ISSN: 1693-6
930
1
An Optim
i
zed
Square
Root
Algorithm
for Im
ple
m
entation in FPGA Hardware (To
l
e Sutikno
)
AN OPTIMIZED SQUARE ROOT ALGORITHM
FOR IMPLEMENTATION IN FPGA HARDWARE
Tole Sutikno
Dep
a
rtme
nt of Electrical En
ginee
ring, Un
iversita
s Ahm
ad Da
hlan (UAD)
Jln. Prof. Soe
pomo, Ja
nturan, Yogyaka
r
ta
55164, Tel
p
. +62
-
274
-3
7941
8, Fax. +62
-
27
4-564
604
e-mail: tole
@ee.uad.a
c
.id
A
b
st
r
a
k
Makal
ah
ini
m
enyajikan m
e
tode
perhi
tungan di
git-dem
i-digit ya
ng dio
p
tim
a
lkan untu
k
m
enyele
s
ai
ka
n pe
rhitu
nga
n a
k
a
r
kua
d
rat ya
ng
rum
i
t dalam
p
e
ra
n
g
kat
ke
ra
s,
sebag
ai al
gorit
m
a
sed
e
rh
ana
yang di
usulka
n untu
k
im
plem
entasi p
a
da field
pro
g
r
am
m
able ga
te array (FP
G
A).
Prinsi
p utam
a dari m
e
tod
e
ini adala
h
operasi
-
o
perasi pe
ngg
eseran
dua
-bit
dan pe
ngu
ra
ng-
m
u
ltipleks, u
n
tuk m
end
ap
atkan
im
plementasi
y
ang l
ebih
sed
e
rha
na d
an p
e
rhitunga
n yang
l
ebih
cep
a
t. Algorit
m
a
ini telah
digun
akan u
n
t
uk im
plem
entasi a
k
a
r
kua
d
rat bi
ner ta
k berta
nda
32-bit
dan 64
-bit b
e
r
ba
sis
FPGA
se
cara sukses. Hasil p
e
n
e
litian m
enun
jukkan b
a
h
w
a m
e
tode ya
ng
diusulkan
pal
ing efi
s
ien
sum
ber d
a
ya
pera
n
g
k
a
s
kera
s, bila
di
bandi
ng
kan
m
e
tode lainn
y
a.
Selain itu, strategi ini dap
at dengan m
udah di
kem
b
ang
kan untu
k
im
plem
entasi aka
r
kuad
rat
yan
g
lebih b
e
s
ar.
Kata kunci
:
perhitu
nga
n a
ngka-ol
eh
-dig
it, FPGA, Squ
a
re Root
A
b
st
r
a
ct
This
pape
r p
r
ese
n
ts a
n
opt
im
ized digit-b
y
-di
g
it calcul
a
t
ion m
e
thod to sol
v
e
com
p
licated
squ
a
re
ro
ot calcul
ation in
hard
w
a
r
e, a
s
a pro
p
o
s
ed
sim
p
le algo
rithm
for im
plem
entation in field
prog
ram
m
able gate
arra
y (FPGA). The
m
a
in prin
cipl
e of p
r
opo
se
d
m
e
thod is t
w
o-bit
shifting
and
subtractin
g-m
u
ltiplexi
ng op
eration
s
, in orde
r
to achi
eve a sim
p
le
r im
plem
entation and fast
er
cal
c
ulatio
n. The propo
se
d algorithm
ha
s con
duc
te
d to im
plem
ent
FPGA based
unsig
ned
32
-bit
and 64
-bit bi
nary squa
re
root succe
s
sfully. T
he results ha
ve sh
own that p
r
o
posed m
e
thod is
m
o
st efficient
of hardware
resou
r
ce com
pare
to
othe
r m
e
thods.
In addition,
the strateg
y
ca
n be
exp
and
ed to large
r
num
be
r easil
y.
Key
w
ords
:
d
i
git-by-di
g
it calcul
ation, FPGA, Square
Root
1. INTRODUCT
I
ON
It is well
-kno
wn that th
e d
i
rect to
rq
ue
cont
rol m
e
tho
d
(
D
T
C
) fo
r
AC moto
rs
h
a
s
simpl
e
stru
cture an
d
good
beh
aviors such a
s
f
a
st
torque re
spo
n
se,
no requireme
nts for
PWM
pul
se
gene
ration, n
o
req
u
irem
en
ts for co
ordi
nate tr
an
sformation, no p
o
sition e
n
cod
e
r an
d cu
rre
n
t
regul
ators [1-7].
The DT
C al
gorithm is
usu
a
lly implement
ed by
serial
cal
c
ulation
s
based on a
Microcontroll
er o
r
Digital
Signal Pro
c
essing
(DSP
) [8-1
1]. The
s
e a
r
e truly
softwa
r
e
-
ba
sed
platform
and
not
ade
quat
e to im
plem
ent a
control
metho
d
s wh
ich
re
quire v
e
ry hi
gh
sp
e
ed
respon
se. A
s
suitabl
e
soluti
on, it is propo
sed
FPGA
to
sup
port
execution very fa
st tasks [1
2-14
].
Ho
wever, it is not ea
sy to implement
DTC
in FP
GA hard
w
a
r
e. One of problem ha
s b
een
addresse
d
mainly in complicated
square ro
ot cal
c
ulatio
n. It is hard t
o
impleme
n
t on
FPGA [15-17]
.
There m
any
algorith
m
s ha
s p
r
o
p
o
s
ed
to solve
sq
ua
re
root,
su
ch
as
Rou
gh
e
s
timation
[18], Babyloni
an m
e
thod
[1
9], expone
ntial ide
n
tity
[20], Taylor-Se
r
i
e
s Exp
a
n
s
ion
Algorith
m
[2
1],
Ne
wton-Ra
ph
son meth
od [
22-2
4
], and
seq
uential al
gorithm (digit
-by-di
git cal
c
ulation meth
od)
[25-29]. Neve
rthele
ss, the
method
s abo
ve usually do
not focus to
solve squa
re
root problem
in
DTC im
plem
entation ba
se
d on FPGA. This pa
pe
r pr
opo
se
s digit-by-digit ca
l
c
ul
ation method
as
a si
mple
st
ra
tegy to
solve
co
mpli
cated
sq
ua
re
ro
ot. The
p
r
op
osed im
pleme
n
t
ation st
rateg
y
is
different co
m
pare
d
to strat
egie
s
in [25-29]. An
optimization i
s
also done by eli
m
inates
circu
i
try
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 8, No. 1, April 2010 : 1 - 8
2
that is not ne
eded. It is ad
dre
s
sed to suppo
rt
DT
C impleme
n
tatio
n
in FPGA hard
w
a
r
e, and
in
hope
s that it gives ri
se si
m
p
ler im
pl
eme
n
tation and fa
ster
cal
c
ulatio
n.
2.
DIGIT-
BY-
D
IGIT CAL
C
U
L
A
TIO
N
METHO
D
In digit-by-di
g
it calculatio
n method, th
e
each digit
of the squ
a
r
e ro
ot is fo
und in a
seq
uen
ce
wh
ere it o
n
ly on
e digit of the
squ
a
re
ro
ot
is ge
nerated
at eac
h iteration [29]. It ha
s
several a
d
va
ntage
s, such
as: eve
r
y di
git of the
r
oot f
ound
is kno
w
n to b
e
co
rre
ct an
d it
will
not
have to
be
ch
ange
d late
r; i
f
the squa
re
root ha
s to
ex
pand, it
will t
e
rmin
ate afte
r the
la
st digit
is
found; an
d th
e algo
rithm
works fo
r any n
u
mbe
r
ba
se
(of cou
r
se the
pro
c
e
s
s dep
e
nds
on n
u
mb
er
base).
(a)
(b)
Figure 1. The
example of digit-by-di
g
it calcul
ati
on to solve squ
a
re
root: (a) resto
r
ing algo
rithm;
(b) n
on re
sto
r
ing algo
rithm
Figure 2. The
example of using mo
dified
non re
st
ori
n
g
digit-by-di
g
it cal
c
ulatio
n algorithm to
solve squa
re
root
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
■
An Optim
i
zed
Square
Root
Algorithm
for Im
ple
m
entation in FPGA Hardware (To
l
e Sutikno
)
3
In gene
ral, th
is metho
d
ca
n be divide
d
in two
cla
s
se
s, i.e. re
stori
ng an
d non
restori
n
g
digit-by-digit
algorith
m
[29
]. In restori
n
g algo
rithm,
the pro
c
e
d
u
r
e is
comp
osed by taki
ng
th
e
squ
a
re
root
obtaine
d
so f
a
r, ap
pen
din
g
01
to it
an
d subtra
cting
it, prop
erly
shifted, fro
m
the
curre
n
t remai
nder. Th
e 0 in 01 co
rresp
ond
s to mu
ltiplying by 2; the 1 is a ne
w gue
ss bit. The
new
root
bit d
e
velope
d is t
r
uly 1, if the re
sulti
ng
rem
a
i
nder is
po
sitive, and vice
versa
is
0, which
the remain
de
r must be re
store
d
by adding the qua
nt
ity just subtracted. It is different, in non
resto
r
in
g alg
o
rithm d
oes
not re
store the subtra
ctio
n
if the re
su
lt was n
egati
v
e. Instead,
it
appe
nd
s a
11
to the
root d
e
velope
d so f
a
r a
nd
on
the
next iteratio
n
it perfo
rm
s a
n
additio
n
. If the
addition
ca
uses a
n
overflo
w
, then o
n
the next iter
at
ion you g
o
b
a
ck to the
su
btractio
n mo
de
[30]. The Fi
g
u
re
1 i
s
the
e
x
ample give
s to take the
b
i
nary
squ
a
re
root of
010
11
101
(eq
u
ivale
n
t
with 93 de
cim
a
l).
A little different than non
restori
ng di
git-by-digit
alg
o
ri
thm in Figu
re
1 (b), a m
odi
fication
as sho
w
n on
Figure 2 ca
n
be co
ndu
cted
to give
simpl
e
r implem
ent
ation and fa
ster cal
c
ul
ation.
In this modification, it only uses subtract
ope
ration
and appe
nd
01, while ad
d operation
and
appe
nd 1
1
is
not used. Thi
s
pa
pe
r ad
op
ts this m
odification to impl
e
m
ent un
sign
e
d
64
-bit bin
a
ry
squ
a
re
root b
a
se
d on FPG
A
.
3.
PROPOSED
SQUA
RE R
O
OT ALG
O
RI
THM
Samavi, et al. [29] has im
proved
cla
s
si
cal no
n-
re
sto
r
ing di
git-by-d
igit squa
re
ro
ot circuit
by eliminate
redu
nda
nt bl
ocks. T
hei
r circuit i
s
refe
rred to a
s
th
e
red
u
ced a
r
e
a
non
re
sto
r
i
ng
circuit. However, it still ba
sed on constant digit of
01
or 11 and add-subtract as
the main
buil
d
ing
block (still ref
e
r
to Figure 1
b). Thi
s
paper offers
a
simple alternat
ive sol
u
tion t
hat it only uses
subtract
s op
e
r
ation a
nd a
p
pend
s 0
1
. As con
s
e
que
n
t, the su
btra
ct-multiplex is u
s
ed
as th
e m
a
in
building bl
ock (refer to Figure 2
)
. The
princi
ple of prop
osed alg
o
rithm can b
e
descri
bed
as
s
h
ow
n
in
F
i
gu
r
e
3
.
Step 0.
Start
Step 1.
Initialization radicand (the n-bit number will be squared root), quotient
(the result of squared root), and remainder. To calculate square root of a
2n bit number, it needs n stage pipelines to implement the proposed
algorithm.
Step 2.
Beginning at the binary point, divide the radicand into groups of two
digits in both direction.
Step 3.
Beginning on the left (most significant bit), select the first group of
one or two digit (If n is odd then the first groups is one digit, and vice
versa)
Step 4.
Choose 1 squared, and then subtract.
Fist developed root is “1” if the result of subtract is positive, and vice
versa is “0”
Step 5.
Shift two bits, subtract guess squared with append 01.
Nth-bit squared is “1” if the result of subtract is positive, and Because
of subtract operation is done
else
Nth-bit squared is “0”, and not subtract
Step 6.
Go to step 5 until end group of two digits
Step 7.
End
Figure 3. The
princi
ple of p
r
opo
se
d algo
rithm to solve squ
a
re
root
A simple ha
rdware imple
m
entation of
the
non-re
storing digit
-
b
y
-digit algo
rithm for
unsi
gne
d 6-b
i
t square roo
t
by an arra
y structu
r
e is sh
own in
Figure 4. The radi
can
d
is P
(P5,P4,P3,P2,P1,P0), U (U2,U1,U
0) a
s
quotient a
nd
R (R4,R3,R2,
R1,R0) a
s
re
mainde
r. It can
be sh
own th
at the implementation ne
eds 3 sta
ge
pipeline
s
. Th
e main buildi
ng blocks of the
array are blo
c
ks called a
s
controll
ed su
btract
-m
ultiplex
(CSM). Fi
gure 5 p
r
e
s
e
n
t the details
of a
CSM. Input of the building
block
is x,y,b and u, and
as output is
bo (bo
r
row) and d (result). If
u=0, then d
<
=x-y-b el
se d
<=x.
The g
ene
rali
zation of
simpl
e
implem
enta
t
ion of
the
no
n-resto
r
ing
di
git-by-di
g
it al
gorithm
for un
sign
ed
n-bit squa
re root by an array stru
ct
ure is shown in Fi
gure
6. Each
row
(sta
ge)
of
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 8, No. 1, April 2010 : 1 - 8
4
the circuit in
Figure 6 ex
ecute
s
on
e-it
eration of
th
e non-re
stori
ng digit-by
-di
g
it squa
re ro
o
t
algorith
m
, wh
ere it only uses s
ubtra
ct
s operation an
d
appen
ds 0
1
.
Figure 4. A simple ha
rd
ware impleme
n
ta
tion of the non-resto
r
ing
digit-by-digit algorith
m
for unsi
gne
d 6-bi
t square ro
ot
Figure 5. Internal st
ru
cture
of
a CSM blo
c
k
Figure 6. A simple ha
rd
ware impleme
n
ta
tion of
the non-resto
r
ing di
git-by-di
g
it algorithm for
unsi
gne
d n-bi
t square ro
ot
To be o
p
timizer
ha
rd
ware
resou
r
ce sa
ving of the impleme
n
tatio
n
above, spe
c
iali
zed
entities
can
be created as buil
d
ing bl
ock compon
ents. It will
eliminate
circuitry that is
not
need
ed. As example, the
implementat
ion in Figur
e
6 for unsig
n
ed 6-bit squ
a
re root can
be
optimize
d
b
e
c
ome
a
s
sh
o
w
n i
n
Fi
gure
7 (i
n thi
s
ca
se, the
remai
n
der is igno
re
d, be
cau
s
e
in
the
DTC d
r
ive, it is not requi
red). The
spe
c
iali
z
ed entiti
e
s A, B, C,
D and E are
minimized
CSM
whe
n
input ybu=100, yu=00, u=0, yu=10, and y=
0
respe
c
tively,
and the rem
a
inder i
s
igno
red.
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
■
An Optim
i
zed
Square
Root
Algorithm
for Im
ple
m
entation in FPGA Hardware (To
l
e Sutikno
)
5
The gen
eralization of op
timized si
mp
le implem
ent
ation of the non-re
storin
g digit-by-digit
algorith
m
for unsi
gne
d n-bi
t square ro
ot is sh
own in Figure 8.
Figure 7. Optimized
simpl
e
hard
w
a
r
e imp
l
ementati
on o
f
the non-rest
oring di
git-by-digit algorith
m
for unsi
gne
d 6-bit sq
ua
re root
Figure 8. Optimized
simpl
e
hard
w
a
r
e imp
l
ementati
on o
f
the non-rest
oring di
git-by-digit algorith
m
for unsi
gne
d n-bit sq
ua
re root
4.
RESULTS A
ND AN
ALYSIS
In the previou
s
sectio
ns, o
p
timized
sim
p
le ha
rdware i
m
pleme
n
tatio
n
metho
d
of the no
n-
resto
r
in
g di
git-by-di
g
it alg
o
rithm for
sq
ua
re
root
and
th
e difficult ta
sk in
DTC to
calcul
ate
squ
a
re
root
were exp
l
ained. In thi
s
se
ction,
sim
u
lation
re
sults of 3
2
-bit a
n
d 64
-b
it squa
re root b
a
se
d
on
Altera APEX
20KE FPGA by using met
hod above are
presented, as shown in
Figure 9. In this
simulatio
n
, P is radicand
and U is qu
o
t
ient. T
he results sho
w
e
d
that the implementation ha
s
su
cceed
ed a
nd wo
rked p
r
operly.
Based
on
co
mpilation rep
o
rt, to implem
ent 32-bit and
64-bit
squ
a
re
root u
s
ing
op
timized
simple
hardware impl
eme
n
t
ation method
of the non-
re
storin
g digit-b
y
-digit algo
rithm are
nee
d
ed
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 8, No. 1, April 2010 : 1 - 8
6
256 and 1
0
2
3
logic elem
e
n
t (LE) re
spe
c
tively. The
compa
r
ison of results obtai
n
ed from different
impleme
n
tation method i
s
sho
w
n in Ta
b
l
e 1.
(a)
(b)
(c
)
(d)
Figure 9. Simulation re
sult
of n-bit squ
a
re root u
s
ing o
p
timized
simp
le hard
w
a
r
e i
m
pleme
n
tatio
n
method of the
non-re
storin
g digit-by-digi
t
algorithm
: (a
) 32-bit in decimal displ
a
y, (b) 32
-bit in
binary di
splay
,
(c) 64
-bit in
decim
al displ
a
y, (d) 64
-bit in binary di
spl
a
y
Table 1. The
comp
ari
s
o
n
o
f
logic eleme
n
t usag
e
No
Method
L
E
U
s
ag
e
32-bit squa
re
root
64-bit squa
re
root
1 Cla
ssi
cal
-
NR
1008
4092
2 Red
u
ced-A
r
e
a
-NR
632
2464
3 Modula
r
-NR
624
2468
4 Simple-X-Mo
dule
648
2488
5 Propo
se
d
256
1023
Note:
Altera APEX 20KE & Xilinx Virtex-E, 1 LC = 1 LE, and 1 CLB = 4 LE [31]
This
co
mpa
r
i
s
on
of LE o
r
logic
cell
(L
C) u
s
ag
e i
s
listed ba
sed
on
refe
ren
c
e
s
[
29] and
[30]. The number of emp
l
oyed LE ind
i
cate
s the
si
ze of the implement
e
d
ci
rcuit “hardwa
r
e
resou
r
ce”. Ta
ble 1 sh
owed
that propo
se
d method
is
most efficient
of hardware resou
r
ce. Based
on Fig
u
re 8,
the strategy i
s
very
ea
sy to be
expan
d
ed for la
rge
r
numbe
r to
so
lve com
p
licated
squ
a
re
root p
r
oble
m
in FPGA implemen
tation.
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
■
An Optim
i
zed
Square
Root
Algorithm
for Im
ple
m
entation in FPGA Hardware (To
l
e Sutikno
)
7
5. CO
NCL
USIO
N
This
cont
ribu
tion pre
s
e
n
te
d digit-by
-dig
it calculation
method a
s
a pro
p
o
s
ed
simple
strategy for i
m
pleme
n
tatio
n
in field pro
g
ramm
able g
a
te array (FP
G
A) ha
rd
ware mainly to solve
compli
cate
d
squ
a
re
root. The
main
prin
ciple
of
prop
osed
m
e
thod i
s
t
w
o
-
bit
shifting
and
subtractin
g-m
u
ltiplexing o
p
e
ration
s. T
h
e
pro
p
o
s
ed
strategy ha
s
co
ndu
cted to i
m
pleme
n
t FP
GA
based
un
sign
ed 3
2
bit
an
d
64
-bit bi
nary
sq
ua
re
r
oot
su
ccessfully. The re
sults
h
a
ve
sho
w
n
th
at
prop
osed met
hod is mo
st e
fficient of hardwa
r
e
resource compa
r
e t
o
other meth
ods. The m
e
thod
also
ca
n b
e
expand
ed to
large
r
n
u
mb
e
r
ea
sily, to solve co
mpli
cated
squa
re
root p
r
obl
em
in
FPGA imple
m
entation.
REFERE
NC
ES
[1].
Taka
ha
shi I, Nog
u
c
h
i T. A New Q
u
ic
k-
Re
spo
n
se an
d High
-Efficie
n
cy Co
ntrol
Strategy o
f
an Indu
ction
Motor.
IEEE
Trans
a
c
t
ions
on Indus
t
ry
A
pplications
. 1986; IA-22(5):
820-8
27.
[2].
Dep
enb
ro
ck
M. Dir
e
ct Self
Co
ntrol
(DS
C
)
of Inv
e
rte
r
-fed Ind
u
ctio
n
Mac
h
ine.
IE
EE Trans
.
on Powe
r Ele
c
troni
cs
. 198
8; 3(4): 42
0-4
29.
[3].
Habetler
TG,
Profumo F,
Pastorelli M,
To
lbert LM.
Direct
Torque
Control of
Induction
Machi
n
e
s
Usi
ng Space Vector Mo
dulati
on.
IEEE Transactions on
Industry Applications
.
1992; 28
(5
): 1045
-10
53.
[4].
Zhong
L, Ra
hman MF, Hu
WY, Lim
K
W
.
Anal
y
s
is
of Di
rect
To
rque
Co
ntrol i
n
Permane
nt
Magnet Syn
c
hrono
us M
o
tor Drive
s
.
IEEE Trans
ac
tions
on Power Elec
tronic
s
. 19
97;
12(3
)
: 528
-53
6
.
[5].
Fren
ch C, A
c
arnley P.
Dire
ct To
rqu
e
Control of Permane
nt Magnet Dri
v
es.
IEEE
Tran
sa
ction
s
on Indu
stry A
pplication
s
.
1996; 32(5): 1
080-108
8.
[6].
Yong
L, Zh
u
ZQ, Howe
D. Di
re
ct To
rqu
e
Control
of Brushle
s
s
DC Drive
s
with
Re
du
ce
d
Torq
ue Ri
ppl
e.
IEEE Trans
ac
tions
on Indus
t
ry
Applic
ations
. 200
5
;
41(2): 59
9-6
08.
[7].
Yong L, Z
h
u
ZQ, Howe
D. Co
mmutat
i
on-T
o
rq
ue
-Ri
pple Mi
nimization in
Dire
ct-To
r
q
ue-
Controlled P
M
Brushle
ss
DC
Drive
s
.
IEEE Transactions
on Indus
try
Applications
. 20
07;
43(5
)
: 101
2-1
021.
[8].
Bos
e
BK, Szc
z
esny PM.
A Microc
omputer-
ba
sed
Control
and
Si
mulation
of A
n
Advan
c
ed
IPM Synchronou
s M
a
ch
ine
Drive S
y
stem for
Electri
c
Veh
i
cle P
r
op
ulsi
on.
IEEE
Tran
sa
ction
s
on
Indu
strial Electro
n
ics
. 1
988; 35(4): 5
47-5
59.
[9].
Lianbi
ng L, Hexu S, Xiaojun W, Yong
qi
ng T.
A High
-Perfo
rm
ance
Dire
ct To
rqu
e
Cont
rol
Based o
n
DS
P in Perm
anent Magnet
Synchrono
us Motor Dri
v
e
.
Proce
eding
s of the 4th
Wo
rld Co
ng
ress on Intellig
ent Control a
nd Automatio
n
. 2002; 2: 16
22-1
625.
[10]. Weijie
L.
Im
plem
entation of
Direct Torque Cont
rol f
o
r Pe
rm
anen
t Magnet S
y
nch
r
on
ou
s
Motor
with Space Vecto
r
Modulatio
n Base
d on
DS
P
. 8th Intern
ational Confe
r
en
ce o
n
Signal Pro
c
e
ssi
ng. 200
6; 4: 101-1
04.
[11].
Cru
z
SMA, T
o
liyat HA, Cardo
s
o A
J
M.
DSP Imple
m
entation of
The Multiple
Referen
c
e
Frame
s
Th
eo
ry for The Di
agno
si
s of Stator
Fault
s
in
A DTC Indu
ction Moto
r Drive.
I
E
EE
Tran
sa
ction
s
on
Energy Co
nve
r
si
on
. 200
5; 20(2
)
: 329-335.
[12].
Monma
s
son
E, Cirstea M
N
. FPGA De
sign M
e
thodo
logy for Indu
strial
Co
ntrol
Systems: A
Rev
i
e
w
.
IEEE Tran
sa
ction
s
on Indust
r
ial Electro
n
ics
. 2
007; 54(4): 1
824-184
2.
[13].
Kowal
ski
CT, Lis J,
Orl
o
wska-K
owalska
T.
FPGA Im
plem
entation o
f
DT
C
Cont
ro
l Metho
d
for the Induction Motor Drive
. Th
e Internatio
nal Confere
n
ce on
Compute
r
a
s
a Tool
(EURO
C
O
N
).
2007:19
16-1
921.
[14].
Colli VD, Di
Stefano R, Marig
netti F, Scara
no M.
Desi
gn of a System
-on
-
Chip PMSM
D
r
iv
e
Se
ns
or
le
ss
C
o
n
t
r
o
l
. IEEE International Symposi
u
m on Indust
rial Elect
r
oni
cs (ISIE).
2007: 23
86-2
391.
[15].
Yamin
L, Wa
nming C.
Im
plem
entation o
f
Single Preci
s
ion
Floatin
g
Point Squa
re
Ro
ot on
FPGAs
. IEEE Sympos
ium on FPGA
for
Cus
t
om
Computing Mac
h
ines
.
Napa, California,
USA. 1997: 2
26-2
32.
[16].
Piromsopa
K, Aporntew
an
C,
Cho
n
g
s
titvatana P.
An
FPGA Im
plem
entation of
A Fixe
d-
Point
Squa
re
Root Ope
r
ati
on.
Int. Symp
. on
Comm
un
ication
s
and
Informatio
n T
e
ch
nolo
g
y
(ISCIT 200
1).
Chian
g
Mai, Thailan
d
. 200
1: 100-1
02.
[17].
Lacho
wicz S,
Pfleiderer
HJ.
Fa
st Eval
uation
of the
Squa
re
Ro
o
t
and
Othe
r
Nonli
nea
r
Functio
n
s in
FPGA
. 4th IEEE International Symposiu
m on El
ect
r
oni
c
Design,
Test
and
Applicatio
ns
(DELTA). 20
0
8
: 474-4
77.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 8, No. 1, April 2010 : 1 - 8
8
[18]. Erc
e
govac
MD.
On Digit-b
y
-Digit Metho
d
s for
Com
puting Ce
rtain
Functio
n
s
. Confere
n
ce
Re
cord of
th
e Fo
rty-First
Asiloma
r
Co
nferen
ce
on
i
n
Signal
s, S
y
stems an
d
Comp
uters
(ACSSC). 20
07: 338-342.
[19]. Koshel
eva
O.
Bab
y
loni
an Method
of Co
m
puting
The
S
quare
Root:
Ju
stificatio
ns Based
on
Fuzzy Techn
i
que
s and o
n
Com
putational Com
p
lexity
. Annu
al Meeting of the No
rt
h
American Fu
zzy Inform
ation Pro
c
e
ssi
n
g
Society (NA
F
IPS). 2009: 1-6.
[20].
Ligon
WB,
Monn IG, St
anzi
one
D,
Stivers F,
Underwo
od K
D
.
Im
plem
entation an
d
Analysis
of Num
e
ri
cal Com
ponents
f
o
r Re
config
u
r
able
Com
p
u
t
ing
. IEEE Proceedi
ngs
Aero
spa
c
e Confere
n
ce.19
99;
2:325-33
5.
[21].
Taek-Jun K,
Sondee
n
J, Drape
r
J.
Floating
-
Point Divi
si
on an
d Sq
uare
Ro
ot
Im
plem
entation Usi
ng A Taylo
r-Se
r
ie
s Expa
nsi
o
n
Algorithm
.
15th IEEE I
n
ternational
Confe
r
en
ce o
n
Electro
n
ics,
Circuits
an
d Systems (I
CECS). 200
8:70
2-70
5.
[22].
Kabuo
H, Ta
nigu
chi T, Mi
yoshi A, Ya
mash
ita
H,
Ura
no
M, Ed
amatsu
H, K
unino
bu S.
Accu
rate
Ro
undin
g
Sche
me for The
Ne
wton-Ra
ph
son Meth
od
Usi
ng Re
dun
dant Binary
Rep
r
e
s
entati
on.
IEEE Trans
ac
tions
on
Computers
. 1
994; 43(1): 4
3
-51.
[23].
Allie M, Lyons R. A Root o
f
Less Evil [Digital Signal Processin
g
].
IEEE Signal Processing
Maga
zine
. 20
05; 22(2
)
: 93-96.
[24].
Liang
-Kai
W,
Sch
u
lte M
J
.
De
cim
a
l
Flo
a
ting-Poi
n
t
Square Root Usi
ng Ne
wto
n
-Raph
so
n
Iteration
. 16th IEEE International
Conferenc
e
on Application-Spec
ific Sys
t
ems
,
Architecture Processo
rs
(ASAP). 2005: 309-315.
[25].
Tcho
umat
che
n
ko
V, Va
ssil
e
va T,
Gurov P.
A FPGA
Based S
qua
re-Ro
o
t Cop
r
oce
s
sor
.
Proceedi
ng
s of the 22
n
d
EUROMICRO
C
onfe
r
e
n
ce B
e
yond
2000:
Ha
rd
ware a
nd
Software
De
sign Strategie
s
. 1996: 520-5
25.
[26].
Taka
gi N,
Taka
gi K.
A VLSI Alg
o
rithm
for Integer Sq
ua
re-Ro
o
ting
. International
Symposium
on Intelligent Signal Processi
ng and Communi
cation
s (ISPACS).
2006: 626-
629.
[27].
Yamin
L, Wa
nming C.
Parallel-Arra
y Im
plem
entation
s
of A Non
-
Re
storin
g Sq
uare
Ro
ot
Algorithm
. IEEE Internatio
nal Co
nfere
n
c
e on
Com
p
uter Desi
gn: VLSI in Com
puters an
d
Processo
rs
(ICCD). 19
97: 690-695.
[28].
Xiumin W, Y
ang Z, Qia
n
g
Y, Shihua
Y.
A New Al
gorithm
for
De
signi
ng S
quare Roo
t
Cal
c
ulato
r
s B
a
se
d on FPG
A
with Pipelin
e Technol
og
y
. Ninth Intern
ational Confe
r
en
ce o
n
Hybrid Intellig
ent Systems (HIS). 2009: 9
9
-10
2
.
[29].
Samavi S, Sadra
badi A,
Fanian
A. Modul
a
r
Array
Structu
r
e Fo
r Non
-
Re
stori
ng Squ
a
re
Root Ci
rcuit.
Jou
r
nal of System
s Archite
c
ture
. 2
008;5
4
(10
)
: 957
-96
6
.
[30].
Compari
ng Altera APEX 20KE & Xilinx Vi
rtex-E
Logi
c Densities. Alte
ra Corporation. 2010.
Evaluation Warning : The document was created with Spire.PDF for Python.