Internati
o
nal
Journal of P
o
wer Elect
roni
cs an
d
Drive
S
y
ste
m
(I
JPE
D
S)
Vol.
6, No. 4, Decem
ber
2015, pp. 906~
918
I
S
SN
: 208
8-8
6
9
4
9
06
Jo
urn
a
l
h
o
me
pa
ge
: h
ttp
://iaesjo
u
r
na
l.com/
o
n
lin
e/ind
e
x.ph
p
/
IJPEDS
Implementation of Space
V
ect
or
Modulat
or
for
Cas
c
ad
ed
H-Bridge Multilevel Inverters
Syamim S
a
n
u
s
i, Auz
a
ni
Jidi
n, Tole
Su
tikn
o,
Kasrul Abd
u
l
Karim
,
Mohd
Luq
m
an Mohd Jamil,
Siti Az
ura Ah
mad T
a
rus
a
n
Facult
y of Ele
c
tr
ica
l
Eng
i
ne
ering
,
Facu
lty of Electrical Eng
i
n
e
ering
,
Un
iv
ersiti Tekn
ik
al Malaysia Melak
a
H
a
ng
Tu
ah
Jay
a
, 761
00
Du
r
i
an
Tun
g
g
a
l, Melak
a
, Malaysia
Article Info
A
B
STRAC
T
Article histo
r
y:
Received
J
u
l 26, 2015
Rev
i
sed
O
c
t 30
, 20
15
Accepted Nov 16, 2015
The Space Ve
ct
or Modulation (
S
VM) technique
has gained wid
e
ac
cept
a
nc
e
for m
a
n
y
AC drive appl
ica
tions,
due to a higher
DC bus voltage utili
zat
ion
(higher outpu
t
voltag
e
when compared with the SPWM), lower harmonic
dis
t
ortions
and
eas
y dig
ita
l re
al
iza
tion.
In r
ecen
t
ye
ars
,
th
e S
V
M
techn
i
qu
e
was extensivel
y adopted in m
u
ltil
evel
inver
t
ers since it off
e
rs great
er
numbers of switching v
ecto
r
s for obt
ain
i
ng fur
t
her improvements of AC
drive perform
an
ces. However, th
e use of m
u
ltilev
e
l inver
t
ers associa
t
ed with
S
V
M
increas
es
the
com
p
lex
i
t
y
o
f
con
t
rol
al
gorithm
(or co
m
putationa
l
burden),
in ob
taining
proper
switching
sequences
and v
ectors. Th
e
complexity
of S
V
M computation causes
a micr
ocontroller or digital sign
al
proces
s
o
r (DS
P
)
to exe
c
ute
the
com
putation
at
a larg
er s
a
m
p
lin
g tim
e. Th
is
consequently
m
a
y
produ
ce er
ro
rs in
computation and hence d
e
grades th
e
control p
e
rfor
m
ances
of AC m
o
tor drives
. This
p
a
per
pres
ents
a
develop
e
ment
of SVM
modu
lator fo
r
three
-
leve
l Cascad
e
d
H-Bridge
Multilev
e
l Inver
t
er (CHMI) using a h
y
brid co
ntrolle
r approa
c
h
, i.e
.
with
combination b
e
tween the DS1104 Controller Bo
ard and FPGA. In
such way
,
the computation
a
l burden can be minimi
zed as
the S
V
M tas
k
s
are dis
t
ribute
d
into two parts, i
n
which ever
y p
a
rt is exe
c
uted
b
y
a singl
e cont
roller
.
This
allows the g
e
n
e
rat
i
on of swit
chi
ng gates performed b
y
FPGA at the
mi
ni
mum sa
mpl
i
ng
t
i
m
e
540
to
o
b
tain
pre
c
ise
d
e
sired
outpu
t
voltag
e
s,
as can
be ver
i
fied via s
imulation and ex
perimental
r
e
sults.
Keyword:
Cascaded h-bri
dge
CHMI
Mu
ltilev
e
l in
v
e
rter
Space vector m
odulation
Th
ree lev
e
ls i
n
v
e
rter
Co
p
y
ri
g
h
t
© 20
15
In
stitu
te
o
f
Ad
van
ced
Engin
eering
a
n
d
Scien
ce.
All rights re
ser
ved.
Co
rresp
ond
i
ng
Autho
r
:
Syam
im
Sanus
i
Faculty of Elec
trical Engineering
Un
i
v
ersiti Tekn
ik
al Malaysia Melak
a
H
a
ng
Tu
ah
Jay
a
, 761
00
Du
r
i
an
Tun
g
g
a
l
Melak
a
, Malaysia
Em
a
il: sya
m
isa
n
u
s
i
@
stud
en
t.u
t
em
.ed
u
.
m
y
1.
INTRODUCTION
The S
p
ace Vec
t
or P
W
M (or
known as S
V
M) is one
of the
m
o
st popula
r
P
W
M technique
s
because of
a hi
ghe
r
o
u
t
p
u
t
vol
t
a
ge a
n
d
fl
exi
b
l
e
t
o
be
im
pl
em
ent
e
d i
n
a
d
v
a
nce
d
ve
ct
or c
o
nt
r
o
l
o
f
AC
m
o
t
o
rs
[
1
-
3
]
.
Tech
ni
cal
l
y
, i
t
was
rep
o
rt
e
d
t
h
at
t
h
e
SVM
i
s
t
h
e m
o
st
fav
o
ra
bl
e m
odul
at
i
on t
e
c
hni
que
am
ong
P
W
M
s
c
hem
e
due t
o
se
veral
adva
nt
ages
, hi
g
h
er o
u
t
put
vol
t
a
ge
, re
duc
i
ng ha
rm
oni
c ri
p
p
l
e
and abl
e
t
o
opt
i
m
i
z
ed t
h
e
switch
i
ng
sequen
ce
[4
]. Besi
des th
at, t
h
is mo
du
latio
n sch
e
me also
offer i
n
o
p
tim
izin
g
th
e
u
s
ed
o
f
d
c
v
o
ltag
e
lin
k
u
tilizatio
n
wh
ich
m
ean
s it can
i
n
crease t
h
e
ratio
n of
o
u
t
p
u
t
m
a
g
n
itud
e
v
o
ltag
e
.
In
ge
neral
,
t
h
e
im
pl
em
ent
a
t
i
on
o
f
S
V
M
i
n
vol
ve wi
t
h
t
h
e
use
d
o
f
DSP
boa
r
d
an
d
req
u
i
r
e
d
sect
o
r
identification whic
h bri
n
gs
i
n
to t
h
e
form
ations
of
rotating s
p
ace
vect
or dia
g
ram
.
[5]. In the
three
pha
se
syste
m
, there are six fractions
in the space vector di
a
g
ram
spinning 360°
whic
h each ha
s equally divided by
6
0
°. Th
is sp
ace v
ecto
r
d
i
agra
m
is a
tran
sform
a
t
i
o
n
fr
o
m
a
b
a
lan
ce of three p
h
a
se
q
u
an
t
ities in
to
two
p
h
a
se
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
PED
S
I
S
SN
:
208
8-8
6
9
4
Impleme
n
tation
of Space Vec
t
or M
o
du
lat
o
r for
Cascade
d
H-Bri
d
ge
M
u
ltilevel Inverters (Syamim Sanusi
)
90
7
syste
m
of
α
-
β
r
e
fere
nce f
r
am
e [4]
.
T
h
e S
V
M
m
a
i
n
ope
rat
i
on i
s
t
o
use
d
t
h
e nea
r
est
t
h
re
e vect
o
r
rec
o
g
n
i
t
i
on
of
th
e
referen
ce v
o
ltag
e
and
determin
ed
th
e co
rresp
ond
ing
on-ti
m
e
u
s
i
ng t
h
e princi
ples of volt s
econd
equi
val
e
nt
[6]
A
m
u
l
t
i
l
e
vel
inve
rt
er of
fers great
er n
u
m
b
er
of v
o
ltage vectors as c
o
mpare
d
to eight
vectors for a
two-level inve
rter. Fi
gure 1
illustrates the space vector
diagram
s
for Se
ctor I, in a two-le
vel inve
rte
r
and
th
ree-lev
e
l
inverter. Th
e sp
ac
e vector
diagra
m
s
can be us
e
d
to co
m
p
are t
h
e i
m
pl
em
ent
a
t
i
on
of
SVM
i
n
t
w
o-
level and t
h
re
e-level inve
rters.
As
com
p
ared to two-le
ve
l space
vect
or di
agram
,
the
sector i
n
t
h
ree
-
level
in
v
e
rter is d
i
v
i
d
e
d
i
n
to
fou
r
i
d
en
tical sm
al
le
r triang
les (i.e.
∆
,
∆
,
∆
and
∆
). T
o
red
u
ce T
HD
(
o
r
⁄
)
and
switchi
ng
losses in m
u
ltil
evel inverter
(i
.e. three-le
vel), it is necessary
to switchi
ng
vectors
whic
h a
r
e the
nearest to t
h
e
refe
rence
vect
or
̅
∗
. Hen
ce, t
h
ree-lev
e
l SVM switch
th
e
v
e
cto
r
s
̅
,
̅
a
n
d
̅
for a
gi
ve
n
refe
rence
vect
o
r
.
The cal
cul
a
t
i
o
n o
f
o
n
-
d
urat
i
on i
n
m
u
l
t
i
l
e
vel
SVM
i
s
qui
t
e
com
p
l
i
c
at
ed and
di
ffe
re
nt
fo
r va
ri
o
u
s
trian
g
l
es
d
u
e
to
sm
all
trian
g
l
es in
th
e sp
ace v
ector d
i
a
g
r
a
m
of t
h
re
e-l
e
vel
i
nve
rt
er i
n
Fi
g.
1 d
o
not
exact
l
y
im
itate the geometry of a sector of
two
-
lev
e
l in
v
e
rter. In
t
w
o-lev
e
l SVM
,
th
e calcu
lation
of on
-du
r
ati
o
n
is
st
rai
ght
fo
rwa
r
d w
h
i
c
h i
s
val
i
d
fo
r eve
r
y
sect
or. H
o
weve
r, t
h
e three
-
level SVM
nee
d
s to m
odify
the referenc
e
v
ector with
n
e
w o
r
i
g
in
po
in
t
to
ap
p
l
y
th
e
t
w
o-lev
e
l b
a
sed
SV
M f
o
r
cal
cu
latin
g
on
-dur
atio
n. A
s
show
n
b
y
Figu
re
1
(b
), t
h
e m
odified
re
fere
nce
vecto
r
̅
∗
with v
ect
o
r
̅
as orig
i
n
p
o
i
n
t
is determ
in
ed
su
ch
th
at t
h
e
calcu
latio
n
of o
n
-d
uratio
n
is si
m
ilar
to
th
at
o
f
two
-
lev
e
l b
a
sed
SVM. Th
e calcu
latio
n
b
e
co
m
e
s co
m
p
lica
t
ed
if
the
re
fere
nce voltage vector
̅
∗
l
i
es in
triang
le
∆
,
whe
r
e t
h
e
o
r
i
e
nt
at
i
on
of
t
r
i
a
n
g
l
e
i
s
di
f
f
ere
n
t
am
on
g ot
hers;
as can
b
e
seen
th
e triang
les
∆
,
∆
and
∆
have t
h
e s
a
m
e
ori
e
nt
at
i
o
n
wi
t
h
a si
ngl
e
t
r
i
a
ngl
e
o
r
Se
ct
or
I i
n
t
w
o
-
l
e
vel
S
V
M
,
as
sho
w
n i
n
Fi
g.
1
.
T
h
e c
o
m
p
lexity increases
as
num
b
er
of le
ve
l
of
i
n
vert
e
r
b
ecom
e
s hi
ghe
r,
e.g
.
i
n
fi
ve-l
ev
el
i
n
vert
er
, t
h
e
r
e a
r
e si
x t
r
i
a
ngl
es
am
ong
si
xt
ee
n
t
r
i
a
ngl
es,
t
h
at
have
di
ffe
rent
ori
e
nt
at
i
on.
Th
e im
p
l
e
m
en
tatio
n
o
f
SVM fo
r m
u
ltile
v
e
l in
v
e
rters req
u
i
re so
m
e
i
m
p
o
r
tan
t
p
a
rts wh
ich
are as
fo
llows; (1) d
e
tectio
n
o
f
sector
, (2
) d
e
tection
o
f
triang
le
∆
, (3
) calcu
latio
n
o
f
on
-du
r
ation
for switch
i
ng
t
h
e
nearest
vect
ors
,
an
d
(
4
)
det
e
r
m
i
n
at
i
on t
h
e
s
w
i
t
c
hi
n
g
se
q
u
e
n
ce
fo
r e
v
ery
s
w
i
t
c
hi
n
g
peri
o
d
.
As
found in literature
, t
h
ere
a
r
e two c
o
m
m
on m
e
thods
to c
a
lculate the
on-durations
. T
h
e
first m
e
thod
is to
d
e
tect th
e trian
g
l
e and
so
lv
e three simu
ltan
e
ou
s e
q
u
a
tio
n
s
of th
e tri
a
n
g
l
e to
d
e
termin
e th
e o
n
-times as
sug
g
est
e
d i
n
[
7
]
.
The sec
o
nd
m
e
t
hod i
s
t
o
det
ect
t
h
e t
r
i
a
n
g
l
e
an
d use
pa
rt
i
c
ul
ar o
n
-
d
u
r
a
t
i
on eq
uat
i
o
ns
st
ore
d
i
n
a l
o
o
k
up t
a
bl
e f
o
r t
h
i
s
t
r
i
a
ngl
e, a
s
p
r
o
p
o
se
d i
n
[
8
]
.
B
o
t
h
of t
h
ese
m
e
t
hods
h
o
we
ver
req
u
i
r
e c
o
m
p
l
e
x
com
putations a
s
the
num
b
er of level i
n
crease
s
.
Al
t
e
rnat
i
v
el
y
,
t
h
e cal
cul
a
t
i
o
n of
on
-
d
u
r
at
i
o
n
can be
obt
ai
n
e
d usi
ng
ge
ner
a
l
al
gori
t
h
m
s
[9]
and
[
10]
.
Specifically, [9] uses a Euclidean vect
o
r
syst
e
m
with
sev
e
ral
m
a
trix
tran
sfo
r
m
a
tio
n
,
provid
e
d
th
at it do
es n
o
t
provide a syste
m
atic approa
ch for
real ti
me SVM im
p
l
e
m
en
tatio
n
.
On th
e o
t
h
e
r way
,
[10
]
calcu
lated
on
-
du
rat
i
o
n an
d o
b
t
a
i
n
ed s
w
i
t
c
h
i
ng st
at
es by
m
eans of c
o
o
r
dinate system
,
whe
r
e the a
x
es
are 60 de
gree
s apart.
Ho
we
ver
,
t
h
e
60
de
g
r
ees t
r
a
n
sf
orm
a
t
i
on l
e
ads t
o
t
h
e
co
m
p
l
e
xi
ty
si
nce t
h
e
v
o
l
t
a
ge
re
fere
nce i
s
c
o
m
m
onl
y
defi
ned
i
n
t
h
e
ort
h
o
g
onal
c
o
o
r
di
nat
e
sy
st
em
.
Recently, a sim
p
le SVM algorithm
for m
u
ltilevel i
nverters base
d on sta
nda
rd two-leve
l SVM was
p
r
op
o
s
ed
in
[11
]
. Th
e two
-
lev
e
l b
a
sed
SVM co
n
cep
t is in
itiated
b
y
[1
2-14
], ho
wev
e
r, th
e calcu
latio
n
o
f
o
n
-
du
rat
i
o
n i
s
based o
n
ori
g
i
n
m
odi
fi
cat
i
on an
d
60 de
gree
s
co
or
di
nat
e
t
r
an
sf
orm
a
t
i
on, w
h
i
c
h can
not
be e
x
t
e
nd
e
d
in
i
m
p
l
e
m
en
ti
n
g
SVM
for h
i
gh
er lev
e
ls, i.e.
greater than three.
Unlik
e t
h
e form
er
m
e
thods, the
Figure
1. Comparis
on Betwe
e
n
(a
) Two-Le
vel
Space Vect
or
Dia
g
ram
and
(b) T
h
ree
-
Le
vel Space
Vect
or
Diagram
,
e.g.
for Sector
I.
(
a
)
(b
)
̅
∗
̅
∗
̅
̅
̅
̅
̅
̅
̅
̅
̅
∆
∆
∆
∆
̅
∗
Sector I
Sector I
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
088
-86
94
I
J
PED
S
Vo
l.
6
,
No
.
4
,
D
ecem
b
er
2
015
:
90
6 – 918
90
8
im
pl
em
ent
a
t
i
o
n
of
SVM
pr
o
p
o
se
d i
n
[
11]
t
h
at
i
n
cl
ude
s t
h
e
det
ect
i
on
o
f
se
ct
or a
n
d
t
r
i
a
n
g
l
e
, an
d cal
c
u
l
a
t
i
o
n
o
f
on
-
d
u
r
at
i
o
n
we
re
deri
ve
d
ge
o
m
et
ri
cal
ly
and
sy
st
em
ati
cal
l
y
whi
c
h s
u
i
t
a
bl
e
fo
r a
n
y
l
e
vel
of
i
nve
rt
er.
.
2.
TOPOLOGY OF
THREE-L
E
VEL C
A
S
C
ADE
D H
-
BR
I
D
GE
MULTILEVEL
INVE
RTER
Th
e sm
allest n
u
m
b
e
r of lev
e
l
for Cascad
ed H-b
r
i
d
g
e
Mu
l
tilev
e
l Inv
e
rter (CHMI) is three lev
e
ls.
Fig
u
r
e
2 show
s a topo
log
y
circu
it of
3
-
lev
e
l
CH
MI
, wh
ich
co
nsists of
t
h
r
e
e fu
ll br
idg
e
i
n
v
e
r
t
ers
(
o
r
kn
ow
n as
H-
bri
dge
).
Eac
h
bri
dge
c
onsi
s
t
s
o
f
t
w
o
l
e
g
s
,
4
po
we
r s
w
i
t
c
hes, i
.
e.
I
G
B
T
s an
d
an
i
d
ent
i
cal
i
s
ol
at
ed
DC
v
o
ltag
e
.
Th
e iso
l
ated
DC
vo
ltag
e
fo
r three-l
e
v
e
l C
H
MI is
d
e
sign
ated as
. Th
e
fi
rst
l
e
g
of
eac
h
bri
dge
i
s
co
nn
ected
t
o
its p
h
a
se or
w
i
nd
ing
of
a three-ph
ase indu
ctio
n
m
o
to
r, wh
ile
anot
her l
e
g i
s
sho
r
t
e
n a
s
a c
o
m
m
on
p
o
i
n
t
wh
ich referred to
as a
n
e
u
t
ral po
in
t,
. In
an
y H-bridg
e
inv
e
rter, it can
produce
three states
of output
v
o
ltag
e
, i.e.
,0
or -
, as gi
ven
i
n
(
1
)
.
.
(1
)
Where
and
a
r
e t
h
e
res
p
ective
switchi
ng st
ates for the
upper s
w
itches
of t
h
e
first
and
second legs
of
th
e
-ph
a
se H-bridg
e
inv
e
rter. No
te
t
h
at
th
e switch
i
ng
o
f
upper
and l
o
we
r
switches i
n
the
sam
e
leg
m
u
st be
co
m
p
li
m
e
n
t
ary
to
each
o
t
h
e
r t
o
av
o
i
d
sho
r
t circu
it cond
itio
ns.
3.
SPAC
E V
E
CTOR
M
O
DULA
T
ION
3.
1.
M
a
p
p
i
n
g
Vect
or
Co
m
p
are with
th
e two-lev
e
l i
n
v
e
rter
b
a
sic co
n
c
ep
t,
th
e t
h
ree-ph
ase
q
u
a
n
t
ities o
f
stato
r
vo
ltag
e
v
ect
o
r
can
b
e
ex
pressed
in
to
a sp
ace v
o
ltag
e
v
ect
o
r
form
b
y
su
b
s
titu
tin
g
t
h
e b
a
sic alg
o
tithm o
f
two
levels p
h
a
se
stator voltages with
e
q
uation
(1), the
s
p
ace
voltage
vector c
a
n als
o
be
writ
ten in term
s of
switching
states and
DC link
v
o
ltage. Th
en it can
b
e
ex
press in
to
- a
n
d
-axi
s c
o
m
ponent
s
of
st
at
or
v
o
l
t
a
ge, a
s
f
o
l
l
o
ws:
3
2
(2
)
1
√
3
(3
)
Fig
.
3
shows vo
ltag
e
v
ect
o
r
s
av
ailab
l
e in
the th
ree-lev
e
l CHMI.
All vo
lta
ge vect
ors a
r
e
m
a
pped
o
n
th
e
- an
d
-ax
i
s p
l
an
e b
y
app
l
yin
g
ev
ery swi
t
ch
in
g state possib
ility in
(2) and
(3).
From
th
is figu
re,
th
e
m
a
ppi
n
g
of
v
o
l
t
a
ge vect
o
r
s
f
o
rm
s si
x sect
o
r
s,
whe
r
e eac
h
sect
or c
o
nt
ai
n
s
fo
u
r
sm
al
l
of t
r
i
a
ng
ul
ars
an
d t
h
e
vol
t
a
ge
vect
or
s
can
be
cat
eg
or
i
zed i
n
t
o
f
o
ur
gr
o
ups
as
fol
l
o
ws:
1) l
o
n
g
am
pl
i
t
ude
o
f
vol
t
a
ge
vect
o
r
s,
i
.
e
.
̅
,
̅
,…,
̅
, 2) m
e
di
u
m
am
pl
it
ude o
f
vol
t
a
ge
vect
o
r
s, i
.
e.
̅
,
̅
,…,
̅
, 3)
sho
r
t
am
pl
i
t
ude of
vol
t
a
ge
ve
ct
ors
i.e.
̅
,
̅
,…,
̅
, a
n
d 4) zero
volta
ge vector, i.e.
̅
. It
should be noted
t
h
at
each long or
m
e
dium
voltag
e
Fig
u
re
2
.
Topolo
g
y
Circu
it of
Th
ree-Lev
e
l C
a
scad
ed
H
-
Bri
d
g
e
Mu
ltilev
e
l
Inv
e
rter
N
Induc
t
ion
Machin
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
PED
S
I
S
SN
:
208
8-8
6
9
4
Impleme
n
tation
of Space Vec
t
or M
o
du
lat
o
r for
Cascade
d
H-Bri
d
ge
M
u
ltilevel Inverters (Syamim Sanusi
)
90
9
vect
o
r
ha
s o
n
l
y
a si
ngl
e o
f
swi
t
c
hi
ng stat
es, howeve
r
, t
h
e num
b
er of
switching sta
t
es increases
as the
am
pl
i
t
ude of v
ect
or red
u
ces
t
o
wa
r
d
t
h
e
o
r
i
g
i
n
of
t
h
e pl
ane
.
3.2.
De
terminati
on of
Sec
t
or and Triangl
e
From
Figure
3, it can
be
noticed that
the m
a
ppi
ng
of
activ
e v
ector
s fo
r
m
s a h
e
x
a
gon
al
bo
und
ar
y, in
whi
c
h t
h
e
hex
a
go
nal
ca
n
be
eq
ual
l
y
di
vi
d
e
d i
n
t
o
si
x
se
ct
ors.
T
h
i
s
m
e
ans t
h
e
di
f
f
ere
n
ce
bet
w
ee
n t
h
e t
w
o
angles
of a
d
jac
e
nt active
voltage
vectors is
60
0
. It
should
be
note
d
t
h
at t
h
e s
p
ace
voltage
vector ca
n
also
be
expresse
d i
n
to
a polar
form
, as bel
o
w:
̅
∗
∗
∠
(4
)
Tak
i
ng
i
n
to acco
un
t t
h
at th
e
u
n
it
o
f
is ex
pressed in rad
i
an wh
ich
v
a
ries
b
e
tween
0
an
d
ra
d
.
Th
e invo
lv
em
en
t of
n
e
g
a
ti
v
e
v
a
lu
e in
variation c
o
m
p
licates the cal
culation. There
f
ore,
a sm
a
ll
m
odi
fi
cat
i
on i
s
m
a
de usi
n
g
a
m
odul
us t
e
c
h
n
i
que,
s
u
ch
t
h
at
t
h
e
eq
ui
val
e
nt
va
ri
at
i
o
n
o
b
t
a
i
n
ed
bet
w
ee
n
0
°
an
d
360
°
. Th
e
fo
llowing
eq
u
a
ti
o
n
s
n
e
ed
th
e tran
sformatio
n
of
i
n
deg
r
ees
an
d t
h
e m
odul
us
o
p
e
r
at
i
o
n
t
o
ha
v
e
th
e po
sitiv
e v
a
lu
e v
a
riatio
n
s
, i.e.
. Th
e
six
secto
r
s can
be easily d
e
ter
m
in
ed
b
y
sub
s
titu
tin
g
t
h
e po
sit
i
v
e
vari
at
i
o
ns
in
to (5
), th
is yield
s
:
60
1
(5
)
W
i
t
h
th
e in
creasem
e
n
t
n
u
m
b
e
r
o
f
lev
e
l, th
e term
in
atio
n
of secto
r
fo
r m
u
ltilev
e
l n
eed
t
o
d
eals
with
th
e ex
isten
c
e
of sm
all seg
m
en
t of t
r
iang
les
(i
.e.
∆
,
wh
er
e
=0,
1,
2
or
3)
i
n
si
d
e
eac
h sect
or
f
o
r
cal
cul
a
t
i
n
g
on
-
d
u
ration
of
vo
ltag
e
v
ect
o
r
s wi
h
i
ch
its sim
ilar
to
th
e t
w
o-level b
a
sed SVM
[11
]
. In
su
ch
way, th
e
d
e
fin
itio
n of
refe
rence
vol
t
a
ge vect
o
r
i
s
ba
sed at
whi
c
h t
r
i
a
ngl
e t
h
e vect
or is located, whe
r
e
th
e calcu
latio
n
of
on-
du
r
a
tion
o
f
vo
ltag
e
v
ect
o
r
s in
a triang
l
e
is treated th
e
sam
e
way
as the vectors
in a
s
ector
for t
w
o-l
e
vel ba
sed SVM. To
obt
ai
n
∆
,
at first, the
angle withi
n
a sector
nee
d
s
t
o
be obt
ai
n
e
d usi
n
g (6
).
Fi
gu
re
3.
V
o
l
t
a
ge
Vect
o
r
s i
n
t
h
e T
h
ree
-
Le
vel
C
H
M
I
O
b
tained
in Eq
u
a
n
tion
s
(2) and
(3) fo
r Ev
ery Switch
i
ng
State Po
ssi
b
ility (S
a1
S
a2
S
b1
S
b2
S
c1
S
c2
)
(
1
001
10
)
̅
(
1
010
01
)
(
0
110
01
)
(
0
110
00
)
(
0
110
10
)
(
0
100
10
)
(
0
101
10
)
(
1
001
00
)
(
1
001
01
)
(
1
000
01
)
(
1
010
00
)
(
0
000
01
)
(
1
000
00
)
(
0
001
01
)
(
0
001
00
)
(
1
000
10
)
(
0
000
10
)
(
0
101
00
)
(
0
100
00
)
(
0
010
10
)
(0
010
00
)
(
0
100
01
)
(
1
010
10
)
(
0
101
01
)
̅
̅
̅
̅
̅
̅
̅
(
00
10
0
1
)
̅
̅
̅
̅
̅
̅
̅
̅
̅
(
0
000
00
)
(
0
001
10
)
̅
√
Sec
I
Sec
I
I
Sec
III
Sec
V
I
Sec
V
Sec
I
V
̅
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
088
-86
94
I
J
PED
S
Vo
l.
6
,
No
.
4
,
D
ecem
b
er
2
015
:
90
6 – 918
91
0
,
6
0
(6
)
Th
en
, let th
e i
d
en
tification
of triang
le is mad
e
by
c
o
nsi
d
eri
n
g t
h
e sect
or at
whi
c
h t
h
e refe
rence
voltage
vect
or
is located. Bas
e
d on t
h
e
- an
d
-axi
s
pl
ane, t
h
e c
o
m
pone
nt
s of
refe
re
nce v
o
l
t
a
ge
vect
or
c
a
n
b
e
written
as:
∗
∗
co
s
(7
)
∗
∗
sin
(8
)
Altern
ativ
ely, th
e
equ
i
v
a
len
t
v
o
ltag
e
v
ector
̅
∗
ba
sed
on
- a
n
d
-a
xi
s
pl
ane
i
n
a
t
r
i
a
ngl
e
i
s
defi
ned
(as
illustrated in z
oom
ed im
ages in Figur
e 4(b) and
(c
)), hence
the
c
o
m
pone
nts
of the
vect
or can
be calc
u
la
ted as:
̅
∗
∗
∗
(9
)
From
t
h
e vect
or di
a
g
ram
s
sho
w
n i
n
Fi
g
u
r
e
4, i
t
can be not
i
ced t
h
at
t
h
e de
fi
ni
t
i
on
of re
fere
nce
v
o
ltag
e
v
ector with
in
a
sm
all trian
g
l
e fo
r tri
a
n
g
l
e
∆
,
∆
an
d
∆
is similar to
th
at
d
e
fin
e
d
i
n
Secto
r
I
fo
r th
e
case o
f
two
-
lev
e
l b
a
sed
SVM. Geo
m
etrically,
th
e
t
r
iang
le
∆
(f
or
=0,
1
o
r
3)
a
n
d
t
h
e
vect
or
c
o
m
pone
nt
s
∗
and
∗
can
be cal
cul
a
t
e
d
usi
n
g t
h
e
fol
l
o
wi
n
g
e
quat
i
o
ns
[
11]
:
∗
∗
2
2
3
(1
0)
∗
∗
√
3
(1
1)
∆
2
(1
2)
whe
r
e
and
ar
e ob
tain
ed using (1
3)
an
d (14
)
:
3
2
∗
∗
√
3
(1
3)
√
3
∗
(1
4)
On
t
h
e
ot
he
r
h
a
nd
, t
h
e
de
fi
ni
t
i
on
of
re
fere
nc
e v
o
l
t
a
ge
vect
o
r
wi
t
h
i
n
a sm
all
t
r
i
a
ngl
e
fo
r t
r
i
a
ngl
e
∆
is
sim
i
l
a
r t
o
t
h
at
defi
ned i
n
Sect
or
IV
fo
r t
h
e c
a
se of t
w
o
-
l
e
v
e
l
based S
V
M
.
It
can be
pr
o
v
e
d ge
om
et
ri
call
y
t
h
at
th
e triang
le
∆
and
t
h
e
vect
o
r
c
o
m
pone
nt
s
∗
and
∗
can be co
m
p
u
t
ed usin
g (1
5)
, (1
6)
an
d (17),
respectively [11].
∗
3
12
∗
(1
5)
∗
√
3
1
∗
(1
6)
∆
2
1
(1
7)
I
n
Fig.4(
a)
, k
1
eq
u
a
tion
is represen
ted
b
y
th
e v
e
rtical lin
e wh
ich
in
clin
es at
1
2
0
°
to
α
ax
is. Th
e resu
lt
o
f
th
is equ
a
tion
will on
ly pro
d
u
c
e two
stat
es, eith
er
k
1
=
0
o
r
k1
= 1.
If
k
1
=
0
,
th
e referen
ce
v
ector is
in
th
e
regi
on o
f
∆
1
. I
ndi
cat
e
a bl
ue col
o
red pa
ral
l
e
l
l
i
n
e
bet
w
ee
n up
pe
r
an
d bel
o
w
t
r
i
a
n
g
l
e
t
o
s
e
parat
e
d
∆
3
fr
om
th
e
rest. T
h
is line
is refe
rre
r as
k
2
has a
di
st
anc
e
val
u
e
o
f
2
√
3
⁄
fr
om
α
p
l
an
e. R
e
f
e
r
t
o
pr
ev
ious Figu
r
e
2
,
th
e
val
u
e of
2
√
3
⁄
is a h
e
igh
t
of ev
ery s
m
all
trian
g
l
e fro
m
th
e
b
a
sed
to
t
h
e tip
s
o
f
trian
g
l
e. To id
en
tify wh
ich
eq
u
a
tion
s
sh
ould
b
e
u
s
ed
in
d
e
term
in
in
g
the trian
g
l
e and
v
ector co
m
p
o
n
en
ts, it is n
ecessary to
ev
alu
a
te th
e
co
nd
itio
n below [11
]
:
∗
√
3
∗
(1
8)
whe
r
e
∗
and
∗
are obtained using
(19) and
(20).
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
PED
S
I
S
SN
:
208
8-8
6
9
4
Impleme
n
tation
of Space Vec
t
or M
o
du
lat
o
r for
Cascade
d
H-Bri
d
ge
M
u
ltilevel Inverters (Syamim Sanusi
)
91
1
∗
3
2
∗
√
3
2
(1
9)
∗
√
3
3
∗
2
2
(2
0)
Fro
m
(18), if t
h
e co
nd
ition
is satisfied
, h
e
n
c
e th
e trian
g
l
e
∆
(f
or
=0,
1
or
3
)
an
d t
h
e
vect
o
r
com
pone
nt
s
∗
and
∗
can
b
e
calcu
lated
u
s
ing
(1
0),
(1
1) an
d (1
2). B
u
t,
if th
e con
d
ition
is
no
t satisfied, the triang
le
∆
and
the vect
or com
p
one
n
ts
∗
and
∗
can
b
e
o
b
t
ained u
s
ing
(
15)
, (
16)
an
d
(1
7)
.
3.
3.
C
a
l
c
ul
a
t
i
o
n of
O
n
-
D
ur
ati
o
n for
Sw
i
t
chi
n
g Vec
t
ors
The
ge
neral
re
prese
n
t
a
t
i
o
n
o
f
vect
o
r
di
ag
ra
m
s
as show
n i
n
Fi
gu
re
4
(b
)
and
Fi
g
u
r
e
4
(
c
) i
s
necessary
for
calcu
latin
g th
e o
n
-duration
o
f
ap
p
lication
for
each
v
ect
o
r
,
̅
,
̅
and
̅
.
In
gen
e
ral, th
e calcu
latio
n of
on
-
d
u
r
at
i
on ca
n be
obt
ai
ned
by
co
nsi
d
e
r
i
n
g
t
h
e vect
o
r
di
a
g
ram
i
a
appl
i
cabl
e
fo
r a
n
y
t
r
i
a
ngl
e i
n
a
n
y
sect
or
.
The
on
d
u
rat
i
o
ns f
o
r swi
t
c
hi
n
g
t
w
o ad
jace
nt
vol
t
a
ge
vect
or
s of
̅
∗
are cal
cul
a
t
e
d by
t
h
e
fol
l
owi
ng e
q
uat
i
o
ns
[1
1]
:
3
2
∗
1
√
3
∗
(2
1)
√
3
∗
∙
(2
2)
(b
)
Zo
o
m
ed
im
a
g
e
̅
∗
∗
∗
(c)
Zo
o
m
ed
im
a
g
e
Fi
gu
re 4.
Defi
ni
t
i
on o
f
R
e
fe
r
e
nce Vol
t
a
g
e
Vect
o
r
of Tri
a
ngl
e
(
∆
and
∆
)
̅
∗
∗
∗
̅
∗
∗
∗
̅
∗
̅
∗
∗
∗
̅
̅
̅
̅
∗
∗
∗
∗
∗
̅
̅
̅
β
Sector 1
α
r
Δ
2
Δ
1
Δ
0
Δ
3
(a)
k
2
k
1
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
088
-86
94
I
J
PED
S
Vo
l.
6
,
No
.
4
,
D
ecem
b
er
2
015
:
90
6 – 918
91
2
It
sh
oul
d be
n
o
t
ed t
h
at
t
h
e sw
i
t
c
hi
ng
peri
od
,
i
s
t
h
e t
o
t
a
l
su
m
m
a
t
i
on o
f
o
n
-
d
u
rat
i
o
ns
fo
r
appl
y
i
n
g
t
w
o
adjace
nt
vol
t
a
g
e
vect
o
r
s a
n
d
a vect
o
r
whi
c
h
i
s
de
fi
ne
d as
ori
g
i
n
poi
nt
t
o
st
at
e t
h
e
̅
∗
, i.e;
̅
,
̅
and
̅
.
Hen
c
e, th
e on-du
ration
fo
r ap
p
l
ying
th
e v
e
cto
r
(as th
e orig
in
po
in
t, as
sh
own
in
Figure 4
)
can
b
e
si
m
p
l
y
obt
ai
ne
d a
s
:
(2
3)
3.
4.
C
a
l
c
ul
a
t
i
o
n of
D
u
t
y
R
a
ti
os
Maj
o
r co
n
s
i
d
eratio
n in
g
e
neratin
g pr
op
er
sw
itch
i
ng
stat
es (o
r
PW
M
sig
n
a
ls)
is to
p
r
ov
id
e
h
i
gh
deg
r
ees
of acc
uracy
a
nd l
i
n
e
a
ri
t
y
t
o
obt
ai
n
desi
re
d o
u
t
p
ut
vol
t
a
ge
s.
At
t
h
i
s
st
age, ap
p
r
o
p
ri
at
e p
u
l
s
e wi
dt
h
o
f
pre
-
swi
t
c
hi
n
g
st
at
es
(
,
,
and
for
eac
h phase a
r
e gene
rated based on
t
h
e
c
o
m
p
arison between
t
h
eir
resp
ectiv
e
d
u
t
y
ratio
an
d
th
e trian
g
u
l
ar wav
e
fo
rm
, as il
lu
st
rated
in
Fig
u
re 4. No
te th
at th
e p
r
e-switch
i
ng
states
are defi
ned t
o
det
e
rm
i
n
e t
h
e on-
d
u
rat
i
o
ns
of v
o
l
t
a
ge
vec
t
ors w
h
i
c
h a
r
e
val
i
d
fo
r eve
r
y
sect
or as w
e
l
l
as
trian
g
l
e.
Based
on
th
e
defin
itio
n
o
f
on
-d
uration
m
a
rk
ed
in
Fi
g
u
re 5, th
e du
ty ratio
s,
i.e.
,
and
for ea
c
h
pha
se i
n
t
e
rm
s of
o
n
-
d
urat
i
ons
i
n
t
h
e ca
s
e
of t
h
ree-level CHMI can b
e
calcu
lated u
s
ing
th
e
fo
llo
wing
equat
i
o
ns:
2
4
(2
4)
2
4
2
(2
5)
2
4
2
2
(2
6)
3.
5.
Gener
a
ti
on
of
Sw
i
t
chi
n
g S
t
a
t
es
Th
e switch
i
ng o
p
e
ration
s
of twelv
e
IGBTs in
th
e
three-lev
e
l CHMI are dr
iv
en
b
y
t
h
e switch
i
ng
status,
,
,
,
,
and
(an
d
als
o
their
respec
tive
com
p
lim
e
nted status
̅
,
̅
,
̅
,
̅
,
̅
and
̅
) w
h
i
c
h
are
ge
nerat
e
d f
r
om
a l
o
o
k
-
u
p t
a
bl
e.
The l
o
o
k
-
u
p t
a
bl
e re
qui
res i
n
f
o
rm
at
i
on o
f
num
ber
of
sect
or
,
t
r
i
a
ngl
e a
nd
pr
e-swi
t
c
hi
ng
st
at
es. The s
w
i
t
c
h
i
ng st
at
us i
n
t
h
e l
o
o
k
-
u
p t
a
bl
e
i
s
m
a
pped s
u
c
h
t
h
at
t
h
e s
w
i
t
c
hi
n
g
vector for e
v
ery sector as wel
l
as triangle satisfies th
e criteria an
d
th
e switch
i
ng
sequ
en
ce. It shou
ld
b
e
tak
i
ng
in
to
acco
u
n
t
that th
e in
form
at
ion
of pre-swit
ching states (
,
,
and
) will d
e
termin
e six
switch
i
ng
states
t
o
dri
v
e
up
pe
r swi
t
c
hi
n
g
de
vi
ces for eac
h l
e
g of t
h
ree H
-
b
r
i
dge i
n
vert
er
s (w
hi
l
e
t
h
e l
o
w
e
r swi
t
c
hi
n
g
d
e
vi
ces
are
drive
n
by t
h
e c
o
m
p
lim
e
nted states).
Fi
gu
re
5.
Ge
ne
rat
i
o
n
o
f
P
r
e-
S
w
i
t
c
hi
n
g
St
at
es
4
2
2
4
1
0
1
0
0
0
1
1
4
2
2
4
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
PED
S
I
S
SN
:
208
8-8
6
9
4
Impleme
n
tation
of Space Vec
t
or M
o
du
lat
o
r for
Cascade
d
H-Bri
d
ge
M
u
ltilevel Inverters (Syamim Sanusi
)
91
3
4.
DESC
RIPTI
O
N OF
E
X
PE
RIME
NTAL
SETUP
Thi
s
sect
i
o
n
d
e
scri
bes
t
h
e t
a
sks
of
ci
rc
ui
t
s
o
r
c
o
m
pone
n
t
s em
pl
oy
ed t
o
set
u
p
t
h
e
e
xpe
ri
m
e
nt
al
platfo
rm
for v
e
rify
ing t
h
e ef
fe
ctiveness
of
S
V
M
co
ntr
o
l al
g
o
rith
m
fo
r three-lev
e
l inv
e
rt
ers. Fro
m
Fig
u
re 6, it
can
b
e
n
o
ticed th
at two
con
t
ro
ller bo
ard
s
are u
tilized
to
p
e
rfo
r
m
th
e tas
k
s of SVM con
t
ro
l alg
o
rith
m. The
reaso
n
o
f
usi
n
g t
w
o c
ont
r
o
l
l
e
r boa
r
d
s i
s
t
h
at
t
o
m
i
nim
i
ze t
h
e com
put
at
ional
b
u
r
de
n o
f
t
h
e
m
a
i
n
cont
rol
l
e
r
boa
r
d
whi
c
h
i
s
DS
11
0
4
R
&
D C
ont
r
o
l
l
e
r
B
o
a
r
d
.
In i
m
pl
em
ent
i
ng t
h
e S
V
M
,
i
t
s
t
a
sks are
di
st
ri
b
u
t
e
d i
n
t
o
t
w
o pa
rt
s, i
n
w
h
i
c
h eac
h pa
rt
i
s
per
f
o
r
m
e
d by
a si
ngl
e c
o
nt
r
o
l
l
e
r bo
ar
d.
In
doi
ng
so
, t
h
e
bu
r
d
en
o
f
cal
c
u
l
a
t
i
on ca
n
be
m
i
nim
i
zed;
t
h
i
s
i
n
t
u
rn al
l
o
ws t
h
e
com
put
at
i
on i
n
t
h
e m
a
i
n
co
nt
r
o
l
l
e
r
boa
rd
pe
r
f
o
r
m
e
d at
hi
g
h
sam
p
l
i
ng fre
q
u
e
ncy
.
4.
1.
D
S
11
04
R
&
D Co
nt
ro
ller B
o
a
r
d
The D
S
1
1
04 R
&
D C
o
nt
r
o
l
l
e
r
B
o
ard i
s
k
n
o
w
n
as t
h
e m
o
st
p
o
we
rf
ul
co
nt
r
o
l
l
e
r boar
d
w
h
i
c
h i
s
wi
del
y
use
d
i
n
i
n
du
st
r
y
and
uni
versi
t
y
fo
r
de
vel
o
pi
n
g
sy
st
em
and
r
a
pi
d
co
nt
r
o
l
pr
ot
ot
y
p
i
n
g.
T
h
e
m
a
jor t
a
s
k
s
of
SVM
algorithm
are execute
d
using
t
h
e DS1104, which
i
n
clude:
Determ
in
atio
n
o
f
sector
, triang
le
∆
, vect
o
r
c
o
m
ponent
s
wi
t
h
i
n
a sect
or
o
r
t
r
i
a
ngl
e
∗
and
∗
.
C
a
l
c
ul
at
i
on of
on
-
d
u
r
at
i
o
n
and
C
a
l
c
ul
at
i
on of
dut
y
rat
i
o
s
,
and
Al
l
t
a
sks l
i
s
t
e
d
ab
ove a
r
e i
m
pl
em
ent
e
d usi
n
g
M
A
TL
AB
-Si
m
uli
nk
bl
oc
k
di
ag
ram
wi
t
h
speci
al
R
eal
-
Tim
e
Int
e
r
f
ac
e (R
T
I)
I
n
p
u
t
/
Out
p
ut
bl
oc
k
s
. F
r
om
Fi
g.
6
,
t
h
e
D
S
1
1
0
4
R
&
D
C
o
nt
r
o
l
l
e
r B
o
ar
d
p
r
ovi
des
in
fo
rm
atio
n
of sector
(3-b
it),
trian
g
l
e
∆
(4
-bit
) a
n
d
pre
-
switc
hin
g
states
,
an
d
in
d
i
g
ital ou
tpu
t
form
. In
th
is case th
e d
ecimal n
u
m
b
e
rs of in
fo
rm
ati
o
n
ar
e conv
er
ted
in
to
G
r
ay Code
to
pr
ev
en
t sp
ur
ious
out
put
by
al
l
o
wi
n
g
onl
y
one
bi
t
o
f
bi
nary
di
gi
t
cha
nge
s f
o
r
t
w
o
su
ccessi
v
e
val
u
e
s
.
Fi
gu
re
6.
Ex
pe
ri
m
e
nt
al
Set
up
DS
11
0
4
R&
D
C
ont
r
o
l
l
e
r B
o
a
r
d
ALTER
A Cyclone
III FP
GA
DEO
Boa
r
d
Gate Driv
er Ci
rcu
its
In
verte
r
Three
-
Phase
Resistiv
e-In
ductiv
e lo
ad
∆
Is
olated
DC
Supplies
4-
b
i
t
3-
b
i
t
1-
b
i
t
1-
b
i
t
1-
b
i
t
̅
̅
̅
̅
̅
̅
̅
̅
Con
t
rolDes
k
Graphical Progr
amming
A
pp
roach
Observation/
ana
l
y
s
is
via Oscilloscope
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
088
-86
94
I
J
PED
S
Vo
l.
6
,
No
.
4
,
D
ecem
b
er
2
015
:
90
6 – 918
91
4
4.
2.
Alter
a
FP
GA
DEO Co
ntr
o
ller
Bo
ard
The
Al
t
e
ra Fi
e
l
d Pr
o
g
ram
m
abl
e
Gat
e
Ar
ray
s
(FP
G
A)
DE
O C
o
nt
r
o
l
l
e
r B
o
ar
d i
s
kn
o
w
n
t
o
ha
ve
hi
g
h
-
spee
d cl
oc
k
w
h
i
c
h i
s
su
pe
ri
o
r
t
o
exec
ut
e l
o
gi
cal
o
r
di
gi
t
a
l
o
p
erat
i
o
n.
Fr
o
m
Fi
gure
6, i
t
can
be see
n
t
h
at
t
h
e
FPGA recei
ves
inform
ation of sector
(3
-b
it),
trian
g
l
e
∆
(4
-b
it) and
pre-switch
i
ng
states
,
an
d
in
d
i
g
ital Gray co
d
e
. Th
e
FPGA is respon
sible to
p
e
rform
t
h
e task
o
f
selectio
n
of ap
propriate switchi
ng states
base
d on
the
i
n
form
ation received.
In
practice, it is co
m
p
u
l
so
ry to
p
r
ov
id
e b
l
ank
i
ng
ti
m
e
fo
r upp
er and
lo
wer IGB
T
s switch
i
ng
ope
rat
i
o
n t
o
a
voi
d
sh
ort
ci
r
c
ui
t
co
n
d
i
t
i
ons
. T
h
e
bl
a
nki
n
g
t
i
m
e i
s
set
app
r
oxi
m
a
t
e
ly at
2
b
y
selectin
g
app
r
op
ri
at
e t
h
r
e
sh
ol
d
val
u
e t
o
be com
p
are
d
wi
t
h
co
u
n
t
e
rs, i
m
pl
em
ent
e
d i
n
FP
GA
. F
o
r c
o
nve
ni
e
n
c
e
, t
h
e
ope
rat
i
o
n of bl
anki
ng
t
i
m
e
generat
i
o
n fo
r on
e
l
e
g of
a
n
y
o
r
-p
hase i
s
desc
ri
be
d,
wi
t
h
t
h
e
ai
d o
f
t
h
e bl
o
c
k
d
i
agram
an
d
timin
g
d
i
ag
ram
,
as illu
strated
i
n
Fig
s
. 7 an
d 8,
resp
ectiv
ely.
From
Fi
gu
re
7, i
t
ca
n
be
n
o
t
i
ced t
h
at
t
h
e
si
gnal
is th
e switch
i
n
g
state for
-p
hase whi
c
h
i
s
obt
ai
ne
d f
r
o
m
t
h
e vol
t
a
ge ve
ct
ors sel
ect
i
on
t
a
bl
e, as
m
e
nt
i
one
d ab
o
v
e. T
h
e bl
oc
k
of M
O
D
1
8 i
s
assi
g
n
ed t
o
di
vi
de t
h
e g
e
n
e
ral
cl
ock
fre
q
u
ency
of
FPG
A 3
3
.
3
3
by
18
,
hence t
h
e ne
w
cl
ock
fre
que
n
c
y
becom
e
s 1.
8
5
(o
r th
e
n
e
w per
i
od
5
40
). T
h
e
ne
w cl
ock freque
ncy signa
l
is de
signated as
as
s
h
o
w
n
i
n
Fi
gu
re
7
.
The
ne
w
cl
oc
k
f
r
eq
ue
ncy
si
g
n
al
is
u
s
ed
to
co
un
t
up
th
e
up
p
e
r coun
ter
(o
r th
e low
e
r cou
n
t
er
) b
y
1 for ev
er
y
54
0
wh
en
is activ
e h
i
gh
(o
r activ
e low), o
t
h
e
rwise
b
o
t
h
co
un
ters will b
e
reset to zero
.
No
te t
h
e
o
u
t
p
u
ts
of
uppe
r a
n
d l
o
we
r c
o
unter a
r
e designated
by the
res
p
ective signals
and
. By com
p
ari
n
g signals
a
an
d
d
with
a con
s
tant v
a
lu
e
o
f
4, the b
l
ank
i
ng
time is app
r
ox
im
a
t
ed
to
2
. I
n
c
r
e
a
si
ng
o
r
decre
a
si
ng t
h
e c
o
n
s
t
a
nt
v
a
lu
e
o
f
4
will en
large or redu
ce th
e
b
l
ank
i
n
g
tim
e, resp
ectiv
ely. No
te that, th
e sig
n
a
ls
and
̅
are for the
up
pe
r a
n
d
l
o
w
e
r
I
G
B
T
s
o
f
a
l
e
g
of
-p
hase
o
f
VS
I.
T
h
e
bl
anki
ng
t
i
m
e ge
nerat
o
r
i
s
d
u
p
l
i
cat
ed f
o
r t
h
e
ot
he
r
pha
ses an
d l
e
g
s
t
o
pe
rf
orm
the sam
e
effect
. The
n
t
h
e
ou
t
put
s o
f
s
w
i
t
c
hi
n
g
st
at
es fr
o
m
t
h
e bl
anki
n
g
t
i
m
e
g
e
n
e
rator are t
h
en fed
t
o
th
e
g
a
te driv
er circu
its.
5
40n
s
h
a
4
d
̅
4
Fi
gu
re
8.
Ti
m
i
ng
Di
a
g
ram
of
B
l
anki
n
g
Ti
m
e
Ge
nerat
i
o
n
fo
r
-
P
h
a
se and
A
ny Leg
Clk
MOD
18
Up
p
er
Counte
r
Clk
Clear1
Lo
w
e
r
Counte
r
Clear1
Comparator
Clk
Clo
c
k
h
a
16-bit
d
16-bit
C
out
1
C
out
4
C
l
ko
ut
uc1
lc1
Cin
1
Cin
4
̅
4
Ci
n
if
4
Ci
n
if
0
1
C
out
Fi
gu
re
7.
B
l
oc
k
Di
ag
ram
of
B
l
anki
n
g
Ti
m
e
Ge
nerat
i
o
n
fo
r
-
P
h
a
se and
A
ny Leg
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
PED
S
I
S
SN
:
208
8-8
6
9
4
Impleme
n
tation
of Space Vec
t
or M
o
du
lat
o
r for
Cascade
d
H-Bri
d
ge
M
u
ltilevel Inverters (Syamim Sanusi
)
91
5
4.
3.
Power Inver
t
e
r
and Gate
Dr
iver Circui
t
s
Fi
gu
re 9
de
pi
ct
s a gat
e
dri
v
e
r
and
po
we
r i
n
vert
er ci
r
c
ui
t
s
whi
c
h are
used
t
o
pr
o
duce
de
si
red
out
put
vol
t
a
ge
s. T
h
e pu
r
poses
of
us
i
ng gat
e
dri
v
e
r
ci
rcui
t
s
are t
o
pr
ovi
de i
s
ol
at
i
on bet
w
ee
n t
h
e el
ect
roni
c co
nt
r
o
l
ci
rcui
t
s
and
po
wer i
n
vert
er ci
rcui
t
s
an
d t
o
pr
ovi
de su
ffi
ci
en
t
powe
r
am
pl
i
f
i
cat
i
on fo
r swi
t
c
hi
n
g
IGB
T
s. I
t
can
be n
o
t
i
ced
fr
o
m
Fi
gure
9, t
h
ere are si
x
uni
t
s
of
H
-
B
r
i
d
ge i
nve
rt
er ci
rc
ui
t
s
t
o
est
a
bl
i
s
h
fi
ve-l
e
v
el
C
H
M
I
. Each
H-B
r
i
d
ge i
n
ver
t
er ci
rcui
t
i
s
s
u
ppl
i
e
d
by
a
n
i
s
ol
at
ed
DC
vol
t
a
ge s
u
ppl
y
.
Fr
om
t
h
i
s
fi
gu
re,
i
t
al
so sh
o
w
s t
h
at
t
h
e
FPGA c
o
ntroll
er
boa
rd and
I/
O inte
rface
DS1104 ca
rd.
4.
4.
Three-Phase
and Series
Connected
Resis
t
ive
and Inductive Loads
A three
-
phase
and series connecte
d
resistive and
in
du
ctiv
e lo
ad
is emp
l
oyed
for analyzing the
per
f
o
r
m
a
nce of SVM
fo
r t
w
o
-
l
e
vel
,
t
h
ree
-
l
e
vel
and fi
ve-l
e
v
el
of i
n
vert
er
s. The l
o
ad i
s
r
e
prese
n
t
e
d
by
a l
o
ad
reactor m
odel
MV 1101 from
TERCO,
whe
r
e its specification and
rated
condition a
r
e
given in Ta
ble 1.
Table 1.
L
o
a
d
Reactor
Pa
rameters
Param
e
ters Value
R
a
t
e
d R
eact
i
v
e
Po
wer
2.
5k
V
A
r
Fre
qua
ncy
50
-
6
0
H
z
Rated
Vo
ltag
e
3
80V
Y
,
22
0V
Y
/
∆
Rated
step
r
e
gu
latio
n
2
8
7
.
81
m
H
, 28.08
2
Ω
5.
RESULT ANALYSIS OF TOTAL
HARM
ON
I
C
DI
S
T
O
R
TI
ON
(
T
HD
)
O
F
O
U
TPU
T
VO
LTAGE
This section e
v
aluates t
h
e total harm
onic distortio
ns (T
HD)
of output
voltage a
n
d the
accura
cy of
fund
am
en
tal o
u
t
pu
t vo
ltag
e
of t
h
ree
-
l
e
vel
i
nve
rt
ers
.
T
h
e e
v
al
uat
i
o
n i
s
ba
sed
on t
h
e si
m
u
l
a
t
i
on
res
u
l
t
s
(i
.e.
t
h
e val
u
e
s
o
f
THD a
n
d f
u
n
d
am
ent
a
l
out
p
u
t
v
o
l
t
a
ge are
obt
ai
ne
d
usi
n
g Fast
F
o
u
r
i
e
r
Trans
f
orm
s
(FFT)
analysis), as
well as com
p
arison with the e
x
perim
e
ntal results.
Th
e ev
al
u
a
tion
on THD of ou
tpu
t
vo
ltage and th
e accu
r
acy
of fu
nda
m
e
n
t
al ou
tput vo
ltag
e
resul
t
e
d
i
n
e
v
e
r
y
l
e
vel
of i
n
v
e
rt
er i
s
al
s
o
ca
rri
ed
o
u
t
at
di
f
f
ere
n
t
m
odul
at
i
on i
ndi
ces
,
0
.
3
and
0.9
. The
sim
u
l
a
t
i
on res
u
l
t
s
obt
ai
ne
d f
r
om
t
h
e eval
ua
t
i
on are
dem
onst
r
at
ed i
n
Fi
g
s
. 1
0
. F
r
om
t
h
ese fi
g
u
res
,
i
t
can be
obs
erved that t
h
e e
xpe
rim
e
ntal results a
r
e i
n
close a
g
reemen
ts with
t
h
e sim
u
lat
i
o
n
resu
lts. Sp
ecifical
ly, th
e
pat
t
e
rns
of
wa
ve sha
p
e an
d f
r
eq
ue
ncy
spect
rum
of out
put
vol
t
a
ge
s i
n
t
h
e
expe
ri
m
e
nt
al
resul
t
s
are si
m
i
l
a
r t
o
th
at ob
tain
ed
i
n
th
e sim
u
latio
n
resu
lts.
Th
e similariti
es b
e
tween
si
m
u
la
tio
n
an
d ex
p
e
rim
e
n
t
al resu
lts, allo
w th
e resu
lts o
f
THD and
fu
n
d
am
ent
a
l
out
p
u
t
v
o
l
t
a
ge
obt
ai
ne
d
vi
a si
m
u
l
a
t
i
on t
o
be
assum
e
d si
m
ilar
wi
t
h
t
h
at
o
f
ex
peri
m
e
nt
al
resul
t
s
with
th
e error
b
e
tween
th
e si
m
u
la
tio
n
an
d calcu
l
ated
v
a
lu
e
s
i
s
i
n
si
gni
fi
ca
nt
an
d a
p
pr
o
x
i
m
at
ely
l
e
ss t
h
a
n
1 %
.
Fig
u
re
9
.
Pho
t
o
g
rap
h
of FPGA,
Gate
Dri
v
er Circu
its an
d Cascad
ed
H-Bri
d
g
e
Mu
ltilev
e
l
Inv
e
rter (C
HM
I)
FPGA
Gate
Di
CHMI
I
/O I
n
terfac
e
DS1104
Evaluation Warning : The document was created with Spire.PDF for Python.