025 0.03 0.035 0.04 0.045 0.05 0.055 0.060
200
400
600
Figure 4.4. Formation of the dc-link voltage vdc and average dc-link volt-
age vdc. As it can be observed, the dc-link voltage is given by the first and second
maximun line-to-line input voltages.
4.3.2 Modulation of the inverter stage
Similar to the rectifier stage, on the load side, the valid inverter switching states
(Tabla 4.2) are considered to obtain the voltage vector which can be applied
to the converter, such as indicated in Fig. 4.5(a). In comparison to the rectifier
modulation, in this case there are three working cycles. The third is related to the
zero vector, which is needed to ensure zero dc-link current and allow proper and
safe commutation on the rectifier stage. Such as in classic voltage source inverters,
the idea in the SVM technique is to synthesize a voltage vector considering the
two adjacent vectors and the zero vector, Fig. 4.5(b). In this case, the position of
the reference vector is determined as:
θo = arctan(vβo /vαo ), (4.14)
θso = θo − ∠inf−sectori . (4.15)
Chapter 4. The indirect matrix converter 53
Table 4.2. Valid switching state on the inverter side.
State Si1 Si2 Si3 Si4 Si5 Si6 vuv vvw vwu idc
1 1 1 0 0 0 1 vdc 0 −vdc iou
2 1 1 1 0 0 0 0 vdc −vdc iou + iov
3 0 1 1 1 0 0 −vdc vdc 0 iov
4 0 0 1 1 1 0 −vdc 0 vdc iov + iow
5 0 0 0 1 1 1 0 −vdc vdc iow
6 1 0 0 0 1 1 vdc −vdc 0 iou + iow
7 1 0 1 0 1 0 0 0 0 0
8 0 1 0 1 0 1 0 0 0 0
Working cycles of adjacent vectors are obtained as follows,
γ1 = mvsin(π/3 − θso),
γ2 = mvsin(θso),(4.16)
where mv, is a variable that relates rectifier and inverter stages as:
mv = m(δ1 + δ2),
0 < m < 1.(4.17)
But in this case, a link between rectifier and inverter stages must be established,
which is given by:
γi1 = γ1/(γ1 + γ2),
γi2 = γ2/(γ1 + γ2),
γi0 = 1 − γi1 − γi2,
(4.18)
τp1 = γi1δr1 τs1 = γi1δr2
τp2 = γi2δr1 τs2 = γi2δr2
τp0 = γi0δr1 τs0 = γi0δr2
. (4.19)
Finaly, the times for each vector application on the rectifier and inverter side are
defined by eq. (4.20) and (4.21) as follows,
tr1 = δr1Ts,
tr2 = δr2Ts,(4.20)
Chapter 4. The indirect matrix converter 54
o
so
Figure 4.5. (a) Available vectors on the inverter side; (b) position of the
output reference vector in sector II.
tp1 = τp1Ts ts1 = τs1Ts
tp2 = τp2Ts ts2 = τs2Ts
tp0 = τp0Ts ts0 = τs0Ts
. (4.21)
With this, and as indicated in Fig. 4.6, it is possible to obtain an average dc-link
voltage vdc, which is used to modulate the inverter side. As mentioned before, the
rectifier commutation is made at zero dc-link current, minimizing switching losses
on the rectifier side in a safe manner.
Chapter 4. The indirect matrix converter 55
r1 r
s
p1
p2
p0
s1
s2
s0
s1
p1
p2
p0
s2
s0
r1
Figure 4.6. Formation of the dc-link voltage vdc and dc-link current idc
within a pulse period. Switching state changes of the input stage do occur at zero
dc-link current.
4.3.3 Simulations results
The method has been tested in open-loop control, considering a modulation index
equal to m = 0.866. As indicated in Fig. 4.7, a source current in phase with its
voltage can be achieved. A PI current controller was adjusted to regulate the
load current acting on the output voltage reference. The schematic diagram of
the current control strategy is presented in Fig. 4.8. By means of this method,
it is possible to set a specific load current reference. The controller will act on
the system setting the appropriate value of the output voltage reference to the
SVM algorithm in order to track the reference signal. As indicated in Fig. 4.9, by
considering a closed-loop control, and when the displacement is equal to φ = 0,
an improved source current can be obtained. In Fig. 4.10. The reference output
current delivered to the PI controller was set sinusoidal with an amplitude of
i∗a=21A and f∗s=60Hz and i∗a=14.7A and f∗s=100Hz after t = 0.12s. The resulting
load current is easily tracked with respect to its reference. The input current is
almost sinusoidal with a small distortion.
Chapter 4. The indirect matrix converter 56
0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1 0.11 0.12 0.13-40
-20
0
20
40
0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1 0.11 0.12 0.13
-20
-10
0
10
20
Figure 4.7. Simulation results SVM technique in open-loop control; (a)
source voltage vsA [V/10] and source current isA [A]; (b) output current ia [A].
f f
fL
L
Figure 4.8. Schematic diagram of the current control strategy added to
the modulation technique; the PI controller is in d-q coordinates.
Chapter 4. The indirect matrix converter 57
0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1 0.11 0.12 0.13-40
-20
0
20
40
0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1 0.11 0.12 0.13
-20
-10
0
10
20
Figure 4.9. Simulation results SVM technique in closed-loop control; (a)
source voltage vsA [V/10] and source current isA [A]; (b) output current reference
i∗a[A] and measured output current ia [A].
Chapter 4. The indirect matrix converter 58
0.06 0.08 0.1 0.12 0.14 0.16 0.18-40
-20
0
20
40
0.06 0.08 0.1 0.12 0.14 0.16 0.18
-20
-10
0
10
20
Figure 4.10. Simulation results SVM technique in closed-loop control; (a)
source voltage vsA [V/10] and source current isA [A]; (b) output current reference
i∗a[A] and measured output current ia [A].
Chapter 5
Model-based predictive control in
an IMC
Predictive control is a very wide class of controllers that have found rather recent
application in static power converters. As described in [18,115], different predictive
methods have been proposed in the literature such as deadbeat, hysteresis-based,
trajectory-based and model-based predictive control (MPC). The main character-
istic of predictive control is the use of the model of the system for the prediction
of the future behavior of the controlled variables. This information is used by the
controller in order to obtain the optimal actuation, according to a predefined op-
timization criterion. The optimization criterion in the hysteresis-based predictive
control is to keep the controlled variable within the boundaries of a hysteresis area,
while in the trajectory-based predictive control method, the variables are forced
to follow a predefined trajectory. In deadbeat control, the optimal actuation is
the one that makes the error equal to zero in the next sampling instant. A more
flexible criterion is used in MPC, expressed as a cost function to be minimized.
The difference between these groups of controllers is that deadbeat-based control
and MPC with continuous control set need a modulator, in order to generate the
required voltage. This will result in having a fixed switching frequency. The other
controllers directly generate the switching signals for the converter, do not need
59
Chapter 5. Model-based predictive control in an IMC 60
a modulator, and present a variable switching frequency. Some of the advantages
of this approach are,
• A simple and intuitive concept.
• Easy inclusion of non-linearities in the model.
• Straightforward treatment of constrains.
• The multiple-input multiple-output case is considered in the formulation.
• The versatility of the method allows its application on a wide variety of
systems.
However, this control approach has some disadvantages, as follows,
• The processing cost is directly related to the number of valid switching
states and hence in the case of the indirect matrix converter will require
a fast processor that uses only a few microseconds to evaluate all possible
predictions.
• It also has a strong dependency on the model used to predict the behavior
of the system.
• Stability can be difficult to probe because of the constraints and non-linearities
included in the model.
Model-based predictive control (MPC) is a control theory composed of a series of
non-linear regulation algorithms that determine the actuation from an optimiza-
tion process, based on predictions from a model of the system [116]. Predictive
control approach considers the benefit of the discrete nature of power converters,
which have a finite number of valid commutation states because they are based
on discrete valves, which have only two states: ON and OFF. Thus, it predicts
the system behavior on real-time for each possible commutation state and selects
the one that minimizes a cost function as future control action. Fundamentally,
and based on [117], the design of the controller consists of the following aspects:
Chapter 5. Model-based predictive control in an IMC 61
1. Converter model, where are identified all possible switching states and their
relation to input-output voltage/current. The total number of switching
states is equal to the number of different combinations of each switch, how-
ever, not all combinations are possible, since each converter has its operating
restrictions and therefore invalid commutation states are not considered.
2. Discrete-time model of the system, to predict the future behavior of the
variables to be controlled. Here it is important to define the variables that
are measurable and those that are not, because occasionally, the variables
that are required for the predictive model are not accessible and therefore
must be estimated.
3. Cost function, which represents the desired behavior of the system, This can
be expressed as the measurement error between a reference and the predicted
variable. One of the advantages of this method is that different kinds of
variables and restrictions control can be included in this cost function.
Once the controller is implemented, the behavior of the controlled variables for
all possible switching states must be predicted and the cost function must be
evaluated for each prediction, selecting the switching state that minimizes this
function and this optimal switching state is applied in the next sampling time.
This approach is known also as a Finite Set MPC (FS-MPC), since the possible
control actions (switching states) are finite. A simplified algorithm for the real-
time implementation of FS-MPC is shown in Fig. 5.1. Multiple variables, system
constraints, disturbances, saturations, and basically, every characteristic that can
be mathematically modeled and measured can be included in the predictive model
and cost function. This is the basis of the great flexibility and control potential
that can be achieved with FS-MPC. Moreover, the fact that power converters
have a reduced and limited number of switching states makes this method feasible
to implement with present-day available microprocessing resources. Since only a
discrete mode of the system is necessary, rather than approximated linear models
together with control system design theory and modulation algorithms, a simpler
and more direct design and implementation of the controller can be achieved.
Chapter 5. Model-based predictive control in an IMC 62
As reported in [18, 115], this family of non-linear control techniques has been
implemented on a wide variety of converters with different applications such as,
1. Current control on a wide variety of converters such as active front end
(AFE) rectifiers, voltage source inverters, neutral point clamped (NPC) mul-
tilevel converters, matrix converters, and others.
2. Torque and flux control, considering an induction motor as a load.
3. Direct control of the active and reactive powers for a three-phase PWM
rectifier.
4. Control of a NPC converter, where it is possible to control the load currents
while balancing the capacitor voltages and reducing the average switching
frequency.
5. Control of a direct matrix converter, where different control strategies have
been developed for the output current control, torque and flux control with
the instantaneous reactive power minimization and the reduction of switch-
ing losses and common mode voltage.
In summary, with its great potential and flexibility, predictive control emerges as
a promising tool for electric power conversion. In the particular case of matrix
converters, a problem observed in predictive control implementations has been
the input current distortion due to the filter resonance, which is altered by the
harmonics of the network and by the converter itself. As mentioned above, FS-
MPC technique does not present a fixed switching frequency, and this may be
situated close to the resonant frequency of the filter, affecting the input currents.
In the following chapters some solutions to avoid this problem will be presented.
Chapter 5. Model-based predictive control in an IMC 63
optg
1...for j n
optk
1k
jg
j n
Figure 5.1. FS-MPC generic algorithm.
Chapter 6
Predictive current control with
reactive power minimization
This chapter presents a current control scheme with instantaneous reactive power
minimization for an indirect matrix converter. The strategy uses the commutation
state of the converter in the subsequent sampling time according to an optimiza-
tion algorithm given by a simple cost functional and the discrete system model.
Simulation and experimental results with a laboratory prototype are provided in
order to validate the control scheme. The effect of a distorted source voltage and
filter resonance is analyzed.
6.1 Control scheme for the IMC
The control scheme for the IMC is represented in Fig. 6.1. The approach pursues
the selection of the switching state of the converter that leads the output currents
closest to their respective references at the end of the sampling period. In addition,
the reactive power on the line side must be minimized and finally the dc-link
voltage must be always positive [98]. First, the control objectives are obtained
and the necessary variables to obtain the prediction model are measured and
calculated, respectively.
64
Chapter 6. Predictive current control with reactive power minimization 65
Figure 6.1. Predictive current control scheme.
The model of the system and measurements are used to predict the behavior
of the variables to be controlled in the subsequent sampling time for each of
the valid switching states and as a final point, the predicted values are used
to evaluate a cost function which deals with the control objectives. After that,
the valid switching state that produces the minimum value of the cost function
is selected for the next sampling period. In order to discretize the differential
equations shown in eq. (4.1)-(4.7), in the following section it will be presented the
model used as the derivative approximation to estimate the value of each function
one sample time in the future (the variables predicted value).
6.2 Input filter and load discrete equations
The predicted values of the input side are:[
vi(k + 1)
is(k + 1)
]
= Φ
[
vi(k)
is(k)
]
+ Γ
[
vs(k)
ii(k)
]
, (6.1)
where,
Φ ∼= eATs , (6.2)
Γ ∼= A−1(Φ − I2x2)B, (6.3)
Chapter 6. Predictive current control with reactive power minimization 66
with,
A =
[
0 1/Cf
−1/Lf Rf
]
, B =
[
0 −1/Cf
1/Lf 0
]
(6.4)
The load current prediction can be obtained using a forward Euler approximation
in eq. (4.7) as:
io(k + 1) = d1vo(k) + d2io(k), (6.5)
where, d1 = Ts/LL and d2 = (1−RLTs/LL) are constants dependent on load para-
meters and the sampling time, Ts [89]. Note that the current is(k+1) and io(k+1)
depends upon the switches state through eq. (4.2) and eq. (4.4), respectively.
6.3 Cost function definition
With the system discretized model, including the load, the input filter and the
IMC, the predictive algorithm is very straight forward to implement. The goal of
this method is to apply the ideal switching state, which is the voltage space vector
vo that produces the least amount of error between the desired load current io∗
and the predicted load current response io in a given sampling time. Hence, if
the dynamic model is accurate the control algorithm will always give the best
performance. A quality function is then defined in order to be able to measure
the error between the reference and the predicted load current response. This
quality function is then computed every sample period for each commutation
state possible on the converter to select the one with the smallest error in order
to apply it at the beginning of the next sample period. The quality function can
be as simple as:
io = (i∗oα − ioα)2 + (i∗oβ − ioβ)2, (6.6)
where ioα and ioβ denotes the load current in α− β coordinates for k + 1 sample
time, and i∗oα and i∗oβ their respective references. An extra term can be added
to this quality function to minimize other parameters which should be subject
to control such as the instantaneous reactive power consumed by the IMC input
along with the filter, the common mode voltage, the commutation losses, the
Chapter 6. Predictive current control with reactive power minimization 67
positive voltage in the dc-link, and so forth. The cost function used to validate
the control scheme in this case is:
g = io + λqqs, (6.7)
which allows the control of the load current and the minimization of the instan-
taneous reactive power on the input side. In eq. (6.7), λq is a weighting factor
and qs denotes the predicted value of the instantaneous reactive power in k+ 1
sampling time, which can be expressed as follows,
qs = (vsαisβ − vsβisα)2, (6.8)
with vsα, vsβ, isα and isβ the source voltages and currents in α − β coordinates,
respectively. The instantaneous reactive power calculation leads to a minimiza-
tion of qs in order to have a unity displacement power factor on the input side.
Noting that g = 0 (for an arbitrary λq) gives perfect tracking of the load current
and unity displacement power factor at the source side, then by minimizing g, the
optimum value for commutation state is guaranteed. In practice, by the appro-
priate selection of the weighting factor λq, a given THD of the input and output
currents is obtained. The principal method for selection of the weighting factors
has been presented in [118].
6.4 Discrete time delay error compensation
Several measured and calculated variables are needed, as well as the knowledge of
the nine rectifier-side and the eight inverter-side valid switching states, to compute
the control scheme algorithm. With these IMC rectifier and inverter side valid
states there are 72 possible switching combinations which must be calculated to
select the one resulting in the less error in the quality function. If the three valid
rectifier-side switching states giving positive dc-link voltage are calculated before
the quality function calculation routine, then only 24 switching combinations must
be computed, resulting in saved computation time, but still a numerical burden
for the microelectronic controller causing an unwanted delay.
Chapter 6. Predictive current control with reactive power minimization 68
The variables measured are vs(k), is(k), vi(k), and io(k), leaving the IMC input
current ii(k) and the IMC output voltage vo(k) as functions of the kth selected
switching state to be calculated. In order to counter the delay error due to the
discrete time computation an effective and simple method is implemented: the
quality function calculation for k + 2. First the variables in k + 1 are predicted
using the already applied switching state S(k), then the variables to be controlled
are predicted for k + 2. The sample time should be sufficient for data acquisition
at time t(k), then compute the variables for k + 1 using S(k) and then calculate
the g(k + 2) to select the optimum S(k + 1), all in the same interval. vs(k + 1) is
considered equal to vs(k) due to its very small change in one sample time [26].
6.5 Simulation results
Two different simulations were carried out to probe the control method feasibility.
Simulations with and without instantaneous reactive power minimization were
done in order to evaluate the effect of introducing the instantaneous reactive power
minimization in the control scheme. The simulation parameters are established
according to the experimental setup available in the laboratory. They are indicated
in Table 6.1 and the sampling period of the control algorithm was set at Ts = 20µs.
The outputs of the controller are used to deliver the gate driver signals for the
IGBTs. These outputs are set directly by the control algorithm and no modulator
is needed. First, the control scheme is simulated without including the term that
minimize the instantaneous reactive power on the input side of the system, so λq =
0 in eq. (6.7). Results in Fig. 6.2(a) show the chaotic behavior of the input current
with a high harmonic distortion such as represented in the spectrum indicated in
Fig. 6.3(b), where it is shown that the resonance of the input filter is situated
in fres = 650Hz, according to the filter parameters. With this it is possible to
observe 1.1%, 87.2% and 91.3% of 3th, 5th and 7th harmonic, respectively. On
the other hand, the output currents follow the reference accurately as indicated
in Fig. 6.2(b). Fig. 6.3(c) shows the spectrum of the load current ia. Fig. 6.2(c)
shows the instantaneous reactive power on the input side.
Chapter 6. Predictive current control with reactive power minimization 69
Table 6.1. Experimental setup parameters
Variables Description Value
Ts Sampling time 20µs
Vs Supply phase voltage 105V
fs Supply frequency 50Hz
Lf Input filter inductance 5.9mH
Cf Input filter capacitance 10µF
Rf Input filter resistance 0.5Ω
RL Load resistance 10Ω
LL Load inductance 15mH
fo Output frequency 50Hz
λq Weighting factor 0; 0.003
Due to the chaotic behavior of the source current, a high reactive power is pre-
sented on the input side, which is not desired. In this case, the ac-supply vsA is
clean with a sinusoidal waveform and not harmonic distortion (Fig. 6.3(a)).
In the second case, Fig. 6.4, the control strategy is evaluated considering λq =
0.003 in eq. (6.7). Fig. 6.4(a) shows an improved input behavior, with almost
sinusoidal current in correct phase with the input phase voltage, fulfilling the con-
dition of unitary displacement power factor, with a reduced harmonic distortion
such as indicated in Fig. 6.5(b). In this case it is possible to observe 0.3%, 2.7%
and 1.2% of 3th, 5th and 7th harmonic, respectively. On the output side, the load
current presents good tracking with respect to its reference, Fig. 6.4(b). Fig. 6.4(c)
shows the improvement in the instantaneous reactive input power minimization,
thus the goal of proposed predictive current control is clearly verified. It must
be acknowledged that the main advantage of the proposed control method is the
simplicity of implementation, since the controller does not need a complex mod-
ulation unit. This can reduce the overall cost of the complete system.
Chapter 6. Predictive current control with reactive power minimization 70
0.4 0.42 0.44 0.46 0.48 0.5-10
0
10
0.4 0.42 0.44 0.46 0.48 0.5-5
0
5
0.4 0.42 0.44 0.46 0.48 0.50
500
1000
Figure 6.2. Simulation results without instantaneous reactive power min-
imization; (a) source voltage vsA/10 [V] and current isA [A]; (b) output current
reference i∗a and measured ia [A]; (c) reactive power qs [VA].
0 200 400 600 800 10000
0.5
1
0 200 400 600 800 10000
0.5
1
0 200 400 600 800 10000
0.5
1
th th
res
Figure 6.3. Simulation results; (a) spectrum of source voltage [pu]; (b)
spectrum of source current [pu]; (c) spectrum of output current [pu].
Chapter 6. Predictive current control with reactive power minimization 71
0.4 0.42 0.44 0.46 0.48 0.5-10
0
10
0.4 0.42 0.44 0.46 0.48 0.5-5
0
5
0.4 0.42 0.44 0.46 0.48 0.50
500
1000
Figure 6.4. Simulation results with instantaneous reactive power mini-
mization; (a) source voltage vsA/10 [V] and current isA [A]; (b) output current
reference i∗a and measured ia [A]; (c) reactive power qs [VA].
0 200 400 600 800 10000
0.5
1
0 200 400 600 800 10000
0.5
1
0 200 400 600 800 10000
0.5
1
th th res
Figure 6.5. Simulation results; (a) spectrum of source voltage [pu]; (b)
spectrum of source current [pu]; (c) spectrum of output current [pu].
Chapter 6. Predictive current control with reactive power minimization 72
6.6 Experimental results
A laboratory IMC prototype designed and built by Universidad Tecnica Federico
Santa Marıa, thanks to the support of the Power Electronics Systems Laboratory
of ETH in Zurich, was used for the experimental evaluation. The converter fea-
tures IGBTs of type IXRH40N120 for the bidirectional switch, standard IGBTs
with anti-parallel diodes IRG4PC30UD for the inverter stage. The control scheme
was implemented in a dSPACE 1103 which is connected to additional boards that
include the FPGA for the commutation sequence generation and the signal con-
ditioning for the measurement of voltages and currents. A brief description of the
experimental setup will be presented in the Appendix.
Similarly to the previous section, first, the control strategy is evaluated considering
λq = 0 in eq. (6.7). Fig. 6.6(a) shows the chaotic behavior of the input current with
a high harmonic distortion such as indicated in the spectrum of Fig. 6.7(b). Here
it is observed that the input filter resonance is situated in fres = 650Hz, approxi-
mately, according to the filter parameters. As mentioned before, it is necessary to
add an input filter to assist the commutation of switching devices and to mitigate
against line-current harmonics. However, the filter configuration, which is shown
in Fig. 4.1, presents a resonance frequency and it can be excited by the utility due
to the potential 5th and 7th harmonics in the ac-source and also by the converter
itself. Due to the available ac-source in the laboratory, the input filter resonance
is reflected in the source voltage as seen in Fig. 6.6(a) and the spectrum of Fig.
6.7(a). A summary of the total harmonic distortion (THD) is presented in Table
6.2.
Table 6.2. Experimental THD results with λq = 0
Harmonic vsA isA ia
THD 36.48% 66.07% 8.80%
3th 2.22% 12.25% 0.20%
5th 3.11% 6.40% 0.23%
7th 2.82% 8.55% 0.31%
Chapter 6. Predictive current control with reactive power minimization 73
Figure 6.6. Experimental results without instantaneous reactive power
minimization; (a) source voltage vsA [50V/div] and current isA [5A/div]; (b) out-
put current reference i∗a and measured ia [5A/div].
0 200 400 600 800 10000
0.5
1
0 200 400 600 800 10000
0.5
1
0 200 400 600 800 10000
0.5
1
th thres
th
th thres
th
Figure 6.7. Experimental results; (a) spectrum of source voltage [pu]; (b)
spectrum of source current [pu]; (c) spectrum of output current [pu].
Chapter 6. Predictive current control with reactive power minimization 74
As was reported in [98]- [119], when a distortion is present in the source volt-
age, the source current is not sinusoidal. For all the aforementioned reasons, it is
necessary to include a term which can help to overcome this problem.
It is known that most industrial application requires unity displacement power
factor in the grid side. For this reason, through the instantaneous reactive power
minimization, the system is forced to work with a unity displacement power factor
on the input side. The results are indicated in Fig 6.8. The measured source current
and voltage of phase A is shown in Fig. 6.8(a) and the reference and measured
output current of phase a in Fig 6.8(b). As expected, the source current fulfils the
condition of unitary power factor showing an almost sinusoidal waveform and, as
a consequence, the instantaneous reactive power is minimized. This is achieved
by increasing the value of the weighting factor from λq = 0 to λq = 0.003 which
has been empirically adjusted as explained in [118], where first it is established
in a value equal to zero in order to prioritize the control of the output current
and later it is increased slowly aiming to obtain unity displacement power factor
in the input currents while maintaining consistent and ideal behaviour on the
output side. In Fig. 6.8(b) it is possible to observe a very good tracking of the
load current ia respect to its reference i∗a. The improvement in the quality of the
source current is remarkable because an important reduction of distortion due to
the mitigation of the input filter resonance is realized. The same effect is observed
in the source voltage spectrum, Fig. 6.9(a). As can be observed in Fig. 6.8(a),
the source currents show a ripple corresponding to the resonance frequency of
the input filter and the harmonic distortion of the ac-supply such as it can be
observed in the spectrum of Fig. 6.9(a). The THD of source voltage and current
and the output current are indicated in Table 6.3.
Chapter 6. Predictive current control with reactive power minimization 75
Figure 6.8. Experimental results with instantaneous reactive power min-
imization; (a) source voltage vsA [50V/div] and current isA [5A/div]; (b) output
current reference i∗a and measured ia [5A/div].
0 200 400 600 800 10000
0.5
1
0 200 400 600 800 10000
0.5
1
0 200 400 600 800 10000
0.5
1
th th resth
th th resth
Figure 6.9. Experimental results; (a) spectrum of source voltage [pu]; (b)
spectrum of source current [pu]; (c) spectrum of output current [pu].
Chapter 6. Predictive current control with reactive power minimization 76
Table 6.3. Experimental THD results with λq = 0.003
Harmonic vsA isA ia
THD 14.82% 21.03% 8.54%
3th 1.57% 10.68% 0.97%
5th 4.90% 2.25% 1.41%
7th 2.21% 5.21% 1.15%
6.7 The problem in the source current with a weak
ac-supply
In Fig. 6.7(a) is shown the spectrum of the source voltage vsA when the term
that minimizes the instantaneous reactive power is not included in the cost func-
tion, λq = 0. In this case, the ac-source was altered, as it was highly distorted
due to the high distortion of the input current and the low-order harmonics. Of
course, the differences between simulation and experimental results are given by
the ac-source. This phenomenon is due to the utilization of a three-phase variac
as the ac-supply, which behaves like a weak ac-source for the system, due to the
associated inductance of the autotransformer connection. On the other hand, Fig.
6.9(a) shows the spectrum of the source voltage vsA when λq = 0.003. Thanks to
the minimization of the instantaneous reactive power, the harmonic distortion of
the source voltage is decreased from a THD of 36.48% to 14.82%. In Fig. 6.6(b),
a distorted source current with a THD of 66.07% was observed, but when the in-
stantaneous reactive power is minimized, a THD of 21.03% is obtained as depicted
in Fig. 6.8(b). The load current THD was 8.80% in the first case, Fig. 6.6(c), but
when the weighting factor λq is considered as λq = 0.003, an output current with
a THD of 8.54% was observed, Fig. 6.8(c).
As well the improvement of the input current is remarkable, the experimental
results do not present a desirable performance yet. For this reason, in the following
chapters, the predictive method is modified in order to improve the behavior of
the input current under a polluted ac-supply.
Chapter 7
Current control for an IMC with
input filter resonance mitigation
In the previous chapter, a predictive control scheme for the indirect matrix con-
verter with instantaneous reactive power minimization was presented. In that case,
the source currents were highly distorted. This effect was observed because the
predictive method does not have any direct control over the source current and
because of distortions in the ac-supply and filter resonance. In this chapter, the
predictive method is improved by including a method to mitigate the resonance ef-
fect of the input filter. The active damping method is based on a virtual harmonic
resistor which damps the filter resonance. Experimental results are presented to
demonstrate that the proposed control method can generate good tracking of the
output current references, achieve unity displacement power factor and reduce the
input current distortion caused by the input filter resonance.
7.1 Current control scheme for the IMC with active
damping approach
As mentioned before, an input filter is necessary to assist the commutation of
switching devices and to mitigate against line-current harmonics. However, the
77
Chapter 7. Current control for an IMC with input filter resonance mitigation 78
filter configuration shown in Fig. 4.1 presents a resonance frequency and it can be
excited by the utility due to potential harmonics in the ac-source and also by the
converter itself.
To suppress the resonances, different propositions have been reported. For exam-
ple, it is feasible to choose a proper filter resonant frequency, which may limit
performance since the LfCf resonant frequency is a function of the power system
impedance, which usually varies with the power system operating conditions. Also,
it is possible to use a high commutation frequency or connect a physical resistor
damper with the filter circuit. The first solution results in output currents featur-
ing low THD, but the converter power losses are increased significantly, wasting
energy unnecessarily and decreasing the converter efficiency. This cannot be tol-
erated in static converters where the energy efficiency is an important issue. The
second one is the classical solution; where a damping resistor physically connected
in parallel to the inductor is used to mitigate a fixed series resonance. There is an-
other strategy with which it is possible to mitigate different resonances, between
the series and parallel resonances, as reported in [19], by using active damping
control.
7.1.1 Active damping approach and implementation
Active damping is a control technique which achieves the attenuation of system
resonance without affecting the efficiency of the converter. The method considers
a virtual harmonic resistive damper Rd, which is immune to system parameter
variations, in parallel with the input filter capacitor Cf as shown in Fig. 7.1,
without affecting the fundamental component [120–126]. The converter draws a
damping current proportional to the capacitor voltage which is extracted by the
converter itself, emulating the damping resistance Rd as indicated by,
id =vi
Rd. (7.1)
As only the harmonics are mitigated and not the fundamental component, the
damping current is calculated using the harmonic capacitor voltage vih. To do
this, the input voltage vi is considered in d− q axes, passing this voltage through
Chapter 7. Current control for an IMC with input filter resonance mitigation 79
a dc-blocker digital filter, deleting the fundamental element and considering only
the harmonic components. The converter is required to draw the current that
produces the input filter resonance in the matrix converter. The transformation in
d−q axes was done by the implementation of a synchronous reference frame-phase
looked loop (SRF-PLL) [127]. Once the voltage harmonics have been obtained,
the current damping harmonics are calculated as indicated in eq. (8.11) where
vihdq corresponds to all harmonic components present in vi.
idhdq =
vihdq
Rd. (7.2)
Then, the active damping in the IMC topology is implemented by passing the
harmonic component effect present on the input side to the output side, adding
this effect to the load current reference [126]. This is possible because in the IMC
topology the input current ii is related to the output current by eq. (4.2) and eq.
(4.3). Thus, the new load current reference can be expressed as:
[
i∗do
i∗qo
]
=
[
I∗do
I∗qo
]
+
[
iddh
iqdh
]
, (7.3)
where, io∗dq =
[
i∗do i∗qo
]T
, Io∗dq =
[
I∗do I∗qo
]T
=[
I∗o 0]T
is the required
load current and the damping reference current is given by eq. (8.11) as idhdq =
[
iddh iqdh
]T
. An essential aspect of the active damping control is that it does
not require any extra measurements and, furthermore, does not incorporate any
modification of the algorithm where only the output current reference has been
modified, such as seen in Fig. 7.1.
7.2 Experimental results
Results without and with the active damping implementation are presented in
this section. In the IMC a ratio transfer equal to 0.866 can be achieved, but in
this case, a different operation point is considered because the only objective of
this experimental validation is to demonstrate the effect of the active damping
implementation.
Chapter 7. Current control for an IMC with input filter resonance mitigation 80
Figure 7.1. Predictive current control with active damping scheme.
The parameters used in the experimental tests have been given in Table 6.1 and
the sampling time is defined as Ts = 20µs. Fig. 7.2(a) shows the measured source
current and voltage of phase A and Fig 7.2(b) shows the reference and measured
output current of phase a. As expected, the source current fulfils the condition of
unitary displacement power factor because it is in phase with respect to its voltage,
so as a consequence, the instantaneous reactive power is minimized. However, the
source current and voltage show a ripple corresponding to the resonance frequency
of the input filter and the harmonic distortion of the ac-supply such as it is
observed in Fig. 7.3. Fig. 7.2(a) shows that the voltage source is altered when the
system is in resonance. As explained in the previous chapter, this phenomenon
is due to the utilization of a three-phase variac as the ac-source, which behaves
like a weak ac-supply for the system, due to the inductance associated with the
autotransformer connection.
Chapter 7. Current control for an IMC with input filter resonance mitigation 81
Figure 7.2. Experimental results without active damping approach; (a)
source voltage vsA [50V/div] and current isA [5A/div]; (b) output current reference
i∗a and measured ia [5A/div].
0 200 400 600 800 10000
0.5
1
0 200 400 600 800 10000
0.5
1
0 200 400 600 800 10000
0.5
1
th th resth
th th resth
Figure 7.3. Experimental results without active damping approach; (a)
spectrum of source voltage [pu]; (b) spectrum of source current [pu]; (c) spectrum
of output current [pu].
Chapter 7. Current control for an IMC with input filter resonance mitigation 82
Table 7.1. Experimental THD results without active damping
Harmonic vsA isA ia
THD 14.82% 21.03% 8.54%
3th 1.57% 10.68% 0.97%
5th 4.90% 2.25% 1.41%
7th 2.21% 5.21% 1.15%
On the other hand, very good tracking of the load current to its reference is
observed in Fig. 7.2(b). The THD of source voltage and current and the output
current are indicated in Table 7.1.
For the second case, the improvement in the quality of the source current and
voltage is noticeable due to an important reduction of distortion by filter reso-
nance mitigation, as shown in Fig. 7.4(a). As well, an almost sinusoidal source
current is obtained, a distortion harmonic due to the unclean ac-supply is still ob-
served which cannot be mitigated by the active damping method. Similarly to the
aforementioned case, the output current follows its reference accurately despite
the distortion added to the reference by the active damping method. In Fig. 7.2
a distorted input current with a THD of 21.03% is observed, but when the reso-
nance mitigation is applied using active damping, a 20.84% of THD is obtained.
This is a relatively small value considering the polluted source with 13.05% of
THD under normal operation. The output current THD is 8.54% without active
damping action, and 6.53% with active damping operation. It is expected that
with a clean ac-source the input and output currents THD can be decreased.
By including the active damping approach, the input and output currents distor-
tion are attenuated considerably. In the IMC topology it is possible to observe the
series resonance, which is produced by the ac-supply and the parallel resonance,
which is generated by the converter itself and the control method.
Chapter 7. Current control for an IMC with input filter resonance mitigation 83
Figure 7.4. Experimental results current control with active damping ap-
proach; a) source voltage vsA [50V/div] and current isA [5A/div]; b) output current
ia and reference i∗a [5A/div].
0 200 400 600 800 10000
0.5
1
0 200 400 600 800 10000
0.5
1
0 200 400 600 800 10000
0.5
1
Figure 7.5. Experimental results with active damping approach; (a) spec-
trum of source voltage [pu]; (b) spectrum of source current [pu]; (c) spectrum of
output current [pu].
Chapter 7. Current control for an IMC with input filter resonance mitigation 84
Table 7.2. Experimental THD results with active damping
Harmonic vsA isA ia
THD 13.05% 20.84% 6.53%
3th 1.36% 10.59% 1.18%
5th 3.96% 4.75% 2.11%
7th 1.17% 4.55% 1.08%
400 600 8000
0.05
0.1
400 600 8000
0.005
0.01
400 600 8000
0.05
400 600 8000
0.05
0.1
400 600 8000
0.005
0.01
400 600 8000
0.05
Figure 7.6. Experimental results current control without and with active
damping approach; a) zoom spectrum of source voltage vsA [pu]; b) zoom spectrum
of source current isA [pu]; c) zoom spectrum of output current ia [pu].
According to the filter parameters and as observed in Fig. 7.3 and Fig. 7.5, the
resonance frequency is located around 650Hz which is amplified by the converter
and the distorted ac-supply. From Fig. 7.6 it is verified that the filter resonance is
mitigated by considering the active damping method. This is reflected in a more
sinusoidal ac-source.
Chapter 7. Current control for an IMC with input filter resonance mitigation 85
Figure 7.7. Experimental results current control without active damping
approach; a) source voltage vsA [50V/div] and current isA [5A/div]; b) output cur-
rent ia and reference i∗a [5A/div] with output frequency reference equal to 100Hz.
Because the 3th, 5th and 7th harmonics are from the ac-source itself, the active
damping method is not able to mitigate these harmonics because the ac-source
and network models are not included in the control strategy, but the operation
of the converter in an industrial application must be independent of the grid
parameters, to which it is connected. For this reason, an active damping technique
is incorporated in the control, where the only parameter to be adjusted is the
virtual resistor regardless of the grid model. As mentioned before, the resonance
and distortion effect is reflected in the load current as it can be observed in Fig.
7.6(f). As shown in Fig. 7.7 and Fig. 7.8, a different output frequency is considered,
which is established in 100Hz. As can be observed, the active damping approach
mitigates the resonance of the input filter. But, similar to the previous case and
due to a weak ac-supply in our laboratory, the source currents are distorted as
well.
Chapter 7. Current control for an IMC with input filter resonance mitigation 86
Figure 7.8. Experimental results current control with active damping ap-
proach; a) source voltage vsA [50V/div] and current isA [5A/div]; b) output current
ia and reference i∗a [5A/div] with output frequency reference equal to 100Hz.
7.3 Comments
In summary, the resonance of the input filter is still a major concern that directly
affects the selection of the design parameters and the modulation method. The
reduction of resonances in the input filter, as described in this chapter, is an
important improvement in the converter’s performance. This approach reduces
power losses as compared to the use of real resistive damping. Experimental results
indicate that the presented strategy allows good tracking of the output current to
its reference and minimizes the instantaneous reactive power on the input side at
the same time. Active damping improves the quality of the input currents even
in the presence of a weakly damped input filter. The ac-supply has an important
influence in the behavior of the source current and better results can be expected
by optimizing the input filter and also with a clean ac-supply. In order to improve
the behavior of the source currents and force them to have a sinusoidal waveform
in spite of the filter resonances and distortions of the ac-supply, a new idea is
proposed for the IMC in the following chapter.
Chapter 8
Imposed sinusoidal source and
load currents for an IMC
A new strategy for indirect matrix converters which allows an optimal control
of source and load currents is presented in this chapter. This method uses the
commutation state of the converter in the subsequent sampling time according
to an optimization algorithm given by a simple cost functional and the discrete
system model. The control goals are regulation of output current according to an
arbitrary reference and also good tracking of the source current to its reference
which is imposed to have a sinusoidal waveform. Simulation and experimental
results support the theoretical development.
8.1 Fundaments
Most of PCC methods applied in matrix converters take into consideration the
output current regulation and the reactive power minimization on the input side,
obtaining input currents in phase with their respective phase voltages. However, as
reported in the previous chapters, this cannot ensure that they present a sinusoidal
waveform, especially when harmonic distortion is present in the source voltage. To
overcome this issue and enhance the quality of the source current, the following
87
Chapter 8. Imposed sinusoidal source and load currents for an IMC 88
pages this chapter illustrate how the PCC can be applied to an IMC and how
both source and load currents waveforms can be directly controlled. First, it is
necessary to define the impedance model of the input filter as:
Zc =1
jwsCf, (8.1)
Zl = Rf + jwsLf , (8.2)
where ws = 2πfs, with fs the source frequency. The load impedance is represented
by:
Zo = RL + jwoLL, (8.3)
where wo = 2πfo, with fo the load frequency. Finally, the filter model in terms of
impedance is given as:
vs = vi + isZl,
is = ii + vi/Zc.(8.4)
8.2 The problem on the input side
SVM and PWM techniques generate a desired output voltage with unity displace-
ment power factor [16,25,80,128,129], but there is a displacement angle between
the source line current is and input current ii due to the filter parameters and con-
sequently a displacement angle between the source voltage and current, requiring
additional controllers to handle this angle [130]. From eq. (8.1), eq. (8.2) and eq.
(8.4), this displacement angle is given as follows,
δ = arctan(
wsCf (Vs −Rf Is))/(Is(1 − w2sLfCf )
)
, (8.5)
where Vs and Is are the source voltage and current fundamental amplitudes re-
spectively. In [130] it has been proposed that two power factor compensation
methods can be used, each one considering direct SVM in order to compensate
the displacement angle δ between source voltage and current with the goal to
obtain a unity displacement power factor from a voltage transfer ratio greater
than or equal to 0.35, but the compensated displacement angle decreases while
the voltage transfer ratio increases and additionally, and the source current does
Chapter 8. Imposed sinusoidal source and load currents for an IMC 89
not present a sinusoidal waveform. In [131] the authors proposed a modified direct
SVM method to control matrix converters with transfer ratio less than 0.5, allow-
ing compensation to a maximum displacement angle of π/6 (30o) but the source
currents are not considered in this work, presenting a distorted waveform as well.
Predictive techniques that have been proposed in the last years have focused on
the minimization of the reactive power on the input side but there are no reports
of additional works based on a source current control [18,23,26,89,115].
In summary, today, most of the work on matrix converters has focused on the
control of the output side while maintaining unity displacement power factor
on the input side but there are no reports of a control of the source current
with imposed waveform like the model proposed in this chapter. In comparison to
classical controllers, by using a predictive algorithm, the controller and modulator
merges in only one block, making it easier to implement than SVM and PWM
methods. The proposed predictive strategy presented in the following sections
suggests that a control of the source current with imposed waveform should be
performed rather than a reactive power minimization and also the active damping
implementation. The predictive algorithm evaluates at every sampling time Ts all
of the 24 possible states and chooses the one that returns the minimal value for
the cost functional g to be applied in the next sampling instant. The minimization
of g guarantees two goals: that the output currents follow their references with
accuracy and that the converter draws sinusoidal input currents with a desired
input displacement power factor according to their references as well.
8.3 Predictive current control for the IMC with imposed
sinusoidal source currents
In order to minimize the computational cost, the α−β linear transform is applied
to all three-phase current and voltage vectors, defined as:
[
uα
uβ
]
=
[
2/3 −1/3 −1/3
0√
3/3 −√
3/3
]
ua
ub
uc
, (8.6)
Chapter 8. Imposed sinusoidal source and load currents for an IMC 90
where the vector [ua ub uc]T is the three-phase current or voltage vector, and
[uα uβ]T is the α− β vector. In [26] a predictive control strategy for an IMC has
been presented, where the approach pursues the selection of the switching state
of the converter that leads the output currents close to their respective references
at the end of the sampling period, while minimizing the instantaneous reactive
power on the input side. As mentioned before, the strategy proposed in [26] cannot
ensure sinusoidal waveform of source current, especially when harmonic distortion
is present in the source voltage or resonances in the input filter. The proposed
MPC scheme is represented in Fig. 8.1, where in comparison to the before men-
tioned strategy, the term which minimizes the reactive power on the input side
is replaced by a direct control of the source current waveforms in order to force
them to follow a sinusoidal reference independently of the distortion present on
the input side.
8.3.1 Prediction model
Since the predictive controller is formulated in discrete time, it is necessary to de-
rive a discrete time model for the load-converter system. As indicated in previous
sections, the input side can be represented by a state space model [89], with the
states variables is and vi obtained from eq. (4.5) and eq. (4.6) as follows,
[
vi
is
]
= A
[
vi
is
]
+ B
[
vs
ii
]
, (8.7)
where,
A =
[
0 1/Cf
−1/Lf −Rf/Lf
]
,
B =
[
0 −1/Cf
1/Lf 0
]
.
(8.8)
Chapter 8. Imposed sinusoidal source and load currents for an IMC 91
Figure 8.1. Predictive source and output current control scheme with
source current reference.
Such as indicated in previous chapters, the discrete-time state space model is
determined as:
[
vi(k + 1)
is(k + 1)
]
= Φ
[
vi(k)
is(k)
]
+ Γ
[
vs(k)
ii(k)
]
, (8.9)
with,
Φ = eATs , Γ = A−1(Φ − I2x2)B. (8.10)
The output current prediction can be obtained using a forward Euler approxima-
tion in eq. (4.7) as:
io(k + 1) = d1vo(k) + d2io(k), (8.11)
where, d1 = Ts/LL and d2 = 1 − RLTs/LL, are constants dependent on load
Chapter 8. Imposed sinusoidal source and load currents for an IMC 92
parameters and the sampling time Ts [89]. Note that the current is(k + 1) and
io(k + 1) depend upon Si(k) through eq. (4.2) and eq. (4.3).
8.3.2 Cost function definition
The error between the predicted load currents and its references can be expressed
as follows,
io(k + 1) = (i∗oα − ioα)2 + (i∗oβ − ioβ)2, (8.12)
where ioα and ioβ denotes the load current in α− β coordinates for k + 1 sample
time, and i∗oα and i∗oβ their respective references. Furthermore, the error between
the reference and predicted value of the source current can be expressed as:
is(k + 1) = (i∗sα − isα)2 + (i∗sβ − isβ)2, (8.13)
where, i∗sα and i∗sβ correspond to the source current references and isα and isβ
are the source current predictions in sample k + 1. Expressions of eq. (8.12) and
eq. (8.13) are merged in a single cost function as indicated in eq. (8.14) which is
evaluated for every switching state, and is applied to the converter the switching
state that minimizes this quality function, as has been explained before. Finally,
eq. (8.12) and eq. (8.13) are combined into a single so-called quality function as
follows,
g = io(k + 1) + γiis(k + 1), (8.14)
where γi is a weighting factor. As indicated in the previous chapters, noting that
g = 0 (for an arbitrary value of γi) gives perfect tracking of the load and source
currents, then by minimizing g, the optimum value for commutation state is guar-
anteed. In practice, by the appropriate selection of the weighting factor γi, a given
total harmonic distortion (THD) of the input and output currents is obtained.
8.4 Generation of the source current reference is∗
Assuming a sinusoidal source waveform, the amplitude of is∗, in terms of the
amplitude of io∗, vs and the system parameters, can be obtained using a power
Chapter 8. Imposed sinusoidal source and load currents for an IMC 93
balance equation between the input and output side of the IMC, i.e.
poη = pi, (8.15)
where po is the output active power, pi is the input active power and η is the
efficiency of the IMC. The active power po can be expressed as,
po =3
2I∗o
2RL, (8.16)
where I∗o is the amplitude of the output current reference. The input power pi is
defined as:
pi =3
2Revi · ii∗, (8.17)
where vi = Viejθvi , ii = Iie
jθii , Vi and Ii are the amplitudes of the input voltage
and current, respectively. Using the equations of the input filter and expressing
the filter parameters as impedances at a fixed source frequency of ωs = 2πfs, the
input voltage and current amplitudes can be expressed as follows,
vi =vs − is(jωsLf +Rf ), (8.18)
ii = is −vi
jωsCf, (8.19)
where vs = Vsejθvs , is = Ise
jθis , Vs and Is are the amplitudes of the source
voltage and current, respectively, and θvi≈ θvs . Replacing (8.18) and (8.19) in
(8.17), replacing (8.17) and (8.16) in (8.15), and solving the quadratic equation
for Is, the obtained solution for the source current amplitude I∗s , using a unitary
displacement power factor in the mains (θis = θvs) is:
I∗s =Vsk1 −
√
V 2s k2 + I∗o
2k3
k4, (8.20)
where,
k1 = 8π2f2sLfCf − 1, k2 = k2
1 ,
k3 = 4ηRLRfk1, k4 = 2Rfk1. (8.21)
In addition, it is necessary to implement a Phase-Locked-Loop (PLL) to obtain
the phase of the fundamental source voltage in order to generate the sinusoidal
Chapter 8. Imposed sinusoidal source and load currents for an IMC 94
reference. Finally, the resulting source current reference is defined as:
i∗sA = I∗s sin(wst+ θ)
i∗sB = I∗s sin(wst− 2π/3 + θ)
i∗sC = I∗s sin(wst+ 2π/3 + θ)
, (8.22)
where θ is the parameter that allows a variable power factor and it is considered
equal to zero in order to obtain unity displacement power factor.
8.5 Simulation and experimental results
Both simulation and experimental results are presented in this section, by consid-
ering the same parameters employed in the previous chapters, which are indicated
in Table 6.1. The control operates with a sample time of Ts = 20µs. The control
strategy is evaluated considering the cost function indicated in eq. (8.14) and
with a weighting factor λi equal to λi = 20. This variable has been empirically
adjusted as explained in [118], where first it is established in a value equal to zero
in order to prioritize the control of the output current and later it is increased
slowly aiming to obtain minimal THD of source and load currents.
Simulation results are presented in Fig. 8.2(a) which show the source current isA
and its respective reference i∗sA and source voltage vsA, where the condition of
unitary power factor is fulfilled. This condition is imposed by the source current
reference i∗sA = I∗s sin(wst+ θ), because the phase of this current reference is the
same of the source voltage. In this method, the source current is forced to have
a sinusoidal waveform with an amplitude of Is = 2.11A. On the output side, the
load current ia presents good behavior with an almost sinusoidal waveform and
4.5A of amplitude, according to its reference (Fig. 8.2(b)). In order to validate
the simulations, Fig. 8.3 shows the experimental results by considering the same
parameters and references.
In Fig. 8.3(a) is shown the source current isA and its respective source voltage
vsA, where the condition of unitary power factor is fulfilled, which is imposed by
the source current reference i∗sA, being both source voltage and current in phase.
Chapter 8. Imposed sinusoidal source and load currents for an IMC 95
0.4 0.41 0.42 0.43 0.44 0.45-5
0
5
0.4 0.41 0.42 0.43 0.44 0.45-5
0
5
Figure 8.2. Simulation results predictive control with imposed sinusoidal
source and load currents; a) source voltage [V/25] and current [A]; b) output
current and reference [A].
Figure 8.3. Experimental results predictive control with imposed sinu-
soidal source and load currents; a) source voltage [50V/div] and current [5A/div];
b) output current and reference [5A/div].
Again, the source current is forced to have a sinusoidal waveform independently
of the distortion present in the source voltage or the resonance of the input filter,
with an amplitude of Is = 2.11A. For this reason, the source current isA is almost
Chapter 8. Imposed sinusoidal source and load currents for an IMC 96
Table 8.1. Experimental THD results with θ = 0
vsA isA ia
THD 15.67% 29.24% 7.79%
3th 1.25% 9.88% 1.03%
5th 3.74% 4.00% 2.67%
7th 1.04% 3.89% 1.84%
sinusoidal and the harmonic distortion and filter resonance are almost mitigated.
As it can be shown in Fig. 8.3(a), the source voltage is not completely clean
because of the utilization of a three-phase variac as the ac-source, which behaves
like a weak ac-supply for the system, due to the inductance associated with the
autotransformer connection. On the output side, the load current ia presents good
behavior with a near sinusoidal waveform and an amplitude of 4.5A according to
its reference Fig. 8.3(b). This method does not involve greater calculations and
the considered sampling time Ts is equal to the Ts used in [26]. With this idea
almost sinusoidal source and output currents can be obtained, realizing desirable
tracking to their respective references. The proposed strategy is immune to source
voltage distortion and/or input filter resonances. A summary of the experimental
THD is presented in Table 8.1. A different output frequency is considered in order
to demonstrate the strategy in a different operation point, such as indicated in
simulation results in Fig. 8.4 and experimental results in Fig. 8.5, where an output
frequency of 100Hz has been imposed on the load. To demonstrate the effectiveness
of the proposed method, two tests have been done such as shown in Fig. 8.6 to
Fig. 8.9. Simulation results with a displacement of θ = 30o and θ = −30o between
source voltage and current, while maintaining the output current control, are
presented in Fig. 8.6 and Fig. 8.8, respectively. Similar experimental results are
indicated in Fig. 8.7 to Fig. 8.9. Again, the source voltage presents an harmonic
distortion due to the ac-supply utilized and the filter resonance.
Chapter 8. Imposed sinusoidal source and load currents for an IMC 97
0.4 0.41 0.42 0.43 0.44 0.45 0.46 0.47 0.48 0.49 0.5-5
0
5
0.4 0.41 0.42 0.43 0.44 0.45 0.46 0.47 0.48 0.49 0.5-5
0
5
Figure 8.4. Simulation results predictive control with imposed sinusoidal
source and load currents; a) source voltage [V/25] and current [A]; b) output
current and reference [A] with an output frequency reference of 100Hz.
Figure 8.5. Experimental results predictive control with imposed sinu-
soidal source and load currents; a) source voltage [50V/div] and current [5A/div];
b) output current and reference [5A/div] with an output frequency reference of
100Hz.
Chapter 8. Imposed sinusoidal source and load currents for an IMC 98
0.4 0.41 0.42 0.43 0.44 0.45-5
0
5
0.4 0.41 0.42 0.43 0.44 0.45-5
0
5
Figure 8.6. Simulation results predictive control with imposed sinusoidal
source and load currents; a) source voltage [V/25] and current [A] 30o displace-
ment angle; b) output current and reference [A].
Figure 8.7. Experimental results predictive control with imposed sinu-
soidal source and load currents; a) source voltage [50V/div] and current [5A/div];
b) output current and reference [5A/div].
Chapter 8. Imposed sinusoidal source and load currents for an IMC 99
0.4 0.41 0.42 0.43 0.44 0.45-5
0
5
0.4 0.41 0.42 0.43 0.44 0.45-5
0
5
Figure 8.8. Simulation results predictive control with imposed sinusoidal
source and load currents; a) source voltage [V/25] and current [A] −30o displace-
ment angle; b) output current and reference [A].
Figure 8.9. Experimental results predictive control with imposed sinu-
soidal source and load currents; a) source voltage [50V/div] and current [5A/div];
b) output current and reference [5A/div].
Chapter 8. Imposed sinusoidal source and load currents for an IMC 100
Table 8.2. Experimental THD results with θ = +30
vsA isA ia
THD 16.81% 19.21% 7.45%
3th 0.80% 5.23% 2.44%
5th 3.45% 1.59% 1.96%
7th 0.63% 1.66% 1.59%
Table 8.3. Experimental THD results with θ = −30
vsA isA ia
THD 16.09% 19.65% 7.84%
3th 0.74% 5.41% 3.02%
5th 3.64% 2.77% 1.66%
7th 0.77% 2.84% 0.77%
Simulation and experimental results verified that it is possible to control both
source and output currents at the same time, while keeping near sinusoidal wave-
forms on both sides in spite of distortions or perturbations in the source voltage.
A summary of the experimental THD are presented in Table 8.2 and Table 8.3
for both cases. It is expected that with a clean ac-source the input and output
currents THD can be decreased.
8.6 Comments
Simulation and experimental results indicate that the presented strategy provides
good tracking of the source and output current to their references, making it
possible to control both source and output currents at the same time, while keeping
almost sinusoidal waveforms on both sides in spite of distortions or perturbations
in the source voltage. Better results can be obtained by considering using a clean
ac-supply.
Chapter 9
Conclusions
The main objective of our research has been to propose, develop and implement
a simple predictive control scheme for the indirect matrix converter that accom-
plishes the standard requirements of other techniques such as unitary power factor
and the improvement of the input current in an operation under abnormal input
conditions.
This document contains a review of the most important ac/ac topologies and the
well established modulation and control techniques for matrix converters. A brief
description of the indirect matrix converter topology and the main space vec-
tor modulation technique throughout simulation results have also been presented.
A predictive current control strategy was proposed which has been extended to
minimize the instantaneous reactive input power. Together with the theoretical
background, the study includes simulation results as well as experimental valida-
tion of the hypothesis and proposed control algorithms. Predictive current control
applied to passive load, effectively controlling the output current from the IMC
to the load and allowing maintenance of the input currents in phase with its in-
put voltage, although they are highly distorted. This distortion was due to the
utilization of a three-phase variac as the ac-supply, which behaves like a weak ac-
source for the system, due to the associated inductance with the autotransformer
connection.
101
Chapter 9. Conclusions 102
An active damping implementation was tested to mitigate the resonance of the
input filter. Active damping as found to improve the quality of the input cur-
rents even in the presence of a weakly damped input filter, but the distortion
from the ac-supply cannot be mitigated with this technique. This issue is still a
major concern that directly affects the selection of the design parameters and the
modulation method. The reduction of resonances in the input filter, as observed
in this thesis, is an important improvement in the converter’s performance.
The ac-supply has an important influence in the behavior of the source current and
better results can be expected by optimizing the input filter in combination with
a clean ac-supply. In order to improve the behavior of the source current with
the available ac-supply, a predictive control method for a conventional indirect
matrix converter has been proposed and the results presented in this thesis. The
algorithm allows simultaneous control of source and output currents with almost
sinusoidal waveforms, according their references. Simulation and experimental re-
sults indicate that the presented strategy provides good tracking of the source
and output current to their references, making it possible to control both source
and output currents at the same time, while keeping near sinusoidal waveforms
on both sides in spite of distortions or perturbations in the source voltage.
In summary, our findings indicate that this predictive control scheme is a simple
and effective alternative to conventional methods for the indirect matrix converter.
The predictive methods presented in this document can be easily implemented,
taking advantage of the present technologies available in digital signal processors.
Finally, with predictive control it is possible to obtain near sinusoidal input and
output currents in presence of resonances on the input filter and a distorted ac-
supply, which presents interesting possibilities in terms of a conceptually different
approach to optimization in the control of power converters.
Appendix
9.1 Publications in journals
1. P. Correa, J. Rodriguez, M. Rivera, J. Espinoza, J. Kolar, Predictive con-
trol of an indirect matrix converter, IEEE Transactions on Industrial
Electronics, Vol. 56 No 6, pp. 1847-1853; June 2009. (Published)
2. J. Rodriguez, P. Wheeler, B. Wu, J. Espinoza, M. Rivera, C. Rojas, Predic-
tive current control with resonance mitigation in a direct matrix
converter, IEEE Transactions on Power Electronics, 2011. (Published)
3. J. Rodriguez, P. Wheeler, M. Rivera, C. Rojas, A. Wilson, Control of a
matrix converter with imposed sinusoidal input currents, SS on Ma-
trix Converters, IEEE Transactions on Industrial Electronics, 2011. (Under
review)
4. J. Rodriguez, P. Wheeler, J. Kolar, M. Rivera, A review of control and
modulation methods for matrix converters, SS on Matrix Converters,
IEEE Transactions on Industrial Electronics, 2011. (Under review)
5. J. Rodriguez, M. Rivera, C. Rojas, Current control for an indirect ma-
trix converter with filter resonance mitigation, SS on Matrix Con-
verters, IEEE Transactions on Industrial Electronics, 2011. (Under review)
6. M. Rivera, J. Rodriguez, J. Espinoza, C. Rojas, A. Wilson, Imposed sinu-
soidal source and load currents for an indirect matrix converter, SS
103
Appendix 104
on Matrix Converters, IEEE Transactions on Industrial Electronics, 2011.
(Under review)
7. A. Wilson, J. Rodriguez, P. Wheeler, L. Empringham, C. Rojas, M. Rivera,
An assessment of model predictive current control and space vec-
tor modulation in a direct matrix converter, SS on Matrix Converters,
IEEE Transactions on Industrial Electronics, 2011. (Under review)
8. J. Espinoza, F. Villarroel, C. Rojas, J. Rodriguez, M. Rivera, Finites states
model predictive control with fuzzy decision making applied to a
direct matrix converter, SS on Matrix Converters, IEEE Transactions
on Industrial Electronics, 2011. (Under review)
9.2 Publications in conferences
1. M. Rivera, J. Espinoza, R. Vargas, J. Rodriguez, Behavior of the pre-
dictive DTC based matrix converter under unbalanced ac supply,
Industrial Applications Society, Annual General Meeting, IAS 2007, New
Orleans, USA. (Published)
2. M. Rivera, R. Vargas, J. Espinoza, J. Rodriguez, C. Silva, Current con-
trol in matrix converters connected to polluted ac voltage supplies,
Power Electronics Specialists Conference, PESC 2008, Rhodes, Greece. (Pub-
lished)
3. R. Vargas, M. Rivera, J. Rodriguez, J. Espinoza, P. Wheeler, Predictive
torque control with input PF correction applied to an induction
machine fed by a matrix converter, Power Electronics Specialists Con-
ference, PESC 2008, Rhodes, Greece. (Published)
4. M. Rivera, P. Correa, J. Rodriguez, I. Lizama, J. Espinoza, Predictive
control of the indirect matrix converter with active damping, 6th
International Power Electronics and Motion Control Conference, IPEMC
2009, Wuhan, China. (Published)
Appendix 105
5. I. Lizama, J. Rodriguez, B. Wu, P. Correa, M. Rivera, M. Perez, Predic-
tive control for current source rectifiers operating at low switching
frequency, 6th International Power Electronics and Motion Control Con-
ference, IPEMC 2009, Wuhan, China. (Published)
6. M. Rivera, P. Correa, J. Rodriguez, I. Lizama, J. Espinoza, C. Rojas, Pre-
dictive control of the indirect matrix converter with active damp-
ing, Energy Conversion Congress and Expo, ECCE 2009, California,
USA. (Published)
7. J. Rodriguez, J. Espinoza, M. Rivera, F. Villarroel, C. Rojas, Predictive
control of source and load currents in a direct matrix converter,
IEEE International Conference on Industrial Technology, ICIT 2010, Val-
paraıso, Chile. (Published)
8. J. Rodriguez, J. Kolar, J. Espinoza, M. Rivera, C. Rojas, Predictive cur-
rent control with reactive power minimization in an indirect ma-
trix converter, IEEE International Conference on Industrial Technology,
ICIT 2010, Valparaıso, Chile. (Published)
9. J. Rodriguez, J. Kolar, J. Espinoza, M. Rivera, C. Rojas, Predictive torque
and flux control of an induction machine fed by an indirect matrix
converter, IEEE International Conference on Industrial Technology, ICIT
2010, Valparaıso, Chile. (Published)
10. J. Rodriguez, B. Wu, M. Rivera, A. Wilson, V. Yaramasu and C. Rojas,
Model predictive control of three-phase four-leg neutral-point-
clamped inverters, IEEE International Power Electronics Conference, IPEC
2010, Sapporo, Japan. (Published)
11. M. Rivera, J. Rodrıguez, Predictive control of an indirect matrix con-
verter, IEEE International Conference on Networking, Sensing and Control,
ICNSC 2010, Chicago, USA. (Published)
12. J. Rodriguez, J. Kolar, J. Espinoza, M. Rivera, C. Rojas, Predictive con-
trol of a direct matrix converter operating under an unbalanced
Appendix 106
ac source, IEEE International Symposium on Industrial Electronics, ISIE
2010, Bari, Italy. (Published)
13. J. Rodriguez, J. Kolar, J. Espinoza, M. Rivera, C. Rojas, Predictive torque
and flux control of an induction machine fed by an indirect matrix
converter, IEEE International Symposium on Industrial Electronics, ISIE
2010, Bari, Italy. (Published)
14. V. Yaramasu, J. Rodriguez, B. Wu, M. Rivera, A. Wilson, C. Rojas, A
simple and effective solution for superior performance in two-level
four-leg voltage source inverters: predictive voltage control, IEEE
International Symposium on Industrial Electronics, ISIE 2010, bari, Italy.
(Published)
15. J. Rodriguez, B. Wu, M. Rivera, C. Rojas, V. Yaramasu, A. Wilson, Predic-
tive current control of three-phase two-level four-leg inverters, In-
ternational Power Electronics and Motion Control Conference, EPE-PEMC
2010, Ohrid, Macedonia. (Published)
16. M. Rivera, J.L. Elizondo, M. E. Macias, O.M. Probst, O.M. Micheloud,
J. Rodriguez, C. Rojas and A. Wilson, Model predictive control of a
double fed induction generator DFIG with and indirect matrix
converter, IECON 2010, Phoenix, USA. (Published)
17. J.L. Elizondo, M. Rivera, M. E. Macias, O.M. Probst, M. Oliver, O.M Mich-
eloud, J. Rodriguez, Model predictive control of a double fed induc-
tion generator with and indirect matrix converter, 41 Principal
Congreso de Investigacion y Desarrollo del Tecnologico de Monterrey, Jan.
2011, Mexico. (Published)
18. M. Rivera, I. Contreras, J. Rodriguez, R. Pena, A simple current con-
trol method for four-leg indirect matrix converters, IEEE European
Conference on Power Electronics and Applications, EPE 2011, Birmingham,
England. (To be published)
Appendix 107
9.3 Projects related with the research
1. FONDECYT 108 0059, Control of Indirect Matrix Converters, J. Ro-
driguez.
2. Basal Project FB021, Predictive Control in a Four-Leg Indirect Ma-
trix Converter, May 2010 - June 2011, M. Rivera.
3. Basal Project FB021, Optimizacion de la operacion de un convertidor
matricial indirecto ante una red distorsionada, May 2011 - June 2012,
M. Rivera.
4. FONDECYT 110 0404, High Performance Control of Electrical Ma-
chines, J. Rodriguez.
Appendix 108
9.4 Experimental setup circuit diagram
In Fig. 9.1 depicts a view of the experimental setup implemented in the laboratory,
which consists of a IMC prototype designed and built by Universidad Tecnica Fe-
derico Santa Marıa, thanks to the support of the Power Electronics Systems Lab-
oratory of ETH in Zurich. This converter features IGBTs of type IXRH40N120 for
the bidirectional switch, standard IGBTs with anti-parallel diodes IRG4PC30UD
for the inverter stage. The control scheme was implemented in a dSPACE 1103
which is connected to additional boards that include the FPGA for the commu-
tation sequence generation and the signal conditioning for the measurement of
voltages and currents.
Fig. 9.2 to Fig. 9.13 present schematic diagrams of the different components con-
sidered in the setup. Fig. 9.2 shows the schematic scheme of the input filter which
consists of an RLC filter with the options to connect an additional inductor in
parallel to the resistor and a resistor in series to the capacitor. As shown in the
schematic, this input filter is connected to the IMC throughout a harting connec-
tor. As mentioned before, this converter is controller with a dSPACE which sends
the switch signals to the FPGA, for this reason it is necessary to add an adapter
between the dSPACE and FPGA. The schematic of this adapter is indicated in
Fig. 9.3. A special card for the FPGA was built with different components as
shown in Fig. 9.4 to Fig. 9.10. This card works with different voltage levels so,
it is necessary include voltage regulators to allow operate with 3.3 V, 2.5 V, and
1.2 V as shown in Fig. 9.4 and Fig. 9.5. A PROM memory is necessary in or-
der to store the code for the commutation and other necessary information in
the FPGA. The schematics of this memory are given in Fig. 9.6. DIP-switches
and some leds have been included and their schematics are indicated in Fig. 9.7.
The schematic of analog-digital converters and digital-analog converters needed
for the implementation are included in Fig. 9.8 and Fig. 9.9, respectively. The
schematics of the input/output connections in the FPGA are shown in Fig. 9.10.
The schematics of the main card which include the power circuit of the IMC and
voltage measurements are indicated in Fig. 9.11 to Fig. 9.13.
Appendix 109
The schematic of the IMC with its rectifier and inverter stages is shown in Fig.
9.12. Finally, the voltage measurements on the input side, are included in the
main card and these signals are directly connected to the dSPACE throughout
SMB connectors as shown in the schematic of Fig. 9.13.
Figure 9.1. Experimental setup in the laboratory.
Appendix 110
100pF
Cfa
10mH
Lfc
100pF
Cfb
100pF
Cfc
10mH
Lfb
10mH
Lfa
12
34
56
78
910
1112
1314
15
Con1
Conector Harting
100pF
Cfc2
100pF
Cfb2
100pF
Cfa2
10mHLfc2
1KRfc
10mHLfb2
10mHLfa2
1K Rfb
1KRfa
1KRfc2
1KRfb2
1KRfa2
Figure 9.2. Schematic of the input filter used in the implementation.
Appen
dix
111
1
1
2
2
3
3
4
4
D D
C C
B B
A A
Title
Number RevisionSize
A4
Date: 19-05-2011 Sheet ofFile: M:\1 - Diseño y armado del IMC\..\Dspace.SchDocDrawn By:
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
Con3
DB-50
1
20
2
21
3
22
4
23
5
24
6
25
7
8
9
10
11
12
13
14
15
16
17
18
19
27
26
Con1
D Connector 25
1B1
1A2
1Y3
2A6
G4
VCC16
3Y11
2Y5
2B7
GND8
4B15
4A14
4Y13
G12
3A10
3B9
U3
SN75173D
4A19
1A1
1Y2
4Y18
NC3
NC17
1Z4
4Z16
EN5
EN15
2Z6
3Z14
NC7
NC13
2Y8
3Y12
2A9
3A11
VCC20
GND10
U1
MC75172BDW
4A19
1A1
1Y2
4Y18
NC3
NC17
1Z4
4Z16
EN5
EN15
2Z6
3Z14
NC7
NC13
2Y8
3Y12
2A9
3A11
VCC20
GND10
U2
MC75172BDW
RU11
CU1
DB15 20DB15 8
DB15 21DB15 9
DB15 22DB15 10DB15 23DB15 11DB15 24DB15 12DB15 25
DB15 19DB15 7
DB15 18DB15 6
DB15 15DB15 3
DB15 16DB15 4
DB15 17DB15 5
DB15 14DB15 2
DB15 1
GND DB50 2
DB50 3
DB50 4
DB50 5
DB50 6
DB50 7
DB50 8
DB50 9
DB50 11
DB50 12
DB50 13
DB50 14
DB50 15
DB50 16
DB50 17
DB50 21
DB50 22
DB50 23
DB50 24
DB50 25
DB50 26
DB50 27
DB50 28
DB50 29
DB50 31
DB50 32
DB50 33
DB50 18
DB50 19
DB50 20
DB50 10
DB50 30
DB50 34
DB50 35
DB50 36
DB50 37
DB50 38
DB50 39
DB50 40
DB50 41
DB50 42
DB50 43
DB50 44
DB50 45
DB50 46
DB50 47
DB50 48
DB50 49
1234
Tx1
Conector Fibra
1234
Tx2
Conector Fibra
1234
Rx
Conector Fibra
RU12
RU13
RU14
RU21
RU22
RU23
RU24
CU2CU3
RU31
RU32
RU33
RU34
5GND
Rrx
5
GNDGND
Rtx12
Rtx115
Qtx1QNPN
GND
GND
Rtx22
Rtx215
Qtx2QNPN
GND
GND
5
5
GND
5
5
5GND
GND
GND
5 5 5
GND GND GND
DB15 12DB15 25
DB15 25DB15 12
DB15 11DB15 24
DB15 24DB15 11
DB15 10DB15 23
DB15 10 DB15 23
DB15 9DB15 22
DB15 22DB15 9
DB15 8DB15 21
DB15 21DB15 8
DB15 20DB15 7
DB15 7 DB15 20
DB15 19DB15 6
DB15 6 DB15 19
DB15 18DB15 5
DB15 5 DB15 18
DB15 4DB15 17
DB15 4 DB15 17
DB15 3DB15 16
DB15 16DB15 3
DB15 15DB15 2
DB15 2 DB15 15
DB15 14DB15 1
DB15 14DB15 1
DB
50 3
4
DB
50 3
5
DB
50 3
6
DB
50 3
7
DB
50 3
8
DB
50 3
9
DB
50 4
0
DB
50 4
1
DB
50 4
2
DB
50 4
3
DB
50 4
4
DB
50 4
5
GND
DB
50 1
7
DB50 1
DB50 18
DB50 2
DB50 19
DB50 3
DB50 20
DB50 4
DB50 21
DB50 46
DB50 6
DB50 22
DB50 5
DB50 1
DB50 47
DB50 9
DB50 26
5
5
D1LED1
1KR
5
GND
Fig
ure
9.3
.Sch
ematic
ofth
eadapter
between
the
dSPA
CE
and
FPG
A.
Appendix 112
VCCINT70
VCCINT88
VCCINT174
VCCINT192
VCCAUX17
VCCAUX38
VCCAUX69
VCCAUX89
VCCAUX121
VCCAUX142
VCCAUX173
VCCAUX193
VCCO_76
VCCO_723
VCCO_632
VCCO_649
VCCO_560
VCCO_573
VCCO_484
VCCO_498
VCCO_3110
VCCO_3127
VCCO_2136
VCCO_2153
VCCO_1164
VCCO_1177
VCCO_0188
VCCO_0201
U7J
XC3S200-4PQ208C
4.7K R30
4.7K R28
330 R3
3.3
3.3
GND
47nF
C23
470nF
C24
4.7uF
C25
3.3
GND3.3
GND3.3
GND3.3
GND3.3
GND3.3
GND
1.2
GND
1nF
C50
47nF
C51
470nF
C52
4.7uF
C53
2.5
GND
1nF
C38
47nF
C39
470nF
C40
4.7uF
C41
3.3
GND
DriveDone=yes
1nF
C22
2.5
1nF
C49
1nF
C37
OE-INIT
CF-PROG
DONE-CE
47nF
C30
470nF
C31
4.7uF
C32
1nF
C29
47nF
C34
470nF
C35
4.7uF
C36
1nF
C33
47nF
C46
470nF
C47
4.7uF
C48
1nF
C45
47nF
C55
470nF
C56
4.7uF
C57
1nF
C54
47nF
C59
470nF
C60
4.7uF
C61
1nF
C58
47nF
C63
470nF
C64
4.7uF
C65
1nF
C62
47nF
C67
470nF
C68
4.7uF
C69
1nF
C66
GND
51
23
J20
PWR2.5
GND
Figure 9.4. Schematic of the FPGA with its different voltage levels.
Appen
dix
113
1
1
2
2
3
3
4
4
D D
C C
B B
A A
Title
Number RevisionSize
A4
Date: 19-05-2011 Sheet ofFile: M:\1 - Diseño y armado del IMC\..\FPGA reguladores.SchDocDrawn By:
4.7uF
C72
D8LED3
68pF
C74
GND
17KR721K
R44
2.2uF
C78
4.7uF
C77
4.7uF
C81
2.2uF
C95
2.2uF
C83
68pF
C80
10K
R75
GND GND
3.3
GNDGND
5
.4KR73
10K
R74
GND GND
1.2
GNDGND
5
10K
R67
GND GND
2.5
GNDGND
5
10KR76
.56KR77
10KR78
5
10KR79
5
10KR80
5
Vin2
SD1
Vout4
SENSE5
GN
D3
LP3966_ADJ
GN
D6
REG1
LP3966
Vin2
SD1
Vout4
SENSE5
GN
D3
LP3966_ADJG
ND
6REG2
LP3966
Vin2
SD1
Vout4
SENSE5
GN
D3
LP3966_ADJ
GN
D6
REG3
LP3966
0.1uF
C27
0.1uF
C28GND
5
GND
-Vin1
+Vin2
3
Com4
-Vo5 6
+Vo7
8
9
10
DC/DC
DC/DC Converter D
0.1uF
C42 15
-15
Reguladores de Voltaje de la tarjeta FPGA
11
22
33
Z2REF30XX
2.048AD10K
Rz5
GND
2.048AD
GND
100nF
Cz2
2.048AD
GND
1nF
Cz1
C5VCap Pol3
GND
5
Fig
ure
9.5
.Sch
ematics
ofvo
ltage
regula
tors
inclu
ded
inth
eFPG
A.
Appendix 114
M154
M055
M256
DONE103
CCLK104
TDO158
TCK159
TMS160
HSWAP_EN206
PROG_B207
TDI208
U7I
XC3S200-4PQ208C
D01
DNC2
CLK3
TDI4
TMS5
TCK6
CF7
OE/RESET8
DNC9
CE10
GND11
DNC12
CEO13
DNC14
DNC15
DNC16
TDO17
VCCINT18
VCCO19
VCCJ20
U6
XCF02SVO20CGND
GNDMaster -serial mode
Turn on pullof resistors of user IO during configuration
3.3
R23
100
R25 100GND
R27 100
3.3 Master serial configuration
DONE-CE
OE-INITCF-PROG
TCK_MTMS_M
TDI_MTCK_MTMS_M
TDO_M
Data_serial
R22 100
R24100
R26 100
TDI_M_in
TCK_M_inTMS_M_in
TDO_M_out
Figure 9.6. Schematic of the PROM included in the FPGA card.
202U7K
3.3
GNDGND
GCK0GCK1
BA
NK
4
IO/VREF_4102
IO_L01N_4/VRP_4101
IO_L01P_4/VRN_4100
IO97
IO/VREF_496
IO_L25N_495
IO_L25P_494
IO93
IO_L27N_4/DIN/D092
IO_L27P_4/D190
IO_L30N_4/D287
IO_L30P_4/D386
IO/VREF_485
IO_L31N_4/INIT_B83
IO_L31P_4/DOUT/BUSY81
IO_L32N_4/GCLK180
IO_L32P_4/GCLK079
U7E
XC3S200-4PQ208C
220 R29220 R31220 R32220 R33
4.7
kR
35
4.7
kR
36
4.7
kR
37
4.7
kR
38
1234 5
678
S1
SW-DIP4FPGA_90FPGA_87FPGA_86FPGA_85
FPGA_81
Data_serial
OE-INIT
Figure 9.7. Schematic of the led and DIP-switches included in the FPGA card.
Appendix 115
DOutA1
VDrive2
DVcc3
Range14
Range05
ADDR6
AGnd7
AVcc8
DCapA9
Vss10
Va111
Va212
Vb213
Vb114
Vdd15
DCapB16
AGnd17
RefSel18
CSNeg19
SClk20
CnvStNeg21
Busy22
DOutB23
DGnd24
CAD4
Conversor AD
5
10uF
C88
0.1uF
C89
0.1uF
C905
GND GND
GND 10uF
C91
0.1uF
C92
GND GND
680nF
C93
GND
680nF
C94
GND
10uFC102
0.1uFC103
GND GND
GNDGND
GND10uFC104
0.1uFC105
GND GND
5CAD4 DOutA
CAD4 ADDR
CAD4 DOutBCAD4 Busy
CAD4 CnvStNegCAD4 Sclk
CAD4 CSNeg
GND
CAD4 Va1 CAD4 Vb115-15
Sock1 2 Sock1 45
CAD4 R0
Figure 9.8. Schematic of an analog-digital converter CAD included in the
FPGA.
Din1
Clk2
CS3
OutA4
AGnd5
Ref6
OutB7
Vdd8
CDA1
Conversor Digital Análogo
5
220 R68
220 R70 100pFC96
100pFC98
GND
GND
CDA1 DinCDA1 ClkCDA1 CS
GND2.048AD
1
2
Cx1
1
2
Cx2GND
GND GND
Cz3
Figure 9.9. Schematic of a digital-analog converter CDA included in the
FPGA.
Appen
dix
116
1
1
2
2
3
3
4
4
D D
C C
B B
A A
Title
Number RevisionSize
A4
Date: 19-05-2011 Sheet ofFile: M:\1 - Diseño y armado del IMC\..\FPGA I-O.SchDocDrawn By:
8
1
4
3
2
AO1ATL082CD
84
75
6
AO1BTL082CD
8
1
4
3
2
AO2ATL082CD
84
75
6
AO2BTL082CD
8
1
4
3
2
AO3ATL082CD
84
75
6
AO3BTL082CD
8
1
4
3
2
AO4ATL082CD
84
75
6
AO4BTL082CD
15
15
15
15
15
15
15
15
-15
-15
-15
-15
-15
-15
-15
-15
R AO1R AO2
R AO3
R AO4
R AO5
GND
R AO6R AO7
R AO8
R AO9
R AO10
GND
R AO11R AO12
R AO13
R AO14
R AO15
GND
R AO16R AO17
R AO18
R AO19
R AO20
GND
C AO1
15GND
C AO2GND -15
C AO3
15GND
C AO4GND -15
R AO21R AO22
R AO23
R AO24
R AO25
GND
R AO26R AO27
R AO28
R AO29
R AO30
GND
C AO5
15GND
C AO6GND -15
R AO31R AO32
R AO33
R AO34
R AO35
GND
R AO36R AO37
R AO38
R AO39
R AO40
GND
C AO7
15GND
C AO8GND -15
CAD1 Vb1
CAD1 Va1
CAD2 Va1
CAD2 Vb1
CAD3 Vb1
CAD3 Va1
CAD4 Va1
CAD4 Vb1
Sock3 1
Sock3 18
Sock3 2
Sock3 17
Sock3 6
Sock3 13
Sock3 7
Sock3 12
Sock3 16
Sock3 3
Sock3 8
Sock3 11
Sock3 4
Sock3 15
Sock3 10
Sock3 9
Sock3 1
Sock3 16Sock3 2
Sock3 15
Sock3 9
Sock3 12Sock3 6
Sock3 11
Sock3 17Sock3 3
Sock3 7
Sock3 10
Sock3 4
Sock3 13
Sock3 18
Sock3 8
Sock1 46Sock1 45Sock1 44Sock1 43
Sock1 1Sock1 2Sock1 3Sock1 4Sock1 5Sock1 6
Sock1 8Sock1 9Sock1 10Sock1 11Sock1 12Sock1 13Sock1 14Sock1 15Sock1 16Sock1 17Sock1 18Sock1 19Sock1 20Sock1 21Sock1 22Sock1 23 Sock1 24
Sock1 25Sock1 26
Sock1 42Sock1 41
Sock1 39Sock1 38Sock1 37Sock1 36Sock1 35Sock1 34Sock1 33Sock1 32Sock1 31Sock1 30Sock1 29Sock1 28Sock1 27
Sock2 1Sock2 2
Sock2 5Sock2 6
Sock2 8Sock2 9Sock2 10Sock2 11Sock2 12Sock2 13Sock2 14Sock2 15Sock2 16Sock2 17Sock2 18Sock2 19Sock2 20Sock2 21Sock2 22Sock2 23 Sock2 24
Sock2 25Sock2 26
Sock2 46Sock2 45
Sock2 42Sock2 41
Sock2 39Sock2 38Sock2 37Sock2 36Sock2 35Sock2 34Sock2 33Sock2 32Sock2 31Sock2 30Sock2 29Sock2 28Sock2 27
Componentes de entrada y salida de la tarjeta FPGA
123456789
1011121314151617181920212223 24
25262728293031323334353637383940414243444546
Sock1
Vertical Socket 23x2
123456789
1011121314151617181920212223 24
25262728293031323334353637383940414243444546
Sock2
Vertical Socket 23x2
1234
Rx2
Conector Fibra
1234
Rx1
Conector Fibra
5GND
Rrx1
3.3
Fibra Rx1
Rrx2
Fibra Rx2
12345678910
1112131415161718
Sock3
Vertical Socket 9x2
GNDGND
5GND
3.3
GND GND GNDGND
53.3
53.3
1234
Tx
Conector Fibra
R01A C01A
GND
R01B C01B
GND
R02A C02A
GND
R02B C02B
GND
R03A C03A
GND
R03B C03B
GND
R04A C04A
GND
R04B C04B
GND
GND
GND
Rtx2
Rtx15
Fibra TxQtxQNPN
Fig
ure
9.1
0.Sch
ematics
ofin
put/
outp
ut
com
ponen
tsin
cluded
inth
eFPG
A.
Appen
dix
117
1
1
2
2
3
3
4
4
D D
C C
B B
A A
Title
Number RevisionSize
A4
Date: 19-05-2011 Sheet ofFile: M:\1 - Diseño y armado del IMC\..\Circuito de Potencia.SchDocDrawn By:
QApIGBT-N
Input Phase u Input Phase v Input Phase w
Output Phase u 1
Output Phase v 1
Output Phase w 1
p
n
Supa Gate Svpa Gate Swpa Gate
Supb Gate Svpb Gate Swpb Gate
Suna Gate Svna Gate Swna Gate
Sunb Gate Svnb Gate Swnb Gate
Sup Gate
Sun Gate
Svn Gate
Svp Gate
Swp Gate
Swn Gate
Supa Emitter Svpa Emitter Swpa Emitter
Suna Emitter Svna Emitter Swna Emitter
Supb Emitter Svpb Emitter Swpb Emitter
Sunb Emitter Svnb Emitter Swnb Emitter
Sun Emitter
Svn Emitter
Swn Emitter
Swp Emitter
Svp Emitter
Sup Emitter
Circuito de Potencia de Convertidor
1
23
QapaIGBT IXRH
1
23 Qapb
IGBT IXRH
1
23
QbpaIGBT IXRH
1
23 Qbpb
IGBT IXRH
1
23
QcpaIGBT IXRH
1
23 Qcpb
IGBT IXRH
1
23
QanaIGBT IXRH
1
23 Qanb
IGBT IXRH
1
23
QbnaIGBT IXRH
1
23 Qbnb
IGBT IXRH
1
23
QcnaIGBT IXRH
1
23 Qcnb
IGBT IXRH
QAnIGBT-N
QBpIGBT-N
QBnIGBT-N
QCpIGBT-N
QCnIGBT-N
Fig
ure
9.1
1.Sch
ematics
ofth
epo
wer
circuit
inth
eIM
C.
Appendix 118
Swpa Emitter
Swnb Emitter
Swpa Gate
Swpb Gate100nF
Ccpa
10ohm
Rcpa
100nF
Ccpb
10ohm
Rcpb
1k
RcpaL
Dcpa
1k
RcpbL
Dcpb
Swpa Emitter
Swpb Emitter
100nF
Cc
5Vgnd
5V100nF
Cc2
3 4
UnB
SN74LS06D
11 10
UnE
SN74LS06D
270ohmRcpaD
5V
5Vgnd
270ohm
RcpbD5V
5Vgnd
Nwpab
Nwpab
1
2
3
4 5
6
7
8
DC/DCc
DC/DC Converter
N/C1
Anode2
Cathode3
N/C4
VEE5
Vo6
Vo7
VCC8
DRVcpa
Driver Single
N/C1
Anode2
Cathode3
N/C4
VEE5
Vo6
Vo7
VCC8
DRVcpb
Driver Single
5V
gnd
Figure 9.12. Schematic of the driver implemented in the IMC for the switches.
Appendix 119
8
1
4
3
2
AO2ATL082CD
84
75
6
AO2BTL082CD
15
15
-15
-15
R AO12
R AO13
R AO14
R AO15
R AO17
R AO18
R AO19
R AO20
C AO3
15
C AO4-15
5Vgnd
5Vgnd
5Vgnd
5Vgnd
5V
gn
d5
Vg
nd
Cx4SMB
Cx3
SMB
R02A C02A
R02B C02B
5V
gn
d5
Vg
nd
R AOA
R AOB
Muvs
Mvus
Mvws
Mwvs
So
ck3
18
So
ck3
17
Figure 9.13. Schematic of the implementation of voltage measurement in
the IMC.
Bibliography
[1] E. Yamamoto, T. Kume, H. Hara, T. Uchino, J. Kang, and H. Krug, “De-
velopment of matrix converter ans its applications in industry,” 35th Annual
Conference of the IEEE Industrial Electronics Society IECON 2009, Porto,
Portugal, 2009.
[2] P. Wheeler, J. Rodriguez, J. Clare, L. Empringham, and A. Weinstein, “Ma-
trix converters: a technology review,” Industrial Electronics, IEEE Trans-
actions on, vol. 49, no. 2, pp. 276 –288, 2002.
[3] M. Venturini, “A new sine wave in sine wave out, conversion technique which
eliminates reactive elements,” in Proc. POWERCON 7, 1980.
[4] M. Venturini and A. Alesina, “The generalized transformer: A new bidirec-
tional sinusoidal waveform frequency converter with continuously adjustable
input power factor,” in Proc. IEEE PESC’80.
[5] J. Rodriguez, “A new control technique for ac-ac converters,” in Proc.
IFAC Control in Power Electronics and Electrical Drives Conf., Lausanne,
Switzerland, 1983.
[6] J. Oyama, T. Higuchi, E. Yamada, T. Koga, and T. Lipo, “New control
strategy for matrix converter,” Power Electronics Specialists Conference,
1989. PESC ’89 Record., 20th Annual IEEE, pp. 360 –367 vol.1, Jun. 1989.
120
Bibliography 121
[7] L. Wei and T. Lipo, “A novel matrix converter topology with simple commu-
tation,” Industry Applications Conference, 2001. Thirty-Sixth IAS Annual
Meeting. Conference Record of the 2001 IEEE, 2001.
[8] M. Hamouda, F. Fnaiech, and K. Al-Haddad, “Space vector modulation
scheme for dual-bridge matrix converters using safe-commutation strategy,”
Industrial Electronics Society, 2005. IECON 2005. 31st Annual Conference
of IEEE, p. 6 pp., 2005.
[9] L. Wei, T. Lipo, and H. Chan, “Matrix converter topologies with reduced
number of switches,” Power Electronics Specialists Conference, 2002. pesc
02. 2002 IEEE 33rd Annual, 2002.
[10] J. Kolar and F. Schafmeister, “Novel modulation schemes minimizing the
switching losses of sparse matrix converters,” Industrial Electronics Society,
2003. IECON ’03. The 29th Annual Conference of the IEEE, vol. 3, pp.
2085 – 2090 Vol.3, 2003.
[11] J. Igney and M. Braun, “A new matrix converter modulation strategy max-
imizing the control range,” Power Electronics Specialists Conference, 2004.
PESC 04. 2004 IEEE 35th Annual, 2004.
[12] ——, “Space vector modulation strategy for conventional and indirect ma-
trix converters,” Power Electronics and Applications, 2005 European Con-
ference on, 0 2005.
[13] S. Kim, S.-K. Sul, and T. Lipo, “Ac/ac power conversion based on ma-
trix converter topology with unidirectional switches,” Industry Applications,
IEEE Transactions on, vol. 36, no. 1, pp. 139 –145, 2000.
[14] J. Itoh, I. Sato, A. Odaka, H. Ohguchi, H. Kodachi, and N. Eguchi, “A novel
approach to practical matrix converter motor drive system with reverse
blocking igbt,” Power Electronics, IEEE Transactions on, vol. 20, no. 6,
pp. 1356 – 1363, 2005.
Bibliography 122
[15] F. Schafmeister and J. Kolar, “Novel modulation schemes for conventional
and sparse matrix converters facilitating reactive power transfer indepen-
dent of active power flow,” Power Electronics Specialists Conference, 2004.
PESC 04. 2004 IEEE 35th Annual, 2004.
[16] M. Jussila and H. Tuusa, “Comparison of simple control strategies of space-
vector modulated indirect matrix converter under distorted supply voltage,”
Power Electronics, IEEE Transactions on, vol. 22, no. 1, pp. 139 –148, 2007.
[17] C. Klumpner, “A hybrid indirect matrix converter immune to unbalanced
voltage supply, with reduced switching losses and improved voltage transfer
ratio,” Applied Power Electronics Conference and Exposition, 2006. APEC
’06. Twenty-First Annual IEEE, p. 7 pp., 2006.
[18] S. Kouro, P. Cortes, R. Vargas, U. Ammann, and J. Rodriguez, “Model pre-
dictive control: A simple and powerful method to control power converters,”
Industrial Electronics, IEEE Transactions on, vol. 56, no. 6, pp. 1826 –1838,
2009.
[19] B. Wu, High-Power Converters and AC Drives, 2006.
[20] I. Takahashi and Y. Itoh, “Electrolytic capacitor-less pwm inverter,” Proc.
International Power Electronics Conference IPEC 90, pp. 131 – 138, Apr.
1990.
[21] M. Baumann and J. Kolar, “Comparative evaluation of modulation methods
for a three-phase/switch buck power factor corrector concerning the input
capacitor voltage ripple,” Power Electronics Specialists Conference, 2001.
PESC. 2001 IEEE 32nd Annual, 2001.
[22] L. Zhang, C. Watthanasarn, and W. Shepherd, “Control of ac-ac matrix con-
verters for unbalanced and/or distorted supply voltage,” Power Electronics
Specialists Conference, 2001. PESC. 2001 IEEE 32nd Annual, vol. 2, pp.
1108 –1113 vol.2, 2001.
Bibliography 123
[23] M. Rivera, R. Vargas, J. Espinoza, and J. Rodriguez, “Behavior of the
predictive dtc based matrix converter under unbalanced ac-supply,” Power
Electronics Specialists Conference, 2008. PESC 2008. IEEE, pp. 202 –207,
Sep. 2007.
[24] K. Mohapatra and N. Mohan, “Open-end winding induction motor driven
with matrix converter for common-mode elimination,” Power Electronics,
Drives and Energy Systems, 2006. PEDES ’06. International Conference
on, pp. 1 –6, 2006.
[25] J. Kolar, T. Friedli, F. Krismer, and S. Round, “The essence of three-phase
ac/ac converter systems,” Power Electronics and Motion Control Confer-
ence, 2008. EPE-PEMC 2008. 13th, pp. 27 –42, 2008.
[26] P. Correa, J. Rodriguez, M. Rivera, J. Espinoza, and J. Kolar, “Predic-
tive control of an indirect matrix converter,” Industrial Electronics, IEEE
Transactions on, vol. 56, no. 6, pp. 1847 –1853, 2009.
[27] P. Zanchetta, P. Wheeler, J. Clare, M. Bland, L. Empringham, and D. Kat-
sis, “Control design of a three-phase matrix-converter-based ac/ac mobile
utility power supply,” Industrial Electronics, IEEE Transactions on, vol. 55,
no. 1, pp. 209 –217, 2008.
[28] B.-H. Kwon, B.-D. Min, and J.-H. Kim, “Novel topologies of ac choppers,”
Electric Power Applications, IEE Proceedings -, vol. 143, no. 4, pp. 323 –330,
Jul. 1996.
[29] P. Ziogas, D. Vincenti, and G. Joos, “A practical pwm ac controller topol-
ogy,” Industry Applications Society Annual Meeting, 1992., Conference
Record of the 1992 IEEE, pp. 880 –887 vol.1, Oct. 1992.
[30] J. Kolar, M. Baumann, F. Schafmeister, and H. Ertl, “Novel three-phase ac-
dc-ac sparse matrix converter,” Applied Power Electronics Conference and
Exposition, 2002. APEC 2002. Seventeenth Annual IEEE, 2002.
Bibliography 124
[31] J. Kolar, F. Schafmeister, S. Round, and H. Ertl, “Novel three-phase
ac/ac sparse matrix converters,” Power Electronics, IEEE Transactions on,
vol. 22, no. 5, pp. 1649 –1661, 2007.
[32] C. Klumpner, M. Lee, and P. Wheeler, “A new three-level sparse indirect
matrix converter,” IEEE Industrial Electronics, IECON 2006 - 32nd Annual
Conference on, pp. 1902 –1907, 2006.
[33] M. Y. Lee, P. Wheeler, and C. Klumpner, “A new modulation method for the
three-level-output-stage matrix converter,” Power Conversion Conference -
Nagoya, 2007. PCC ’07, pp. 776 –783, 2007.
[34] J. Kolar, H. Ertl, and F. Zach, “Design and experimental investigation of
a three-phase high power density high efficiency unity power factor pwm
(vienna) rectifier employing a novel integrated power semiconductor mod-
ule,” Applied Power Electronics Conference and Exposition, 1996. APEC
’96. Conference Proceedings 1996., Eleventh Annual, vol. 2, pp. 514 –523
vol.2, Mar. 1996.
[35] R. Erickson and O. Al-Naseem, “A new family of matrix converters,” Indus-
trial Electronics Society, 2001. IECON ’01. The 27th Annual Conference of
the IEEE, 2001.
[36] C. Klumpner, “Hybrid direct power converters with increased/higher than
unity voltage transfer ratio and improved robustness against voltage supply
disturbances,” Power Electronics Specialists Conference, 2005. PESC ’05.
IEEE 36th, pp. 2383 –2389, 2005.
[37] J. Rodriguez, E. Silva, F. Blaabjerg, P. Wheeler, J. Clare, and J. Pontt, “Ma-
trix converter controlled with the direct transfer function approach: analy-
sis, modelling and simulation,” Taylor and Francis-International Journal of
Electronics, vol. 92, no. 2, pp. 63 –85, Feb. 2005.
[38] S. Bernet, S. Ponnaluri, and R. Teichmann, “Design and loss comparison
of matrix converters, and voltage-source converters for modern ac drives,”
Bibliography 125
Industrial Electronics, IEEE Transactions on, vol. 49, no. 2, pp. 304 –314,
Apr. 2002.
[39] J. Rzasa, “Capacitor clamped multilevel matrix converter controlled with
venturini method,” Power Electronics and Motion Control Conference,
2008. EPE-PEMC 2008. 13th, pp. 357 –364, Sep. 2008.
[40] Y. Mei and L. Huang, “Improved switching function modulation strategy
for three-phase to single-phase matrix converter,” Power Electronics and
Motion Control Conference, 2009. IPEMC ’09. IEEE 6th International, pp.
1734 –1737, May. 2009.
[41] S. Lopez Arevalo, P. Zanchetta, P. Wheeler, A. Trentin, and L. Empringham,
“Control and implementation of a matrix-converter-based ac-ground power-
supply unit for aircraft servicing,” Industrial Electronics, IEEE Transac-
tions on, vol. 57, no. 6, pp. 2076 –2084, Jun. 2010.
[42] G. Roy and G.-E. April, “Cycloconverter operation under a new scalar con-
trol algorithm,” Power Electronics Specialists Conference, 1989. PESC ’89
Record., 20th Annual IEEE, pp. 368 –375 vol.1, Jun. 1989.
[43] J. Rodriguez, “High performance dc motor drive using a pwm rectifier with
power transistors,” Electric Power Applications, IEE Proceedings B, vol.
134, no. 1, p. 9, Jan. 1987.
[44] J. Itoh, I. Sato, A. Odaka, H. Ohguchi, H. Kodatchi, and N. Eguchi, “A novel
approach to practical matrix converter motor drive system with reverse
blocking igbt,” Power Electronics Specialists Conference, 2004. PESC 04.
2004 IEEE 35th Annual, vol. 3, pp. 2380 – 2385 Vol.3, Jun. 2004.
[45] K. Mohapatra, P. Jose, A. Drolia, G. Aggarwal, and S. Thuta, “A novel
carrier-based pwm scheme for matrix converters that is easy to implement,”
Power Electronics Specialists Conference, 2005. PESC ’05. IEEE 36th, pp.
2410 –2414, Jun. 2005.
Bibliography 126
[46] Z. Idris, M. Hamzah, and A. Omar, “Implementation of single-phase matrix
converter as a direct ac-ac converter synthesized using sinusoidal pulse width
modulation with passive load condition,” Power Electronics and Drives Sys-
tems, 2005. PEDS 2005. International Conference on, vol. 2, pp. 1536 –
1541, Nov. 2005.
[47] Y.-D. Yoon and S.-K. Sul, “Carrier-based modulation technique for matrix
converter,” Power Electronics, IEEE Transactions on, vol. 21, no. 6, pp.
1691 –1703, Nov. 2006.
[48] T. Satish, K. Mohapatra, and N. Mohan, “Steady state over-modulation of
matrix converter using simplified carrier based control,” Industrial Electron-
ics Society, 2007. IECON 2007. 33rd Annual Conference of the IEEE, pp.
1817 –1822, Nov. 2007.
[49] P. Loh, R. Rong, F. Blaabjerg, L. Shan, and P. Wang, “Carrier-based modu-
lation schemes for various three-level matrix converters,” Power Electronics
Specialists Conference, 2008. PESC 2008. IEEE, pp. 1720 –1726, Jun. 2008.
[50] S. Thuta, K. Mohapatra, and N. Mohan, “Matrix converter over-modulation
using carrier-based control: Maximizing the voltage transfer ratio,” Power
Electronics Specialists Conference, 2008. PESC 2008. IEEE, pp. 1727 –1733,
Jun. 2008.
[51] Y. Miura, S. Horie, S. Kokubo, T. Amano, T. Ise, T. Momose, and Y. Sato,
“Application of three-phase to single-phase matrix converter to gas engine
cogeneration system,” Energy Conversion Congress and Exposition, 2009.
ECCE 2009. IEEE, pp. 3290 –3297, Sep. 2009.
[52] P. C. Loh, R. Rong, F. Blaabjerg, and P. Wang, “Digital carrier modula-
tion and sampling issues of matrix converters,” Power Electronics, IEEE
Transactions on, vol. 24, no. 7, pp. 1690 –1700, Jul. 2009.
[53] T. Sanada, S. Ogasawara, and M. Takemoto, “Experimental evaluation of
a new pwm control scheme for matrix converters,” Power Electronics Con-
ference (IPEC), 2010 International, pp. 2366 –2371, Jun. 2010.
Bibliography 127
[54] P. Kiatsookkanatorn and S. Sangwongwanich, “A unified pwm strategy for
matrix converters and its dipolar pwm realization,” Power Electronics Con-
ference (IPEC), 2010 International, pp. 3072 –3079, Jun. 2010.
[55] T. Yamashita and T. Takeshita, “Pwm strategy of single-phase to three-
phase matrix converters for reducing a number of commutations,” Power
Electronics Conference (IPEC), 2010 International, pp. 3057 –3064, Jun.
2010.
[56] J. Itoh and K. Maki, “Single-pulse operation for a matrix converter synchro-
nized with the output frequency,” Power Electronics Conference (IPEC),
2010 International, pp. 2781 –2788, Jun. 2010.
[57] S. Ishii, H. Hara, T. Higuchi, T. Kawachi, K. Yamanaka, N. Koga, T. Kume,
and J. Kang, “Bidirectional dc-ac conversion topology using matrix con-
verter technique,” Power Electronics Conference (IPEC), 2010 Interna-
tional, pp. 2768 –2773, Jun. 2010.
[58] P. Potamianos, E. Mitronikas, and A. Safacas, “A modified carrier-based
modulation method for three-phase matrix converters,” Power Electronics
and Applications, 2009. EPE ’09. 13th European Conference on, pp. 1 –9,
Sep. 2009.
[59] Y.-D. Yoon and S.-K. Sul, “Carrier-based modulation method for matrix
converter with input power factor control and under unbalanced input
voltage conditions,” Applied Power Electronics Conference, APEC 2007 -
Twenty Second Annual IEEE, pp. 310 –314, Mar. 2007.
[60] L. Huber and D. Borojevic, “Space vector modulated three-phase to three-
phase matrix converter with input power factor correction,” Industry Ap-
plications, IEEE Transactions on, vol. 31, no. 6, pp. 1234 –1246, Nov. 1995.
[61] F. Blaabjerg, D. Casadei, C. Klumpner, and M. Matteini, “Comparison of
two current modulation strategies for matrix converters under unbalanced
input voltage conditions,” Industrial Electronics, IEEE Transactions on,
vol. 49, no. 2, pp. 289 –296, Apr. 2002.
Bibliography 128
[62] O. Simon, J. Mahlein, M. Muenzer, and M. Bruckmarm, “Modern solutions
for industrial matrix-converter applications,” Industrial Electronics, IEEE
Transactions on, vol. 49, no. 2, pp. 401 –406, Apr. 2002.
[63] D. Casadei, G. Serra, A. Tani, and L. Zarri, “Matrix converter modulation
strategies: a new general approach based on space-vector representation of
the switch state,” Industrial Electronics, IEEE Transactions on, vol. 49,
no. 2, pp. 370 –381, Apr. 2002.
[64] L. Helle, K. Larsen, A. Jorgensen, S. Munk-Nielsen, and F. Blaabjerg, “Eval-
uation of modulation schemes for three-phase to three-phase matrix convert-
ers,” Industrial Electronics, IEEE Transactions on, vol. 51, no. 1, pp. 158 –
171, Feb. 2004.
[65] D. Casadei, G. Serra, A. Tani, and L. Zarri, “A novel modulation strategy
for matrix converters with reduced switching frequency based on output
current sensing,” Power Electronics Specialists Conference, 2004. PESC 04.
2004 IEEE 35th Annual, vol. 3, pp. 2373 – 2379 Vol.3, Jun. 2004.
[66] S. Pinto and J. Silva, “Direct control method for matrix converters with
input power factor regulation,” Power Electronics Specialists Conference,
2004. PESC 04. 2004 IEEE 35th Annual, vol. 3, pp. 2366 – 2372 Vol.3, Jun.
2004.
[67] M. Jussila, M. Salo, and H. Tuusa, “Induction motor drive fed by a vector
modulated indirect matrix converter,” Power Electronics Specialists Con-
ference, 2004. PESC 04. 2004 IEEE 35th Annual, vol. 4, pp. 2862 – 2868
Vol.4, 2004.
[68] L. Wang, D. Zhou, K. Sun, and L. Huang, “A novel method to enhance
the voltage transfer ratio of matrix converter,” TENCON 2004. 2004 IEEE
Region 10 Conference, vol. D, pp. 81 – 84 Vol. 4, Nov. 2004.
[69] K.-B. Lee and F. Blaabjerg, “Reduced-order extended luenberger observer
based sensorless vector control driven by matrix converter with nonlinearity
Bibliography 129
compensation,” Industrial Electronics, IEEE Transactions on, vol. 53, no. 1,
pp. 66 – 75, Feb. 2005.
[70] M. Jussila and H. Tuusa, “Space-vector modulated indirect matrix converter
under distorted supply voltage - effect on load current,” Power Electronics
Specialists Conference, 2005. PESC ’05. IEEE 36th, pp. 2396 –2402, Jun.
2005.
[71] F. Gao and M. Iravani, “Dynamic model of a space vector modulated matrix
converter,” Power Delivery, IEEE Transactions on, vol. 22, no. 3, pp. 1696
–1705, Jul. 2007.
[72] S. Kwak, T. Kim, and O. Vodyakho, “Space vector control methods for
two-leg and three-leg based ac to ac converters for two-phase drive sys-
tems,” Industrial Electronics, 2008. IECON 2008. 34th Annual Conference
of IEEE, pp. 959 –964, Nov. 2008.
[73] H.-H. Lee, H. Nguyen, and T.-W. Chun, “New direct-svm method for matrix
converter with main input power factor compensation,” Industrial Electron-
ics, 2008. IECON 2008. 34th Annual Conference of IEEE, pp. 1281 –1286,
Nov. 2008.
[74] D. Casadei, G. Serra, A. Tani, and L. Zarri, “Optimal use of zero vectors for
minimizing the output current distortion in matrix converters,” Industrial
Electronics, IEEE Transactions on, vol. 56, no. 2, pp. 326 –336, Feb. 2009.
[75] H. She, H. Lin, X. Wang, and S. Xiong, “Space vector modulated matrix
converter under abnormal input voltage conditions,” Power Electronics and
Motion Control Conference, 2009. IPEMC ’09. IEEE 6th International, pp.
1723 –1727, May. 2009.
[76] F. Bradaschia, E. Ibarra, J. Andreu, I. Kortabarria, E. Ormaetxea, and
M. Cavalcanti, “Matrix converter: Improvement of the space vector modula-
tion via a new double-sided generalized scalar pwm,” Industrial Electronics,
2009. IECON ’09. 35th Annual Conference of IEEE, pp. 4511 –4516, Nov.
2009.
Bibliography 130
[77] E. Ibarra, J. Andreu, I. Kortabarria, E. Ormaetxea, and E. Robles, “A fault
tolerant space vector modulation strategy for matrix converters,” Industrial
Electronics, 2009. IECON ’09. 35th Annual Conference of IEEE, pp. 4463
–4468, Nov. 2009.
[78] M. Y. Lee, P. Wheeler, and C. Klumpner, “Space-vector modulated mul-
tilevel matrix converter,” Industrial Electronics, IEEE Transactions on,
vol. 57, no. 10, pp. 3385 –3394, Oct. 2010.
[79] E. Ormaetxea, J. Andreu, I. Kortabarria, U. Bidarte, I. Martinez de Alegria,
E. Ibarra, and E. Olaguenaga, “Matrix converter protection and computa-
tional capabilities based on a system on chip design with an fpga,” Power
Electronics, IEEE Transactions on, vol. PP, no. 99, pp. 1 –1, 2010.
[80] R. Cardenas-Dobson, R. Pena, P. Wheeler, and J. Clare, “Experimental
validation of a space vector modulation algorithm for four-leg matrix con-
verters,” Industrial Electronics, IEEE Transactions on, vol. PP, no. 99, pp.
1 –1, 2010.
[81] I. Takahashi and T. Noguchi, “A new quick response and high efficency con-
trol strategy for an induction motor,” Industrial Electronics, IEEE Trans-
actions on, vol. 22, no. 5, pp. 820 –827, Sep. 1986.
[82] M. Kazmierkowski, R. Krishnan, and F. Blaabjerg, “Control in power elec-
tronics,” Academic Press, 2002.
[83] D. Casadei, G. Serra, and A. Tani, “The use of matrix converters in direct
torque control of induction machines,” Industrial Electronics, IEEE Trans-
actions on, vol. 48, no. 6, pp. 1057 –1064, Dec. 2001.
[84] C. Ortega, A. Arias, C. Caruana, J. Balcells, and G. Asher, “Improved
waveform quality in the direct torque control of matrix-converter-fed pmsm
drives,” Industrial Electronics, IEEE Transactions on, vol. 57, no. 6, pp.
2101 –2110, Jun. 2010.
Bibliography 131
[85] D. Xiao and F. Rahman, “A modified dtc for matrix converter drives using
two switching configurations,” Power Electronics and Applications, 2009.
EPE ’09. 13th European Conference on, pp. 1 –10, Sep. 2009.
[86] K.-B. Lee and F. Blaabjerg, “Sensorless dtc-svm for induction motor driven
by a matrix converter using a parameter estimation strategy,” Industrial
Electronics, IEEE Transactions on, vol. 55, no. 2, pp. 512 –521, Feb. 2008.
[87] J.-X. Wang and J. guo Jiang, “Variable-structure direct torque control for
induction motor driven by a matrix converter with overmodulation strat-
egy,” Power Electronics and Motion Control Conference, 2009. IPEMC ’09.
IEEE 6th International, pp. 580 –584, May. 2009.
[88] D. Xiao and F. Rahman, “An improved dtc for matrix converter drives
using multi-mode isvm and unity input power factor correction,” Power
Electronics and Applications, 2009. EPE ’09. 13th European Conference
on, pp. 1 –10, Sep. 2009.
[89] S. Muller, U. Ammann, and S. Rees, “New time-discrete modulation
scheme for matrix converters,” Industrial Electronics, IEEE Transactions
on, vol. 52, no. 6, pp. 1607 – 1615, Dec. 2005.
[90] M. Rivera, R. Vargas, J. Espinoza, J. Rodriguez, P. Wheeler, and C. Silva,
“Current control in matrix converters connected to polluted ac voltage sup-
plies,” Power Electronics Specialists Conference, 2008. PESC 2008. IEEE,
pp. 412 –417, Jun. 2008.
[91] R. Vargas, J. Rodriguez, U. Ammann, and P. Wheeler, “Predictive cur-
rent control of an induction machine fed by a matrix converter with reac-
tive power control,” Industrial Electronics, IEEE Transactions on, vol. 55,
no. 12, pp. 4362 –4371, Dec. 2008.
[92] R. Vargas, U. Ammann, J. Rodriguez, and J. Pontt, “Predictive strategy to
control common-mode voltage in loads fed by matrix converters,” Industrial
Electronics, IEEE Transactions on, vol. 55, no. 12, pp. 4372 –4380, Dec.
2008.
Bibliography 132
[93] R. Vargas, U. Ammann, and J. Rodriguez, “Predictive approach to increase
efficiency and reduce switching losses on matrix converters,” Power Elec-
tronics, IEEE Transactions on, vol. 24, no. 4, pp. 894 –902, Apr. 2009.
[94] F. Morel, J.-M. Retif, X. Lin-Shi, B. Allard, and P. Bevilacqua, “A predictive
control for a matrix converter-fed permanent magnet synchronous machine,”
Power Electronics Specialists Conference, 2008. PESC 2008. IEEE, pp. 15
–21, Jun. 2008.
[95] P. Gamboa, J. Silva, S. Pinto, and E. Margato, “Predictive optimal matrix
converter control for a dynamic voltage restorer with flywheel energy stor-
age,” Industrial Electronics, 2009. IECON ’09. 35th Annual Conference of
IEEE, pp. 759 –764, Nov. 2009.
[96] F. Barrero, M. Arahal, R. Gregor, S. Toral, and M. Duran, “One-step mod-
ulation predictive current control method for the asymmetrical dual three-
phase induction machine,” Industrial Electronics, IEEE Transactions on,
vol. 56, no. 6, pp. 1974 –1983, Jun. 2009.
[97] Y. Li, N.-S. Choi, H. Cha, and F. Peng, “Carrier-based predictive current
controlled pulse width modulation for matrix converters,” Power Electronics
and Motion Control Conference, 2009. IPEMC ’09. IEEE 6th International,
pp. 1009 –1014, May. 2009.
[98] J. Rodriguez, J. Kolar, J. Espinoza, M. Rivera, and C. Rojas, “Predictive
current control with reactive power minimization in an indirect matrix con-
verter,” Industrial Technology (ICIT), 2010 IEEE International Conference
on, pp. 1839 –1844, Mar. 2010.
[99] R. Vargas, J. Rodriguez, C. Rojas, and P. Wheeler, “Predictive current
control applied to a matrix converter: An assessment with the direct transfer
function approach,” Industrial Technology (ICIT), 2010 IEEE International
Conference on, pp. 1832 –1838, Mar. 2010.
[100] J. Rodriguez, J. Espinoza, M. Rivera, F. Villarroel, and C. Rojas, “Pre-
dictive control of source and load currents in a direct matrix converter,”
Bibliography 133
Industrial Technology (ICIT), 2010 IEEE International Conference on, pp.
1826 –1831, Mar. 2010.
[101] J. Rodriguez, J. Pontt, R. Vargas, P. Lezana, U. Ammann, P. Wheeler, and
F. Garcia, “Predictive direct torque control of an induction motor fed by
a matrix converter,” Power Electronics and Applications, 2007 European
Conference on, pp. 1 –10, Sep. 2007.
[102] R. Vargas, M. Rivera, J. Rodriguez, J. Espinoza, and P. Wheeler, “Predictive
torque control with input pf correction applied to an induction machine
fed by a matrix converter,” Power Electronics Specialists Conference, 2008.
PESC 2008. IEEE, pp. 9 –14, Jun. 2008.
[103] C. Ortega, A. Arias, and J. Espina, “Predictive vector selector for direct
torque control of matrix converter fed induction motors,” Industrial Elec-
tronics, 2009. IECON ’09. 35th Annual Conference of IEEE, pp. 1240 –1245,
Nov. 2009.
[104] T. Geyer, G. Papafotiou, and M. Morari, “Model predictive direct torque
control; part i concept, algorithm, and analysis,” Industrial Electronics,
IEEE Transactions on, vol. 56, no. 6, pp. 1894 –1905, Jun. 2009.
[105] G. Papafotiou, J. Kley, K. Papadopoulos, P. Bohren, and M. Morari, “Model
predictive direct torque control; part ii implementation and experimental
evaluation,” Industrial Electronics, IEEE Transactions on, vol. 56, no. 6,
pp. 1906 –1915, Jun. 2009.
[106] R. Vargas, U. Ammann, B. Hudoffsky, J. Rodriguez, and P. Wheeler, “Pre-
dictive torque control of an induction machine fed by a matrix converter
with reactive input power control,” Power Electronics, IEEE Transactions
on, vol. 25, no. 6, pp. 1426 –1438, Jun. 2010.
[107] J. Rodriguez, J. Kolar, J. Espinoza, M. Rivera, and C. Rojas, “Predictive
torque and flux control of an induction machine fed by an indirect matrix
converter,” Industrial Technology (ICIT), 2010 IEEE International Confer-
ence on, pp. 1857 –1863, Mar. 2010.
Bibliography 134
[108] J. Itoh, K. Koiwa, and K. Kato, “Input current stabilization control of a ma-
trix converter with boost-up functionality,” Power Electronics Conference
(IPEC), 2010 International, pp. 2708 –2714, Jun. 2010.
[109] H. Karaca, R. Akkaya, and H. Dogan, “A novel compensation method based
on fuzzy logic control for matrix converter under distorted input voltage
conditions,” Electrical Machines, 2008. 18th International Conference on,
pp. 1 –5, Sept. 2008.
[110] K. Park, E.-S. Lee, and K.-B. Lee, “A z-source sparse matrix converter with
a fuzzy logic controller based compensation method under abnormal input
voltage conditions,” Industrial Electronics (ISIE), 2010 IEEE International
Symposium on, pp. 614 –619, Jul. 2010.
[111] P. Q. Dzung and L. M. Phuong, “A new artificial neural network - direct
torque control for matrix converter fed three-phase induction motor,” Power
Electronics and Drives Systems, 2005. PEDS 2005. International Confer-
ence on, vol. 1, pp. 78 –83, 2005.
[112] H. H. Lee, P. Q. Dzung, L. M. Phuong, and L. D. Khoa, “A new artifi-
cial neural network controller for direct control method for matrix convert-
ers,” Power Electronics and Drive Systems, 2009. PEDS 2009. International
Conference on, pp. 434 –439, Nov. 2009.
[113] P. Zanchetta, M. Sumner, J. Clare, and P. Wheeler, “Control of matrix
converters for ac power supplies using genetic algorithms,” Industrial Elec-
tronics, 2004 IEEE International Symposium on, vol. 2, pp. 1429 – 1433
vol. 2, May 2004.
[114] P. Zanchetta, J. Clare, P. Wheeler, M. Bland, L. Empringham, and D. Kat-
sis, “Control design of a three-phase matrix converter mobile ac power
supply using genetic algorithms,” Power Electronics Specialists Conference,
2005. PESC ’05. IEEE 36th, pp. 2370 –2375, Jun. 2005.
Bibliography 135
[115] P. Cortes, M. Kazmierkowski, R. Kennel, D. Quevedo, and J. Rodriguez,
“Predictive control in power electronics and drives,” Industrial Electronics,
IEEE Transactions on, vol. 55, no. 12, pp. 4312 –4324, dec. 2008.
[116] E. Camacho and C. Bordons, “Model predictive control,” Springer, 1999.
[117] J. Rodriguez, J. Pontt, C. Silva, P. Correa, P. Lezana, C. P., and U. Am-
mann, “Predictive current control of a voltage source inverter,” Industrial
Electronics, IEEE Transactions on, vol. 54, no. 1, pp. 495 –503, 2007.
[118] P. Cortes, S. Kouro, B. La Rocca, R. Vargas, J. Rodriguez, J. Leon,
S. Vazquez, and L. Franquelo, “Guidelines for weighting factors design in
model predictive control of power converters and drives,” Industrial Tech-
nology, 2009. ICIT 2009. IEEE International Conference on, pp. 1 –7, feb.
2009.
[119] A. Timbus, P. Rodriguez, R. Teodorescu, M. Liserre, and F. Blaabjerg,
“Control strategies for distributed power generation systems operating on
faulty grid,” Industrial Electronics, 2006 IEEE International Symposium
on, vol. 2, pp. 1601 –1607, jul. 2006.
[120] D. Casadei, J. Clare, L. Empringham, G. Serra, A. Tani, A. Trentin,
P. Wheeler, and L. Zarri, “Large-signal model for the stability analysis of
matrix converters,” Industrial Electronics, IEEE Transactions on, vol. 54,
no. 2, pp. 939 –950, 2007.
[121] J. Wiseman and B. Wu, “Active damping control of a high-power pwm
current-source rectifier for line-current thd reduction,” Industrial Electron-
ics, IEEE Transactions on, vol. 52, no. 3, pp. 758 – 764, 2005.
[122] A. Qiu, Y. W. Li, B. Wu, N. Zargari, and Y. Liu, “High performance current
source inverter fed induction motor drive with minimal harmonic distor-
tion,” Power Electronics Specialists Conference, 2007. PESC 2007. IEEE,
pp. 79 –85, 2007.
Bibliography 136
[123] Y. W. Li, B. Wu, N. Zargari, J. Wiseman, and D. Xu, “Damping of pwm
current-source rectifier using a hybrid combination approach,” Power Elec-
tronics, IEEE Transactions on, vol. 22, no. 4, pp. 1383 –1393, 2007.
[124] Y. W. Li, “Control and resonance damping of voltage-source and current-
source converters with lc filters,” Industrial Electronics, IEEE Transactions
on, vol. 56, no. 5, pp. 1511 –1521, May 2009.
[125] S. Mariethoz and M. Morari, “Explicit model-predictive control of a pwm
inverter with an lcl filter,” Industrial Electronics, IEEE Transactions on,
vol. 56, no. 2, pp. 389 –399, 2009.
[126] M. Rivera, P. Correa, J. Rodriguez, I. Lizama, and J. Espinoza, “Predic-
tive control of the indirect matrix converter with active damping,” Power
Electronics and Motion Control Conference, 2009. IPEMC ’09. IEEE 6th
International, pp. 1738 –1744, May 2009.
[127]
[128] R. Pena, R. Cardenas, E. Reyes, J. Clare, and P. Wheeler, “A topology for
multiple generation system with doubly fed induction machines and indirect
matrix converter,” Industrial Electronics, IEEE Transactions on, vol. 56,
no. 10, pp. 4181 –4193, 2009.
[129] T. Friedli and J. Kolar, “Comprehensive comparison of three-phase ac-
ac matrix converter and voltage dc-link back-to-back converter systems,”
Power Electronics Conference (IPEC), 2010 International, pp. 2789 –2798,
2010.
[130] M. Nguyen, H. Lee, and T. Chun, “Input power factor compensation al-
gorithms using a new direct-svm method for matrix converter,” Industrial
Electronics, IEEE Transactions on, vol. PP, no. 99, pp. 1 –1, 2010.
[131] H. Nguyen, H.-H. Lee, and T.-W. Chun, “An investigation on direct space
vector modulation methods for matrix converter,” Industrial Electronics,
Bibliography 137
2009. IECON ’09. 35th Annual Conference of IEEE, pp. 4493 –4498, nov.
2009.