Beruflich Dokumente
Kultur Dokumente
Elias G. Strangas
I. INTRODUCTION
Although a 3-phase 3-leg inverter is most commonly used
to power AC machines, there are applications were a nleg, n-phase (n > 3) inverter is needed. One such application is where pole-phase modulation is used to control
an induction motor. Pole-phase modulation extends the
speed-torque range of an induction motor by adjusting the
pole-phase ratio. For example, a nine-phase symmetrically
wound induction machine can operate as either a four- or
twelve-pole machine. This is accomplished by reconfiguring
the nine-leg inverter as either a nine-phase or a three-phase
voltage source. Extended speed-torque range is desired for
several automotive induction motor applications. The purpose of this paper is to provide a comprehensive analysis
of a multi-leg, multi-phase inverter and propose techniques
for its implementation.
First, the n-leg, inverter will be examined in terms of
the load equivalent circuit during a 2?/n switching period.
Unlike the three-phase inverter, which has only one load
equivalent circuit, a n-leg inverter has (n - 1)/2 load equivalent circuits. Each one of these equivalent circuits can be
operated in a n-step mode to produce a unique discrete ac
or stair-step voltage waveform. These waveforms differ
in terms of harmonic content and magnitude of the fundamental component. As shown by Holtz [3] the fundamental component of the three-phase 6-step voltage waveform
is 2vdc/7r. On the other hand, a nine-phase inverter has
four different such voltage waveforms, with fundamental
components of O.64Vdc, O.56Vdc, 0.41Vdc, O.22Vdc.
For most electrical machines, the windings are assumed
sinusoidally distributed in the stator. Voltage space-vectors
are defined based on the voltages applied to the windings.
For the conventional three-phase inverter, connected to an
0-7803-709
1-O/O1/$ 10@200
1 I EEE
11. ANALYSIS
OF NINE-LEG
INVERTER
The nine-leg inverter of figure 1 requires eighteen gating
signals, gl-g18. The gating combination determines the
phase voltage across each of the nine separate phases of
the load, which are connected at a point to form a star.
1I-
a) 4-5 Coifigiration
. . .
c) 2-7 Configuration
b) 3-6 Configuration
d) 1-8 Configuration
lower switch open, 'off'. The other two possibilities are the
2-7 configuration and the 1-8 configuration, figure 2-c,d.
1
1
1
0
0
1
1
0
0
0
0
1
0
1
1
This sequence produces three phase voltages that are displaced 120'and that appear in 'stair-step' shape, as shown
in figure 3. For the nine-leg inverter a phase can attain
-2vs
vdc,
(3)
Table I shows the equivalent impedance, source current
and the maximum phase current for the four different load
circuits. The corresponding quantities for a three-phase
inverter were included for comparison. Of the four possible
TABLE I
EQUIVALENT
CIRCUIT
COMPARISON
148
D. Harmonic Content
For a three-phase inverter, switching every 60, the phase
voltage has a fundamental of 2/7rVdc. Table I1 shows the
result of a fundamental magnitude of the waveforms of figures 4,5, 6 and 7.
For a nine-leg inverter operating in eighteen-step mode,
the 4 5 switching set waveform has the highest fundamental
content. The 1-8 switching set has the lowest fundamental
content.
111. EXTENSION
OF SPACE VECTOR THEORY
TO 12
PHASES
Symmetric time-varying electrical quantities operating
on spatially distributed functions (e.g. windings) give rise
to the concept of space-vectors. In an AC machine stator, voltages are assigned a direction (that of the winding
spatial distribution) and their vectorial sum defines the corresponding voltage space-vector.
149
Fundamental
n=l
3" order
n=3
n=5
3 phase
0.6366=2/pi
0.1273
0.0909
Four-Five
Three-Six
Two-Seven
One-eight
0.636bUpi
0.2122
0
0.2122
0.1273
0.1715
0.0909
0.5598
0.4155
0.221 1
0.2122
Configuration
5IhOrder
7* Order
NineQhaseVdlageSpace V m o e
./
/.
I.
;.
/.
/.
!.
0.2303
0.0677
0.2619
0.1951
0.1709
TABLE I1
HARMONIC
CONTENT
to switching states from the [4-51 switching set. Specifically, these space-vectors correspond to the 18 switching
states of the (4'5 muz} subset. This subset consists of
only those switching states where the switches operate 40"
out of phase. The {4,5 muz} switching states are shown
below.
{ 0 0 0 0 0 1 1 1
0 0 0 0 1 1 1 1
1
0
(5)
P-'
'
8
'
0 1 1 1 1 1 0 0 0
1 1 1 1 1 0 0 0 0 }
f .
E] 2
=
[e::]
e 3
-1 -1
1
1
-1
1 -1
1 -1
1 -1 -1 -1
1
-:I
(4)
-are presented.
Three-phase SVP-WM, [3], increases the DC bus utilization over PWM based on Sine-Triangle comparison [l]. By
injecting third-harmonic into the reference voltage signal,
the fundamental of the phase. voltage can be increased.
SVPWM is: implemented by averaging-the. time spent in
adjacent switching' states.
-
[I101
7. .. ..
,,
a)[O0001 1 1 1 111
UW
+--
b)-[OO010 1 1 1 1 11
12. The polygons enclose all the subsets of the four switching sets.
In general, the number of polygons is equal t o the number of equivalent load circuits. In section IV it will be
shown how vectors that form these concentric polygons are
used in a multi-phase Space Vector Pulse Width Modulation algorithm.
IV.
SPACE-VECTOR
PWM
For a n-phase system, nth harmonic signal can be injected by adding V, o f f s e t to the voltage reference.
tending the Unified method, this offset signal is
v, offset = max
[".I
vdc
- min
vdc
[".$I
vdc
(7)
IMPLEMENTATION
Pulse Width Modulation is a general technique for realizing a voltage reference via an inverter. The duration
of the gating pulses controls the frequency and magnitude of the inverter output voltage. Space Vector Pulse
Width Modulation (SVPWM) is a PWM technique which
uses space-vectors to generate these gating duration. Section IV-A reviews three-phase SVPWM. In section IV-B
a proposed general algorithm is presented, which extends
present three-phase SVPWM to n-phase SVPWM. In section IV-C,several SVPWM techniques, specific to n-phase
. d l 1
I 1
1 1
1 1
1 1
l l e
1 1 1
1 1
1 1 1 1 1 1 1
111
1 1
11
v
3 6 max
V*-
K[(ts + t 3 e j g ) ]
VTotal
Fig. 15. The Space Vector PWM Realization of the vector l.0G300
The switching states generated by this algorithm are exclusively from the four maximum subsets described in section 111. Also apparent in figure 15 is that only one switch
changes state at a time. Figure 16 shows the switching
state sequence, for half a sampling period, in terms of corresponding space-vectors.
Fig. 16. Vector sequence for the realization of the voltage space
vector 1 . 0 ~ 3 0 ~
=2fs [Vi,s m a z ( t l e a 2
+
v3,6
max(t7ea2
+ tge@) + ~ 2 , 7m a x ( t 3 e a 2 + t g e j a l )
+tzeja1) +v ~ m
, a~x ( + 5 e ~ 2+ t 4 e 3 a i ) ]
(8)
1
to = - - tl
2fs
- t 8 - t 2 - t7
-t 3 - t6
- t4 - t5
Fig. 17. a)Windings of a 3/9 phase machine in 3-phase operation b)9phase operation c)Windings of a conventional 3-phase machine
Figure 17c shows the winding configuration of a conventional three-phase machine with the same number of
turns. The resultant winding distribution factor for one
phase in figure 17a will be less than the winding distribution factor of figure 17c. Therefore, the resulting voltage
space-vectors will have lower magnitude. For a conventional three-phase machine the six possible non-zero volt-
y,
(16)
..fi
=
(17)
For the equivalent system of figure 17a, the maximum possible modulation is
0.507
4
0.02
0.04
0.06
Fig. 19. Reference Voltage signals and respective voltage offset for
nine and three phase SVPWM
(19)
The extended SVPWM scheme of figure 14 was implemented using a dsp processor, a field programmable gate
array (FPGA), a nine-phase inverter and a nine-phase load.
The dsp supplied the FPGA with nine voltage commands.
The actual SVPWM algorithm was implemented on the
FPGA, which supplied eighteen pulses to gating circuitry.
Figure 18 is a plot of four of the nine phase voltages of an
RC load.
0.08
0.1
seconds
The proposed technique allows for the fi order harmonic to be injected into a n-phase system. The advantage
of this technique is the increase fundamental content of the
output voltage due t o the lower order harmonic injection.
Howerver, f i separate neutrals have t o be used.
In the case of nine-phase system, the nine-phase symmetrical load is treated as three, three-phase systems displaced
40, each producing one third of the total voltage spacevector. Conventional three-phase SVPWM is applied to
each of the three systems to realize the voltage command.
When operating with one neutral, experimental results
showed the phase current had a significant amount of harmonic distortion. Figure 20 shows one of the nine phase
currents.
This harmonic distortion is caused by circulating currents in the load. These currents are a result of unbalanced
3rd order harmonic voltages. Each of the three three-phase
SVPWM algorithms injected a 3rd harmonic component
that was forty degrees with respect t o each other. Figure
21 shows three of the nine voltage reference signals and
their respective offset.
As seen in figure 21, the three voltage offsets are displaced 40, resulting in the three injected 3rd harmonic
voltages being out of phase, resulting in circulating currents. This problem can be solved by separating the ninephase load neutral into three isolated neutrals.
153
Fig. 21. Reference and offset voltages for a three and nine-phase
SVPWM systems.
Fig. 22.
{V4,}5
,,,
2
= (-)(COS(lOo))
= 0.63
71
mg$
(20)
0
1
1
1
1
1
0
154
0
0
0
1
0
0
0
0
0
0
1
0
0
0
0
0
0
1
0
0
0
0
0
0
1
0
0
0
0
0
1
1
1
0
0
0
1
1
1
1
1
0
0
1
1
1
1
1
0
0
1
1
1
1
1
0
(21)
V. CONCLUSIONS
0
0
0
1
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
1
1
1
0
0
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
/
i'
20
REFERENCES
[l] A.B.Plunkett. A current-controlled pwm transistor inverter drive.
Conference Recording IEEE/IAS Annual Meetmg, pages 786-
155