IECON 2005 Matrix Converter Tutorial
November 2005
Matrix Converter Technology Dr Pat Wheeler and Prof Jon Clare Power Electronics, Machines and Control Group School of Electrical and Electronic Engineering University of Nottingham, UK Tel. +44 115 951 5591
Email.
[email protected]
Presentation Outline I Basic Matrix Converter Concepts (Jon Clare) Power Circuit Implementation (Pat Wheeler) • Bi-directional switch implementation and available semiconductor device products • Status of Devices: SiC, Reverse Blocking IGBTs • Current Commutation strategies • Power circuit protection • Practical circuit layout issues
Modulation Algorithms (Jon Clare) • • • •
Mathematical model Basic Modulation problem and solution Voltage ratio limitation Principal modulation methods: Venturini, Space vector, Max-mid-min, Fictitious DC Link
School of Electrical and Electronic Engineering, University of Nottingham, UK
IECON 2005 Matrix Converter Tutorial
November 2005
Presentation Outline II Design Issues (Jon Clare) • Comparison of modulation methods • Input Filter design • Matrix Converter losses and comparisons with other topologies
Two-Stage Matrix Converters (Pat Wheeler) • • • •
Basic Principle of Operation Circuit topologies and device count Comparison of Sparse Matrix Converter Topologies Modulation Schemes
Experimental Matrix Converters and applications (Pat Wheeler) • Application Examples • Industrial Products
Potential Future Application Areas (Jon Clare and Pat Wheeler)
Jon Clare
School of Electrical and Electronic Engineering, University of Nottingham, UK
IECON 2005 Matrix Converter Tutorial
November 2005
Matrix Concept Input filter
3-phase supply
Bidirectional switch
Load
Variable frequency Variable voltage 3-phase output
Basic Ideas Switching pattern and commutation control must avoid line to line short circuits at the input Switching pattern and commutation control must avoid open circuits at the output Each output phase can be connected to any input phase at any time Switch duty cycles are modulated so that the “average” output voltage follows the desired reference (for example a sinusoidal reference) Modulation is arranged so that the “average” input current is sinusoidal when the input voltage, output reference and output current are sinusoidal
School of Electrical and Electronic Engineering, University of Nottingham, UK
IECON 2005 Matrix Converter Tutorial
November 2005
Nomenclature Phase Labelling Convention
A B
SAa
C a
b
c
Load
Example Switching Pattern Possible arrangement SAa (on)
SCa (on)
SBa (on) tBa
tAa
tCa SCb (on)
SBb (on)
SAb (on) tAb
tBb
SAc (on)
tCb SBc (on)
tAc
tBc
SCc (on) tCc
Tseq (sequence time)
Output phase a Output phase b Output phase c
Repeats
Switching frequency = 1/Tseq Modulation strategy ensures that tAa - tCc are generated so that the average output voltage during each sequence equals the target output voltage. The sequence time is constant.
School of Electrical and Electronic Engineering, University of Nottingham, UK
IECON 2005 Matrix Converter Tutorial
November 2005
Illustrative Output Waveforms Fin > Fout
Output line to supply neutral voltage
360 Volts
50Hz in - 25Hz out switching frequency 500Hz
240 120 0
-120 -240 -360 0
Volts
Time (ms)
40
Output line to line voltage
600
Low switching frequency shown for visual clarity
20
400 200 0 -200 -400 -600 0
10
Time (ms)
20
Illustrative Output Waveforms Fin < Fout
Output line to supply neutral voltage
360 Volts
50Hz in - 100Hz out switching frequency 1kHz
240 120 0
-120 -240 -360 0
10
Time (ms)
20
Output line to line voltage 600
Low switching frequency shown for visual clarity
Volts
400 200 0
-200 -400 -600 0
School of Electrical and Electronic Engineering, University of Nottingham, UK
10
Time (ms)
20
IECON 2005 Matrix Converter Tutorial
November 2005
Illustrative Input Waveforms 1.2 0.8
Input current (unfiltered) 50Hz in - 25Hz out
0.4 0 -0.4 -0.8
Low switching frequency shown for visual clarity
-1.2 0
20
40
60
Time(ms) 80
0
5
10
15
Time(ms) 20
1.2 0.8
Input current (unfiltered) 50Hz in - 100Hz out
0.4 0 -0.4 -0.8 -1.2
Example Spectra 100 % 80
50Hz in - 25Hz out
Output voltage 25Hz Sidebands around multiples of the switching frequency
60 40
2kHz switching
20 0 0
Exact nature of spectra depends on modulation method
1
100 50Hz
% 80
2
3
4
kHz
5
Input Current Sidebands around multiples of the switching frequency
60 40 20 0 0
School of Electrical and Electronic Engineering, University of Nottingham, UK
1
2
3
4
kHz
5
IECON 2005 Matrix Converter Tutorial
Modulation Control A number of modulation strategies have been proposed. All of them allow flexible control with the following features: • Continuous control of output voltage amplitude from zero up to a maximum limit • Continuous control of output frequency up to a maximum feasible limit of approximately 1/10 of the switching frequency • Control of input displacement factor: unity, leading and lagging regardless of output power factor
DC-AC and AC-DC conversion is an inherent feature by setting either the input or output frequency to zero
Matrix Converter Features Direct conversion - No DC link - “all silicon solution” No restriction on input and output frequency within limits imposed by switching frequency Inherent bi-directional power flow in all modes with 4 quadrant voltage-current characteristics at both ports “Sinusoidal” input and output currents Potential for high power density if switching frequency is high enough Output voltage limited to 87% of input voltage (for most modulation schemes) Higher semiconductor count than other AC-AC configurations
School of Electrical and Electronic Engineering, University of Nottingham, UK
November 2005
IECON 2005 Matrix Converter Tutorial
November 2005
Alternatives Rectifier DC link Inverter
3-Phase Supply
3-Phase Load
Industry “workhorse” - made from a few kW to MW Unidirectional power flow Poor AC supply current waveforms DC link capacitor is often 30% - 50% of the power circuit volume at 20kW upwards
Alternatives “Back to Back” DC link Inverter
3-Phase Supply
3-Phase Load
Bi-directional power flow PWM control of input bridge with line inductors gives sinusoidal input currents Large DC link capacitor and line inductors Matrix converter provides the same functionality
School of Electrical and Electronic Engineering, University of Nottingham, UK
IECON 2005 Matrix Converter Tutorial
November 2005
Perceived and Actual Limitations Voltage Transfer Ratio • Output voltage is limited to 86% of the input voltage • Only a problem if standard motors are used from a standard supply
Device Count • Normally requires 18 fully controllable switching devices for a 3-phase to 3-phase converter • Compares to 12 switching devices and large reactive components for a back-to back inverter circuit
Control Algorithms • Considered complex by some researchers • Have been reported as processor intensive • No longer really and issue
Device Count
Topology
Fully Controlled Devices
Fast Diodes
Rectifier Diodes
Large Electrolytic Capacitors
Large Inductors
Matrix Converter
18
18
0
0
0
Back-toBack Inverter
12
12
0
1
3
Inverter with Diode Bridge
6
6
6
1
0 or 1
Conventional rectifier DC Link inverter • Has poor supply current waveforms • Provides no regenerative capability • Requires a DC link capacitor
School of Electrical and Electronic Engineering, University of Nottingham, UK
Back to back inverter • • •
Provides regenerative capability Has sinusoidal supply currents Requires a DC link capacitor
IECON 2005 Matrix Converter Tutorial
November 2005
Pat Wheeler
Presentation Outline
Power Circuit Implementation • Bi-directional switch implementation and available semiconductor device products • Current Commutation strategies • Practical circuit layout issues • Power circuit protection
School of Electrical and Electronic Engineering, University of Nottingham, UK
IECON 2005 Matrix Converter Tutorial
November 2005
Matrix Concept
Bidirectional Switch
Motor
The Bi-directional Switch • Must be able to conduct positive and negative currents • Must be able to block positive and negative voltages
Possible Switch Configurations Diode Bridge • High conduction losses » Two diodes and a switching device conducting
• Only one switching device per switch
School of Electrical and Electronic Engineering, University of Nottingham, UK
IECON 2005 Matrix Converter Tutorial
November 2005
Possible Switch Configurations Back to Back Switch • Two switching devices per switch • Conduction losses of only one diode and one switching device • Common Collector » Pair of switching devices arranged with collectors connected » Diodes required for reverse blocking capability
Possible Switch Configurations Back to Back Switch • Common Emitter » Pair of switching devices arranged with emitters connected » Both devices can be gated from the same isolated power supply
• Can Control Direction of Current Flow within each Switch » Useful for most current commutation strategies
• Diodes can be Si or SiC » SiC may offers lower conduction losses, depending on device rating
School of Electrical and Electronic Engineering, University of Nottingham, UK
IECON 2005 Matrix Converter Tutorial
November 2005
Possible Switch Configurations Back to Back Switch • Reverse Blocking IGBTs » Pair of reverse blocking IGBTs » Lower conduction losses » Reverse recovery can be an issue and may lead to higher switching losses
• Simpler Power Semiconductor Module Design » Increase in theoretical reliability?
• Can Control Direction of Current Flow within each Switch
Matrix Converter Device Packaging A Bi-directional Switch in a Single Package • Two IGBTs and associated diodes • A rearranged ‘Inverter leg’ • 200Amp samples available from Dynex Semiconductors
A Matrix Converter Output Leg in a Single Package • Possible to have 3 bi-directional switches in a single package » One package per output leg of the converter » Possible advantages in the minimisation of inductance between devices
• Can be built as specials by Dynex and Semelab • Products from Fuji, IXSY and Mitsubishi using Reverse blocking IGBTs
A Complete Matrix Converter in a Single Package • Suitable for lower power levels • Eupec had a 400V, 7.5kW matrix converter ‘ECONOMAC’ module
School of Electrical and Electronic Engineering, University of Nottingham, UK
IECON 2005 Matrix Converter Tutorial
November 2005
A Bi-directional Switch in a Single Package
Dynex 200Amp Bi-directional Module DIM200MBS12-A
Nine packages for a 3-phase to 3-phase Matrix Converter Used for larger converters, say >200Amps
Common Emitter
A Matrix Converter Output Leg in a Single Package
600V, 300A (SEMELAB)
1700V, 600A (DYNEX)
Three packages for a 3-phase to 3-phase Matrix Converter Used for medium converters, say 50Amps to 600Amps
School of Electrical and Electronic Engineering, University of Nottingham, UK
IECON 2005 Matrix Converter Tutorial
November 2005
A Complete Matrix Converter in a Single Package
EUPEC 35 Amp Matrix Converter Module
One package for a 3-phase to 3-phase Matrix Converter Used for small converters, say >50Amps
A Complete Matrix Converter in a Single Package EUPEC 35 Amp Matrix Converter Module
A three phase to three phase matrix converter 7.5kW from a 400V supply
School of Electrical and Electronic Engineering, University of Nottingham, UK
IECON 2005 Matrix Converter Tutorial
November 2005
The Current Commutation Problem
3-phase input
Motor
Two Rules • Do not short circuit input lines » will short circuit the supply
• Do not open circuit output lines » will open circuit inductive load
The Two Rules for Safe Current Commutation • Do not short circuit input lines 2-phase input
Load
• Do not open circuit output lines 2-phase input
Load
School of Electrical and Electronic Engineering, University of Nottingham, UK
IECON 2005 Matrix Converter Tutorial
November 2005
Switch Cells for a 2-Phase to 1-Phase Converter
SA1
A1 A2 B1 B2
SA2 SB1
RL Load
SB2
2-Switch Converter Commutation Options Switch states for a 2 to 1 matrix Converter • Allowable conditions for each state is given
Commutation path just has to follow the allowable conditions
V1 V2
Io
School of Electrical and Electronic Engineering, University of Nottingham, UK
1 1 1 1 V1=V2
1 1 1 0 V1>V2
1 0 1 1 V1V2
1 0 1 0 Io +ve
0 1 0 1 Io -ve
1 1 0 0 Any
0 0 1 1 Any
0 0 0 0 Io = 0
1 0 0 0 Io +ve 2
0 1 0 0 Io -ve
0 0 1 0 Io +ve
1 0 0 1 V1V2
1 1 0 1 V1 0
• V1 = +2.5 Volts and V2 = -1.2 Volts If IL < 0
and V2 = +2.5 Volts
School of Electrical and Electronic Engineering, University of Nottingham, UK
Only devices which are conducting are turned on
Forms a Two Step Commutation Strategy •
• S2 and D2 are conducting • S1 and D1 are reverse biased
• V1 = -1.2 Volts
Turns off all Devices Which are Not Conducting •
• S1 and D1 are conducting • S2 and D2 are reverse biased
Direct measurement of actual current flowing Current direction information passed between cells
Minimisation of switch state change delays
IECON 2005 Matrix Converter Tutorial
November 2005
Current Direction
Current [mA]
Current Detection Circuit Output During Increasing Current 0.6 0.4 0.2 0 0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
Time [ms]
Current [mA]
Current Detection Circuit Output During Decreasing Current 0.6 0.4 0.2 0 0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6 Time [ms]
Experimental Results 40
30Hz Output
30
Loa d Curre nt (A)
20
2kHz Switching
10 0 -10 -20 -30 -40 0
5
10
15
20
25 Time (ms)
30
35
40
45
50
5
10
15
20
25 Time (ms)
30
35
40
45
50
400 300
Loa d Volta ge (V)
200 100 0 -100 -200 -300 -400 0
School of Electrical and Electronic Engineering, University of Nottingham, UK
IECON 2005 Matrix Converter Tutorial
November 2005
Input Voltage Based Commutation Uses Input Voltages to Make Current Commutation Decisions • Relies on knowledge of relative magnitude order of the input voltages • Requires accurate and balanced measurement of input voltage waveforms required
Example: 4-Step Voltage Commutation • Must avoid critical areas where input voltages are close » Prevention method » Replacement method
4-Step Voltage Based Commutation SA1
SA1
SA1 0V
SA2
SA2
SA2
SB1
SB1
SB1 100V
SB2
SB2
SB2 SA1
SA2
SA1
SA2 SB1
SB2
School of Electrical and Electronic Engineering, University of Nottingham, UK
SB1
SB2
IECON 2005 Matrix Converter Tutorial
November 2005
4-Step Voltage Based Commutation Critical areas VB
VA
VC
Problems may occur when voltages are very close •
Critical areas
•
Could commutated via the other voltage
•
Could rearrange commutation sequence
−
−
Extra losses and unwanted pulses Waveform quality issues unless inherent in control algorithm
4-Step Voltage Based Commutation VA
VA
VB
VB
VC
VC
…A
…A – C – B – B – C – A…
B – C – A \
/
B – C… \
C
/ C
Extra states
Critical Step Prevention Method •
Rearrange commutation sequence
School of Electrical and Electronic Engineering, University of Nottingham, UK
Critical Step Replacement Method •
Commutated via the other voltage
IECON 2005 Matrix Converter Tutorial
November 2005
Comparison of Commutation Methods Output Current Commutation Methods • Relies on measurement of the output current direction on each output leg • Output line open circuit if a commutation error occurs » Overvoltage clamp used
Input Voltage Commutation Methods • Relies on measurement of the relative input voltages • Longer commutation times • Input line short circuit is a commutation error occurs »?
Some Protection Issues Fault conditions • Overcurrent due to short circuit » Commutation failure
• Loss of supply • Output power overload
Protection strategies • No natural freewheeling paths • Have to provide energy storage in event of turning-off all devices » Overvoltage clamp » Freewheeling with the matrix converter circuit
School of Electrical and Electronic Engineering, University of Nottingham, UK
IECON 2005 Matrix Converter Tutorial
November 2005
Matrix Converter Protection
a b c
Input filter
Cin
CClamp
Lin
Clamp circuit line
A B C
3x3 matrix of bi-directional switches
Auxiliary circuits supply unit (gate-drivers, transducers, control)
SMPS
IM 3~
motor
Capacitor is typically very small depends on nature of load
For a 3kW Matrix Converter Drive for an Aircraft Actuator (shown later) machine inductance = 1.15mH maximum output current is, say, 30Amps capacitor required is 2µF
Power Circuit Layout Minimisation of mutual inductance between input lines Inclusion of local capacitance between input lines Laminated input line bus bars • Simplifies power circuit assembly Lstray Clocal Lstray Clocal
Clocal Lstray
School of Electrical and Electronic Engineering, University of Nottingham, UK
IECON 2005 Matrix Converter Tutorial
November 2005
IGBT Turn-off Voltage when using Laminated Input Power Planes
Device Voltage (V)
600 500 400 300 200 100 0 0
200
400
600
800
Time (ns)
Jon Clare
School of Electrical and Electronic Engineering, University of Nottingham, UK
1000
IECON 2005 Matrix Converter Tutorial
November 2005
Presentation Outline
Modulation Algorithms • Mathematical model • Basic Modulation problem and solution • Voltage ratio limitation • Principal modulation methods » Venturini, Space vector, Max-mid-min, Fictitious DC Link
Ideal Switch Matrix vA Input
vB vC
iA iB
SAa
iC va
vb
vc
ia
ib
ic
Output
Assume voltage fed input and current sink output - inductors represent inductive load Measure all voltages with respect to a hypothetical star (wye neutral) point of the supply
School of Electrical and Electronic Engineering, University of Nottingham, UK
IECON 2005 Matrix Converter Tutorial
November 2005
Mathematical Model Assuming instantaneous and perfect commutation v a ( t ) S Aa ( t ) v ( t ) = S ( t ) b Ab v c ( t ) S Ac ( t )
S Ba ( t ) S Bb ( t )
i A ( t ) S Aa ( t ) i (t ) = S (t ) B Ba i C ( t ) S Ca ( t )
S Ab ( t )
S Ca ( t ) v A ( t ) S Cb ( t ) v B ( t ) S Cc ( t ) v C ( t )
S Bc ( t )
S Ac ( t ) i a ( t ) S Bc ( t ) i b ( t ) S Cc ( t ) i c ( t )
S Bb ( t ) S Cb ( t )
where the switching
function
S Kj ( t ) is 1 when the switch
joining input line K to output line j is ON and is 0 otherwise. Voltage and current constraint
∑S
K = A ,B ,C
Ka
(t ) =
∑S
K = A ,B ,C
Kb
(t ) =
∑S
K = A ,B ,C
rules require that :
Kc
( t ) =1
Example Switching Pattern SBa =1
SAa=1 (on)
tBa
tAa SAb=1
tCa
SBb=1
tAb
SCb=1
tBb
SAa=1
tBc
Output phase a Output phase b
tCb SBc=1
tAc
Sca=1
SCc=1 tCc
Tseq (sequence time)
Output phase c Repeats
Switching frequency fsw = 1/Tseq Many different ways of sequencing the switches are possible – depends on modulation strategy Define the modulation duty cycle for each switch as mAa(t) = tAa/Tseq etc
School of Electrical and Electronic Engineering, University of Nottingham, UK
IECON 2005 Matrix Converter Tutorial
November 2005
Low Frequency Modulation Model Switching function model gives instantaneous relationships - not immediately useful for studying modulation Assume that the input frequency and output frequency (fi, fo) a2 : input is inductive (lagging displacement factor) a1 < a2 : input is capacitive (leading displacement factor)
Assuming unity displacement factor solution, allows the switch duty cycle calculation to be reduced to: mKj =
1 2v K v j 1 + 2 3 Vim
School of Electrical and Electronic Engineering, University of Nottingham, UK
for K = A, B,C and j = a, b, c
IECON 2005 Matrix Converter Tutorial
November 2005
Voltage Ratio Limitation Average output voltage taken over each switching sequence equals the target voltage Target voltage must fit within input voltage envelope Input voltage envelope Target output voltages
1.2 0.8 0.4 0 -0.4 -0.8 -1.2
0
90
180
270
360
Basic algorithm has a voltage ratio limitation of 0 < q < 0.5
Optimised Voltage Ratio Modify target output voltages to use all the input volt-second area. Target voltages become:
1.2 0.8 0.4 0 -0.4 -0.8 -1.2
cos(ωot ) − 61 cos(3ωot ) + 1 cos(3ωi t ) 2 3 [vo (t )] = qVim cos(ωot + 2π / 3) − 61 cos(3ωot ) + 2 13 cos(3ωi t ) cos(ω t + 4π / 3) − 1 cos(3ω t ) + 1 cos(3ω t ) o o i 6 2 3
Target output voltages with q=0.866
0
90
180
270
360
Input voltage envelope
Maximum voltage increased to 87% of input Added triple harmonics cancel in the output line to line voltages
School of Electrical and Electronic Engineering, University of Nottingham, UK
IECON 2005 Matrix Converter Tutorial
November 2005
Added Voltage Cancellation Vim cos(ω i t )
Matrix Converter
qVim 6
cos(3ωot )
im − qV cos(3ωi t ) 2 3
qVim cos(ω o t )
Venturini Optimum Amplitude Method Extension to original method to allow use of the modified target waveform set Input displacement factor control is at the expense of voltage ratio Algorithm can be simplified for unity displacement factor to yield:
m Kj =
1 2v K v j 4q + sin( ω i t + β K ) sin( 3 ω i t ) 1 + 3 Vim 2 3 3
for K = A, B,C and j = a, b, c β K = 0,2 π/3,4 π/3 for K = A,B,C respectively and v j includes the triple harmonic addition
School of Electrical and Electronic Engineering, University of Nottingham, UK
IECON 2005 Matrix Converter Tutorial
November 2005
Cyclic Venturini Method (1) Original Venturini method uses a “single-sided” fixed switching sequence S11 t11
S21
S12
S13
t12
t13 S23
t21
S22 t22
t23
S31
S32
S33
t31
t32
t33
tseq
S11 ≡ SAa, S12 ≡ SBa etc
Cyclic Venturini Method (2) Cyclic Venturini method uses a “double-sided” switching sequence S13
S12
S12
S11
t12/2
t12/2
t11/2 S21
S11 t11/2
t13/2 S23
S21
t23/2
S33
t33/2
t21/2 S31 t31/2
tseq/2
S22 t22/2 S32 t32/2
S22 t22/2 S32 t32/2
t21/2
S13 t13/2 S23 t23/2
S31
S33
t31/2
t33/2
tseq/2
“Cyclic” refers to the fact that the selection order of input voltages (3-12-2-1-3 above) is changed every 60O of input period. Input voltage with largest absolute magnitude (1 above) is always placed in the middle. Duty cycle calculations are identical to standard (optimum) Venturini method.
School of Electrical and Electronic Engineering, University of Nottingham, UK
IECON 2005 Matrix Converter Tutorial
November 2005
Cyclic Venturini Method (3) Line to Line Voltage
Non-cyclic (standard)
Cyclic
Cyclic method eliminates sub-optimal vectors
Space Vector Concept Space vector concept allows a 3-phase set of quantities to be represented by a single vector on a complex plane Define space vector of (Va, Vb, Vc) as: 2 Vo (t ) = v a (t ) + v (t )e j 2π / 3 + v c (t )e j 4π / 3 b 3
Geometrically, this amounts to plotting the instantaneous values of the three voltages along axes displaced by 120O
School of Electrical and Electronic Engineering, University of Nottingham, UK
IECON 2005 Matrix Converter Tutorial
November 2005
Space Vector Illustration Assume target voltages are:
v a = qVim cos(ω o t ), v b = qVim cos(ω o t + 2π / 3), v c = qVim cos(ω o t + 4π / 3) Result is that Vo(t) - the target output voltage space vector has constant length qVim and rotates at ωO when plotted in the complex plane imd0046.html Space vector of input current is defined in the same way 2 I (t ) = ia(t ) + i (t )e j2π / 3 + ic (t )e j 4π / 3 i b 3
Target space vector of input current is normally chosen to line up with the input voltage space vector (unity displacement factor), and rotates at ωi
Matrix Converter Space Vectors 27 possible vectors can be split into 3 groups Group I: each output line is connected to a different input line. Space vectors of output voltage rotate at ωi Space vectors of input current rotate at ωO
Group II: two output lines are connected to a common input line, the remaining output line is connected to one of the other input lines. Space vectors of output take one of 6 fixed positions (varying amplitude) Space vectors of input current take one of 6 fixed positions (varying amplitude)
Group III: all output lines are connected to a common input line. All space vectors are at the origin (zero length)
Group I is not useful, only Groups II (18 vectors) and III (3 vectors) are used
School of Electrical and Electronic Engineering, University of Nottingham, UK
IECON 2005 Matrix Converter Tutorial
November 2005
Group II Space Vectors Vector Number +1 -1 +2 -2 +3 -3 +4 -4 +5 -5 +6 -6 +7 -7 +8 -8 +9 -9
Conducting Switches SAa SBa SBa SCa SCa SAa SBa SAa SCa SBa SAa SCa SBa SAa SCa SBa SAa SCa
SBb SAb SCb SBb SAb SCb SAb SBb SBb SCb SCb SAb SBb SAb SCb SBb SAb SCb
SBc SAc SCc SBc SAc SCc SBc SAc SCc SBc SAc SCc SAc SBc SBc SCc SCc SAc
Output Phase Voltages vb vc va vB vB vA vA vA vB vB vC vC vB vB vC vA vA vC vC vC vA vA vB vB vB vA vA vB vC vC vC vB vB vC vA vA vA vC vC vB vA vB vA vB vA vC vB vC vB vC vB vA vC vA vC vA vC
Output Line to Line Voltages vab vbc vca 0 -vAB vAB -vAB 0 vAB vBC 0 -vBC 0 -vBC vBC 0 -vCA vCA 0 -vCA vCA -vAB vAB 0 0 vAB -vAB 0 -vBC vBC 0 vBC -vBC -vCA vCA 0 0 vCA -vCA 0 -vAB vAB 0 vAB -vAB 0 -vBC vBC 0 vBC -vBC 0 -vCA vCA 0 vCA -vCA
Input Line Currents IA Ia Ib+Ic 0 0 Ib+Ic Ia Ib Ia+Ic 0 0 Ia+Ic Ib Ic Ia+Ib 0 0 Ia+Ib Ic
IB Ib+Ic Ia Ia Ib+Ic 0 0 Ia+Ic Ib Ib Ia+Ic 0 0 Ia+Ib Ic Ic Ia+Ib 0 0
IC 0 0 Ib+Ic Ia Ia Ib+Ic 0 0 Ia+Ic Ib Ib Ia+Ic 0 0 Ia+Ib Ic Ic Ia+Ib
Modulation Calculations Calculations are performed at a regular sampling frequency. Target output voltage space vector rotates, but can be assumed to be fixed at a particular magnitude and position during each sampling period. Output voltage space vectors that the converter can produce are fixed in position (or zero). Time weighted switching between adjacent vectors, produces the correct target “average” output voltage vector during each sampling period. Use of 4 (non-zero) vectors in each sampling period allows input current space vector direction to be controlled as well (for unity displacement factor). Any extra time in the sampling period not occupied by active vectors is filled with zero vectors. Sequence of the 4 active vectors is chosen to minimise commutations.
School of Electrical and Electronic Engineering, University of Nottingham, UK
IECON 2005 Matrix Converter Tutorial
November 2005
Target Vector Synthesis ±4, ±5, ±6
±2, ±5, ±8
±7, ±8, ±9
ωo
±1, ±4, ±7
Target vector
±1, ±2, ±3
±3, ±6, ±9
Input current space vectors
Target ±1, ±2, ±3 vector
Output voltage space vectors
±3, ±6, ±9 ±4, ±5, ±6
±7, ±8, ±9
±1, ±4, ±7
ωi ±2, ±5, ±8
For any condition, using 4 vectors allows control of output voltage magnitude and angle and input current angle (displacement factor) In this case vectors are 5, 6, 8, 9
Vector Sequences S13
S11
t13/2 S23 t23/2 S33
t31/2
t33/2 01
V1
S12 S12 t12/2 t12/2 S22 S22 t22/2 t22/2 S32 S32 t32/2 t32/2
t11/2 S21 t21/2 S31 V2
02
V3 V4
03
03
S11
S12
t11/2 S21
t12/2
S22
t23/2
t21/2
t22/2
S32 t32/2
t33/2 01
V1
V2
S23 t23/2
V4 V3
02
S33
t33/2 V2
V1
01
tseq/2
t13/2 S23 S33
S13 t13/2
t21/2 S31 t31/2
tseq/2 S13
S11 t11/2 S21
V3 V4
tseq/2
School of Electrical and Electronic Engineering, University of Nottingham, UK
02
S12 t12/2 S22 t22/2 S32 t32/2 02
S11 t11/2 S21 t21/2
V4 V3
V2
tseq/2
S13 t13/2
S23
t23/2
S33 t33/2 V1
Double sided 3-zero states V1 → V4 are active states 01 → 03 are zero states
Double sided 2-zero states
01
IECON 2005 Matrix Converter Tutorial
November 2005
Space Vector Comments Selection of vector sequence is not unique - different implementations possible Different implementations give different high frequency (distortion) characteristics at the input and output port Common mode addition to output target is inherent with space vector method → 87% voltage ratio Freedom to control input current vector position can be beneficial under distorted/unbalanced load/supply conditions
Min-Mid-Max Method Oyama et al Attempts to minimise switching loss Minimise commutations by having only 2 output phases switched in each sampling period Minimise voltage change at each commutation through optimum selection of switching sequence S11
S11 t11/2
t11/2
S21
S22
S23
S23
S22
S21
t23/2
t22/2
t23/2
t23/2
t22/2
t21/2
S31 S32 t31/2 t32/2 tseq/2
School of Electrical and Electronic Engineering, University of Nottingham, UK
S33
S33
S32
t33/2
t33/2
t32/2 t31/2
tseq/2
S31
IECON 2005 Matrix Converter Tutorial
November 2005
Fictitious DC Link Modulation 1 Modulation considered as a two step process
[v o (t )] = ([A][v i (t )])[B ] First step - multiply by A, second step - multiply result by B [A] and [B] are given by:
cos(ω i t ) [A] = α cos(ω i t + 2π / 3) cos(ω i t + 4π / 3)
T
cos(ω o t ) [B ] = β cos(ωo t + 2π / 3) cos(ω o t + 4π / 3)
Fictitious DC Link Modulation 2 First step yields the “fictitious DC link” and is analogous to rectification 3αVim [ A][v i (t )] = 2 Second step modulates this DC constant at the output frequency and is analogous to conventional inversion using PWM
cos(ω o t )
[A][v i (t )][B ] = 3αβVim cos(ωot + 2π / 3) 2
cos(ω o t + 4π / 3)
Theoretical maximum values of a and b are:
α MAX =
4 3 2 , β MAX = 2π π
yielding a maximum voltage transfer ratio of 1.053!
School of Electrical and Electronic Engineering, University of Nottingham, UK
IECON 2005 Matrix Converter Tutorial
Fictitious DC Link Modulation 3 For q > 0.87 the mean output voltage in each sequence cannot equal the target voltage → Increased low frequency distortion in output and/or input As q → 1.05 input current and output voltage approach quasi-square wave For q < 0.87, method is similar to others Sparse Matrix Converter makes the distinction between [A] and [B] in hardware - but still without DC energy storage
Modulation - Observations Practical implementation of switching schemes (any of them) with a modern DSP is straightforward Switch duty cycles are normally calculated at each sampling instant based on input voltage measurement (all methods) Low frequency distortion/unbalance in input voltage does not appear at output (Instantaneous power out) = (Instantaneous power in) at all instants in a matrix converter
School of Electrical and Electronic Engineering, University of Nottingham, UK
November 2005
IECON 2005 Matrix Converter Tutorial
November 2005
Modulation - Conclusions No restriction on input and output frequency within limits imposed by switching frequency Inherent bi-directional power flow in all modes with 4 quadrant voltage-current characteristics at both ports “Sinusoidal” input and output currents Input displacement factor can be controlled Output voltage limited to 87% of input voltage (for most modulation schemes) Schemes for which q > 0.87 have significant performance penalties
Jon Clare
School of Electrical and Electronic Engineering, University of Nottingham, UK
IECON 2005 Matrix Converter Tutorial
Presentation Outline
Design Issues • Comparison of modulation methods • Input Filter design • Matrix Converter losses and comparisons with other topologies
Comparison - Introduction Define: • Modulation frequency (fm) = frequency at which switching pattern repeats • Sampling frequency (fsamp) = frequency at which modulation duty cycles are calculated • Switching frequency (fsw) = average frequency at which each bidirectional switch commutates
Comparison of modulation methods not straightforward since: • Often fm ≠ fsamp ≠ fsw • Ratio fm/fsw, fsamp/fsw etc depends on modulation method • Even for equal fsw, different modulation methods can give vastly different switching losses
School of Electrical and Electronic Engineering, University of Nottingham, UK
November 2005
IECON 2005 Matrix Converter Tutorial
November 2005
Comparison (1) Comparison of output voltage weighted THD for equal commutation frequency (8kHz)
WTHD =
n max
∑
n =2
f1 I (fn ) fn I (f1 )
Sampling frequencies Vent (8kHz – single sided) SVM 3z (6kHz – double sided) SVM 2z (7kHz – double sided) MMM (9kHz – double sided)
Comparison (2)
Comparison of input current weighted THD for equal commutation frequency (8kHz)
School of Electrical and Electronic Engineering, University of Nottingham, UK
2
IECON 2005 Matrix Converter Tutorial
November 2005
Comparison (3)
Comparison of losses for 30kW converter Balance between conduction and switching loss depends on devices chosen – relatively slow devices used in this example
Input Filter Design R
L C
Matrix Converter
C chosen to limit voltage distortion at converter terminals L chosen to limit current distortion at supply R chosen to give adequate damping • Limit overshoot on turn-on • Avoid excitation of resonance by supply or converter
School of Electrical and Electronic Engineering, University of Nottingham, UK
IECON 2005 Matrix Converter Tutorial
November 2005
Simple Filter Analysis Iin L Assume harmonic current flows entirely in C to calculate distortion on Vin
Vin In
C
Use calculated distortion on Vin to determine distortion on Iin Enables C and L to be determined directly from weighted THD curves and target THD for Iin and Vin
∑ ((f
ITHD1 =
/ f n )I (f n ))
fn ≠ fi
I (f i )
∑ ((f
ITHD 2 =
Power I C = THD1 2 VinTHD 6π fiVll
2
i
i
/ fn )2 I(fn )
)
I 1 L = THD2 3C (2π f )2 I in THD i
2
I (f i )
fn ≠ fi
Simple Example 4.5
0.40
Input current weighted (1/f) THD Venturini optimum method, q =0.8
Weighted THD %
3.5
Input current weighted (1/f 2) THD Venturini optimum method, q =0.8
0.35 Weighted THD %
4.0
0.30
3.0
0.25
2.5
I THD2
0.20
I THD1
2.0
0.15
1.5 1.0
0.10
0.5
0.05
0.0
0.00
0
50
100
150
f sw /f i
200
0
50
100
150
f sw /f i 200
Example: 415V line to line input at 50Hz, 15kW power level at q=0.8, 8kHz switching frequency Target distortions: Input current THD 5%, Converter input voltage THD 5% Data from curves at fsw/fi = 160: ITHD1 = 0.35%, ITHD2 = 0.004% Component values: C = 6µF, L = 210µH Space vector or cyclic Venturini modulation would yield smaller values
School of Electrical and Electronic Engineering, University of Nottingham, UK
IECON 2005 Matrix Converter Tutorial
November 2005
Comparison of AC to AC Converter Losses
Research programme looking at 30kW integrated matrix converter induction motor drive 3 configurations studied Rectifier PWM drive Active front-end PWM drive Matrix converter drive Conduction and commutation losses considered in detail
Voltage Source Inverter Drives Drive application supplying a 30kW induction motor is considered A 400V induction motor load is used with the inverter drives
Ls
Ls 400V 50Hz
IM
400V 50Hz
IM
≡
≡
Rectifier input PWM Inverter Drive
Active front-end Inverter Drive
School of Electrical and Electronic Engineering, University of Nottingham, UK
IECON 2005 Matrix Converter Tutorial
November 2005
Matrix Converter Drive • Maximum voltage transfer ratio of matrix converter is 0.866 • A 340V induction motor load is therefore used for the matrix converter drive v1i
400V 50Hz
v2i v3i
i1i i2i i3i
S11
S21
S31
S12
S22
S32
S13
S23
S33
340V 30kW
≡
OR
IM
Bi-directional Switch 1200V, 200A IGBTs
Matrix Converter Drive
Device Conduction Losses • Fit curve to the IGBT and diode forward voltage drop characteristics. • Matrix Converter - output current flows through a series combination of an IGBT and a diode at all times. • Inverter – Dependant on the output fundamental displacement angle. • Diode bridge – Dependant on supply impedance.
School of Electrical and Electronic Engineering, University of Nottingham, UK
IECON 2005 Matrix Converter Tutorial
November 2005
Device Commutation Losses • Simulations for each converter were used to identify switching instants • IGBT turn-on, turn-off losses and diode recovery energy loss included • Soft turn-on, turn-off instances due to zero current switching • Matrix Converter – switching voltage dependant upon the switching instants • A linear relationship of switching loss with voltage and current at commutation instant was assumed
Results (1) 3000
Total ) ( loss W t (w) pu t u O d et a R t a s e s s o L r et r ev n o C
DB-Inverter AFE-Inverter Venturini M.C. S VM 2z S VM 3z
2500
2000
Note:
1500
THD of SVM method < Venturini at equal sampling frequency
1000
500
0
0
5
10
15
Modulation fre que ncy (kHz)
Variation of total converter loss against sampling frequency at rated load
School of Electrical and Electronic Engineering, University of Nottingham, UK
IECON 2005 Matrix Converter Tutorial
November 2005
75
Load Current (%)
50
25
0 0
2.5
5
7.5
10
15 12.5
Frequency (kHz)
Rectifier Input PWM Inverter
4500 4000 3500 3000 2500 2000 1500 1000 500 0 150 125 100
Total Converter Losses (W)
4500 4000 3500 3000 2500 2000 1500 1000 500 0 150 125 100
Total Converter Losses (W)
Total Converter Losses (W)
Results (2)
75
50
Load Current (%)
25
0 0
2.5
5
7.5
10
15 12.5
Frequency (kHz)
4500 4000 3500 3000 2500 2000 1500 1000 500 0 150 125 100
75
50
25
Load Current (%)
Active front-end Inverter
0
0
2.5
5
7.5
10
15 12.5
Frequency (kHz)
Matrix Converter
Total Converter Loss against load current and sampling frequency
Loss Comparison - Conclusions
• Highest efficiency obtained with diode rectifier PWM inverter • Matrix converter is more efficient than the active front-end drive that has similar characteristics
School of Electrical and Electronic Engineering, University of Nottingham, UK
IECON 2005 Matrix Converter Tutorial
November 2005
Pat Wheeler
Presentation Outline
Two-Stage Matrix Converters (Sparse) • Basic Principle of Operation • Circuit topologies and device count reduction • Comparison of Sparse Matrix Converter Topologies • Modulation Schemes
School of Electrical and Electronic Engineering, University of Nottingham, UK
IECON 2005 Matrix Converter Tutorial
November 2005
Two-Stage Matrix Converters ‘DC’ Link Voltage Bi-directional Switches
Output Line Voltage
3-Phase Supply
3-Phase to 2-phase Matrix Converter
3-Phase Load
Also known as the ‘Sparse’ Matrix Converter Same Functionality as a Matrix Converter Exception: rotating vectors are not possible, ie. different input phase connected to each output phase
In this form it has the same number of devices as a Matrix Converter
Two-Stage Matrix Converters Input Voltage [Volts/10] Unfiltered Input Current [Amps]
‘DC Link’ Voltage [Volts]
Output Voltage (L-N) [Volts]
Output Currents [Amps]
School of Electrical and Electronic Engineering, University of Nottingham, UK
IECON 2005 Matrix Converter Tutorial
November 2005
Sparse Matrix Converters
Single-Stage and Two-Stage Converters a b c
Input filter
Cin
CClamp
Lin
Clamp circuit
A B C
IM 3~
3x3 matrix of bi-directional switches
Auxiliary circuits supply unit (gate-drivers, transducers, control)
SMPS
line
Clamp circuit
Lin Cin
CClamp
IM 3~ Auxiliary circuits supply unit (gate-drivers, transducers, control)
SMPS
motor
line
Both Converters need LC input filter, clamp circuit, Vout/Vin < 0.87! ☺ Save diodes for clamp circuit on load side ☺ Flexible design of rectifier stage ☺ Dead-time commutation in inversion stage ☺ Possible ZCS of rectifier stage during a zero-voltage vector ☺ Conduction losses are load dependent Cannot produce rotating vectors ZCS ⇒ Rectifier stage decrease max. voltage transfer ratio Higher conduction losses at rated power
School of Electrical and Electronic Engineering, University of Nottingham, UK
motor
IECON 2005 Matrix Converter Tutorial
November 2005
Indirect Modulation Model Indirect modulation model for MC = two stage transformation • a rectification stage, to provide a (constant) DC-link voltage • an inversion stage, to produce the three output voltages
Rectification stage p
a b c
Inversion stage
A B C
Upn
[R]=[Sa, Sb, Sc]
n
[T] = [R]⋅[I]
[I]=[SA, SB, SC]T
Known PWM modulation methods may apply easily
Rectifier Stage SV-Modulation Combine adjacent current vectors for sharing the constant output power to the input lines ⇒ sine wave Va ab
Line c b a
REC = ca
P=c
Lin
Cclamp
Cin ac
Iδ dδ⋅Iδ
θ*in
cb
N=a
Iin
dγ⋅Iγ
Rectification Stage ⇒VPN
bc Iγ
Vc
Vb ca
ba
π d γ = mI ⋅ sin − θ*in 3
( )
dδ = mI ⋅ sin θ
* in
School of Electrical and Electronic Engineering, University of Nottingham, UK
Sector γ-sequence:
1
2
3
4
5
ac
0
bc
ba
ca
cb
ab
VP
Va
Vb
Vb
Vc
Vc
Va
VN
Vc
Vc
Va
Va
Vb
Vb
Vline- γ
Vac
Vbc
Vba
Vca
Vcb
Vab
ab
ac
bc
ba
ca
cb
VP
Va
Va
Vb
Vb
Vc
Vc
VN
Vb
Vc
Vc
Va
Va
Vb
Vline- δ
Vab
Vac
Vbc
Vba
Vca
Vcb
δ-sequence:
IECON 2005 Matrix Converter Tutorial
November 2005
Inverter Stage SV-Modulation Line
Combine adjacent voltage vectors for accurate generation of the reference voltage vector
REC = ca
c b a
P=c
INV=011
Lin 001
101
Vβ
Vout
dβ⋅Vβ θ*out
011
Cclamp
Cin
IDC
=“acc”
100
dα⋅Vα
Inversion Stage
Vα
α-sequence
β-sequence
0
100 = IA
110 = -IC
0 IA -IC IA 0
1
110 = -IC
010 = IB
0 -IC IB -IC 0
2
010 = IB
011 = -IA
0 IB -IA IB 0
3
011 = -IA
001 = IC
0 -IA IC -IA 0
4
001 = IC
101 = -IB
0 IC -IB IC 0
5
101 = -IB
100 = IA
0 -IB IA -IB 0
Sector 010
110
π dα = mU ⋅ sin −θ*out 3 * dβ = mU ⋅ sin θ out
(
C=c B=c A=a
N=a
)
IDC [0-α-β -α-0]
Pulse Width Generation Removing the Zero Current Vector from REC Stage = maintain dutyREC proportion Rectification stage duty-cycles
d γR = VPN =
dγ
d δR =
d γ + dδ
dδ dγ + dδ
π dα = mU ⋅ sin −θ *out 3 dβ = mU ⋅ sin (θ *out )
d γR⋅Vline- γ + d δR ⋅Vline- δ dγ
Rectifier Stage
-
-
d1
d0 = dγR ⋅ 1 − ( dγ + dδ ) ⋅ ( dα + d β )
δ -
d2
α
0
Inversion stages duty-cycles
dδ
γ d0
Inverter Stage
mU = 2 ⋅Vout VPN
-
β
d3
- d4
α
d1 = dγ ⋅ dα
0 Overflow
d 2 = (dγ + dδ ) ⋅ d β
Reload
d3 = dδ ⋅ dα
Timer Equivalent switching sequence
0
-
αγ
-
School of Electrical and Electronic Engineering, University of Nottingham, UK
βδ
-βγ -
αγ
-0
d4 = dδR ⋅ 1 − ( dγ + dδ ) ⋅ ( dα + d β )
IECON 2005 Matrix Converter Tutorial
November 2005
Pat Wheeler
Matrix Converter Product The Yaskawa Matrix Converter • The first commercial Matrix Converter product • Launched in 2004 • Aimed at Lift and hoist applications • An important milestone in the development of Matrix Converter • Some circuit optimisation still required, for example in size and wieght
School of Electrical and Electronic Engineering, University of Nottingham, UK
IECON 2005 Matrix Converter Tutorial
November 2005
Matrix Converter Modules
600V, 300A SEMELAB Leg Module
1200V, 35A EUPEC Matrix Converter Module
1200V, 200A Dynex Switch Module
Applications?
Integrated Motor Drives • No DC link capacitor • Voltage ratio not a limitation
Industrial Applications • Lifts and Hoists • Power density • Regeneration
Aerospace • Power density • Temperature tolerance
Electric Military Vehicles • Weight and volume • Bi-directional power flow
School of Electrical and Electronic Engineering, University of Nottingham, UK
1700V, 600A DYNEX Leg Module
IECON 2005 Matrix Converter Tutorial
November 2005
An EHA using a Matrix Converter Permanent Magnet Motor Drive
Aims • Produce a 3kW Matrix Converter to drive an EHA • Demonstrate the actuator as part of the TIMES programme
Testing • Prototype EHA has been tested on 400Hz and variable frequency supplies over a range of realistic loading conditions • Converter has also been tested as a motor drive under various supply conditions found on aircraft
An EHA using a Matrix Converter Permanent Magnet Motor Drive (2) EHA Control Loops Voltage transducers
Matrix Converter
Supply
Supply Voltage
LEMs
PM Motor
Resolver
Actuator
Motor Current
Control (DSP and FPGA)
Motor Speed
Ram Position Demand
School of Electrical and Electronic Engineering, University of Nottingham, UK
Ram Position
LVDT
IECON 2005 Matrix Converter Tutorial
November 2005
An EHA using a Matrix Converter Permanent Magnet Motor Drive (3) Matrix converter driving two 400Hz induction motor fans, V/f mode 10
24
5
20
A 16
0
12
-5
8
-10
4
-15
A
Output current (400Hz)
-20 Input
0
current
-25 (360Hz)
-4 -8 0.001
0.0015
0.002
0.0025
0.003
-30 0.004
0.0035
An EHA using a Matrix Converter Permanent Magnet Motor Drive (4)
Supply Loss Operation
Speed reversal at 9600rpm 15000
Motor shaft speed (rpm)
5000 0 -5000 -10000
Iq ref[Amps]
-15000 0.00
4 2 0 -2 -4 -6 -8 -10 -12 0.00
0.05
0.10
0.15
0.20
0.25
0.30
7500
0 .0
0 .1
0 .2
0 .3
0 .4
0 .5
0.10
0.15
0.20
0.25
0.30
Phase A current
5
Iq
15 10 5 0 .0
0 .1
0 .2
0 .3
0 .4
0.10
0.15
0.20
0.25
0.30
Phase B current
Io 2 [Amps]
10 5 0 -5 -10 0.05
0.10
0.15
0.20
15
0.25
0.30
0 .6
0 .7
0 .8
Input supply voltages
5 0 -5 -10 0.05
0.10
0.15
0.20
Time [secs]
School of Electrical and Electronic Engineering, University of Nottingham, UK
0.25
200 150 100 50 0 -50 -100 -150 -200 0.0
0.1
0.2
0.3
0.4 T ime [secs]
Phase C current
10
Input Supply [Volts]
0.05
15
Io 3 [Amps]
0 .5
-5
-10
-15 0.00
0 .8
0
0 -5
-15 0.00
0 .7
20
q-axis current 0.05
10
-15 0.00
0 .6
25
15
Io 1 [Amps]
Motor speed 8000
7000
Iq [Amps]
Speed [rpm]
10000
Motor Speed [rpm]
8500
0.30
0.5
0.6
0.7
0.8
IECON 2005 Matrix Converter Tutorial
November 2005
Integrated Electromechanical Actuator (EMA) Technology Demonstrator
Electronics
Motor
To design and build an Integrated Electro Mechanical Actuator (EMA) intended as a technology demonstrator for a rudder actuator on a large, twin-engined, civil aircraft. Need to continuously deploy rudder under some flight conditions drives thermal design (stationary motor with high torque) Natural cooling considered
Integrated EMA Technology demonstrator 30kW matrix converter integrated with ballscrewheatsink Switching Signals Gate Drive Circuits Voltage Clamp Capacitors Voltage Clamp Diodes Input Filter Capacitors
Ballscrew housing
School of Electrical and Electronic Engineering, University of Nottingham, UK
IECON 2005 Matrix Converter Tutorial
November 2005
Integrated EMA Technology demonstrator
Bespoke PM motor designed and constructed Speed limited to 4950rpm by use of existing actuator for demonstrator
Integrated EMA Technology demonstrator
School of Electrical and Electronic Engineering, University of Nottingham, UK
IECON 2005 Matrix Converter Tutorial
November 2005
100kW Direct Converter PM Motor Drive
Water-cooled direct power converter 100kW vector controlled PM motor 360Hz-800Hz input, dc-1200Hz output 230V phase voltage input 120kVA rating Aerospace power quality targets Bespoke semiconductor packaging Preliminary results
Dynex/Nottingham collaboration
Entire system designed and developed at Nottingham Control system Control electronics Detailed modelling Power circuit
100kW Direct Converter PM Motor Drive
Input Current [Amps]
200 150 100 50 0 -50 -100 -150 -200 0
0.002
0.004
0.006
0.008
0.01
Time [secs]
Converter on test in USA, May 2005
Input Voltage [Volts]
400 300 200 100 0 -100 -200 -300 -400 0
0.002
0.004
0.006 Time [secs]
School of Electrical and Electronic Engineering, University of Nottingham, UK
0.008
0.01
IECON 2005 Matrix Converter Tutorial
November 2005
An Integrated Matrix Converter Induction Motor Drive (1)
Power Electronics house in the motor end plate
=
+
IGBTs, diodes and filter capacitors Redesigned end plate
Induction Motor
Matrix Converter
Integrated Motor Drive (Power Electronics housed in a redesigned End Plate)
Extra fins to cool the devices
Specially packaged devices (Dynex Semiconductors) 200 Amp Bi-directional Switch module
Integrated Drives above 7.5kW are not feasible within the same motor space envelope DC Link Capacitors form about 40% of the volume
Matrix Converter will give same functionality as a back-to-back inverter drive Regeneration to supply Input current waveform quality BUT no large capacitors or inductors
Bi-directional Switch Modules
Redesign Motor End Plate
Integrated Motor Drive
Bi-directional Switches and Output Connections Power Planes and Input Filter Capacitors
Complete Converter with Gate Drives
School of Electrical and Electronic Engineering, University of Nottingham, UK
IECON 2005 Matrix Converter Tutorial
November 2005
An Integrated Matrix Converter Induction Motor Drive (3)
Output Voltages
Power Circuit fits in available space
2500 2000
Output Voltages [Volts]
1500
Input inductor fits into a slightly modified terminal box
1000 500 0
Cooling requirement known – design for appropriate end plate exists
-500 -1000 -1500 -2000 -2500 0
5
10
15
20
25
30
35
40
45
50
Viability of 30kW integrated drive using matrix converter has been demonstrated
Time [ms ecs ]
Output Current
Input Currents
80
Output Currents [Amps]
60 40 20 0 -20 -40 -60 -80 0
5
10
15
20
25
30
35
40
45
50
Time [msecs]
A 130kW Matrix Converter Vector Controlled Induction Motor Drive Control Platform • Infineon C167 control platform • FPGA based Current Commutation control • Fibre-optic connections from control card to to gate drives
Power Circuit • Water cooled heat sinks • Laminated input power planes Controller Board
Gate Drivers
Work done in collaboration with the US Army Research Labs Design and construction of a large Matrix Converter power circuit
FPGA
Micro Contr.
(6)
PWM
(6)
Bidirectional Switches Current Direction (3)
Current Direction Sensor
School of Electrical and Electronic Engineering, University of Nottingham, UK
Input voltage
(6)
Results from 150kVA tests with an Induction Motor Load under v/f control Closed loop vector control of a 150HP Induction Motor
D/A
Motor
Speed Encoder
Fiber Optic Links (27)
Desired voltage, freq.
PC Controller
Serial Link
IECON 2005 Matrix Converter Tutorial
November 2005
A 130kW Matrix Converter Vector Controlled Induction Motor Drive (2)
Results from a 600Amp, 1200V IGBT Matrix Converter
Output Currents 500 400
150HP Induction Motor Load, 480Volt supply Output Power 129kW (156kVA)
300 200
Amps
100 0 -1 0 0
Switching Frequency: 4kHz
-2 0 0 -3 0 0 -4 0 0 -5 0 0
0
5
1 0
1 5
2 0
2 5
30
35
40
4 5
5 0
Output Voltages 1750 1500 1250
134.0kW
Output Power
129.5kW
Total converter losses
1000 750 500 250
Volts
Input Power
0 -250 -500 -750 -1000 -1250 -1500 -1750 0
5
10
15
20
25
30
35
40
45
50
4530W
Output Power Factor
0.835
Efficiency
96.2%
Input Voltage (L to L)
475V
Input Current
172A
Input Power Factor
0.985
Output Voltage
362V
Output Current
256
Time, m illiseconds
A 130kW Matrix Converter Vector Controlled Induction Motor Drive (3)
Speed Demand
ωref
id
Compensation terms
*
input voltages
*
Id Current Control
vd
Iq Current Control
vq
jθ
e Speed Control
iq
*
3-Phase Supply
MICRO-CONTROLLER Infineon SAB80C167
Flux Current Demand
vα 2/3
vβ
*
va vb vc
vAB vBC
Closed Loop Vector Control of a 150HP Induction Machine
Voltage A to D Input Filter
Matrix Converter Control Algorithm
Matrix Converter Power Circuit
Gate Drives
• Natural regeneration • Low cost Micro-controller control platform
ωr i q*
ωsl
ωe
τ i d*
PWM
dt
id iq
e-jθ
iα iβ
3/2
ia ib
FPGA
Current A to D
ic 1000
ωr
A⊕B Timers
Up/Down
800
FPGA
A
Closed Loop Motor Control
Closed Loop Vector Scheme applied to the Matrix Converter Induction Motor Drive
B
motor
Speed [rpm]
Rotor Speed
600 400 200
Encode
0
800
Id, Iq [Amps]
600 400 200 0
-200 -400
Output Currents [Amps]
600 400 200 0 -200 -400 -600 0
1
2
3 Time [secs]
Control Platform
School of Electrical and Electronic Engineering, University of Nottingham, UK
4
5
IECON 2005 Matrix Converter Tutorial
November 2005
Field Power Supply Using a Four-Output Leg Matrix Converter 250 200
• • • • •
150
Matrix Converter Power Circuit Variable Speed Diesel Engine Permanent Magnet Generator Designed for 10kVA Load 50Hz, 60Hz or 400Hz Output Frequency
Output Line to Line Voltages [V]
Field power supply
100 50 0 -50 -100 -150 -200
DIESEL ENGINE
Matrix 10kW
Gen
Load
-250 0.000
0.001
0.002
0.003
0.004
0.005
0.006
0.007
0.008
Time [s]
400Hz Output Voltage Waveforms FILTER
FILTER
MATRIX CONVERTER
Input Voltage
Space Vector Modulator
Output Current
Modulation D,Q, Control and Engine Demand
Engine Speed Control
Output Voltage
• • • • •
IGBT based Matrix Converter 25kHz Sampling Frequency DSP/FPGA Control Platform LC Output Filter Output Voltage Control Loop designed using a Genetic Algorithm Optimisation • A collaborative project with the US Army Research Labs
Conclusions Matrix converters can offer advantages • Size • Regenerative operation • Sinusoidal input/output Modulation control is not difficult New power devices (eg Silicon Carbide) will increase the attractiveness of matrix converters Current research is application orientated Ongoing research into derived circuits
School of Electrical and Electronic Engineering, University of Nottingham, UK
IECON 2005 Matrix Converter Tutorial
November 2005
Book A Book entitled “Matrix Converters” is due for publication in 2006 • Authors: » Prof Jon Clare » Dr Pat Wheeler » Dr Christian Klumpner » Dr Lee Empringham
• Publisher: » Springer
School of Electrical and Electronic Engineering, University of Nottingham, UK