Int
ern
at
i
onal
Journ
al of
El
e
ctrical
an
d
Co
mput
er
En
gin
eeri
ng
(IJ
E
C
E)
Vo
l.
8
, No
.
6
,
Decem
ber
201
8
, p
p.
4922
~
4931
IS
S
N: 20
88
-
8708
,
DOI: 10
.11
591/
ijece
.
v8
i
6
.
pp
4922
-
49
31
4922
Journ
al h
om
e
page
:
http:
//
ia
es
core
.c
om/
journa
ls
/i
ndex.
ph
p/IJECE
Analysis
of CMO
S Comp
arato
r in 90nm T
ec
hn
olo
gy with
Differ
ent Power
Reducti
on
Tec
hn
i
qu
es
An
il
Kh
ata
k
1
, Ma
noj
Kum
ar
2
, Sanj
ee
v Dh
ull
3
1
Depa
rtment of
Biom
edi
cal
Eng
i
nee
ring
,
GJ
US
&T
,
Ind
ia
2
US
ICT,
Guru G
obind
Singh I
ndr
apr
astha Unive
rs
ity
,
India
3
Depa
rtment of
ECE
,
GJ
US
&T
,
HIS
AR,
India
Art
ic
le
In
f
o
ABSTR
A
CT
Art
ic
le
history:
Re
cei
ved
Ja
n
26
, 2
01
8
Re
vised
Jun
17
, 201
8
Accepte
d
J
ul
22
, 2
01
8
To
red
u
ce
pow
er
consum
pti
on
of
reg
en
erati
ve
compara
tor
thr
ee
dif
fer
en
t
te
chn
ique
s
are
i
ncor
pora
t
ed
in
t
his
work.
The
se
te
chn
ique
s
provi
de
a
wa
y
to
ac
hi
eve
low
po
wer
consum
pti
on
through
th
ei
r
m
ec
hani
sm
that
alter
s
th
e
oper
ation
of
the
ci
rcu
it
.
The
se
t
e
chni
ques
ar
e
pse
udo
NM
O
S,
CV
SL
(ca
sc
ode
volt
ag
e
sw
it
ch
l
ogic
)/DCVS
(diff
ere
nt
ia
l
ca
scod
e
volt
ag
e
sw
it
ch
)
&
power
gat
ing
.
Init
i
all
y
reg
ene
r
at
iv
e
c
om
par
at
or
is
sim
ula
te
d
at
90
nm
CM
OS
te
chno
log
y
wit
h
0.
7
V
suppl
y
vo
ltage.
Re
sults
show
s
to
ta
l
powe
r
consum
pti
on
of
15.
02
µW
with
conside
rab
l
y
la
r
g
e
leaka
ge
cur
re
nt
of
52.
0
3
nA.
Further,
w
it
h
pseudo
N
MO
S
te
chni
que
tot
a
l
power
c
onsum
pti
on
inc
re
ase
s
to
126.
53
µW
while
CVS
L
sho
ws
t
ota
l
power
cons
um
pti
on
of
18.
94
µW
with
le
aka
g
e
cur
ren
t
o
f
1270.
13
nA.
More
the
n
90%
red
uction
is
at
t
ai
ned
in
tot
a
l
power
co
nsumpti
on
and
l
ea
ka
ge
cur
ren
t
b
y
e
m
plo
y
ing
th
e
power
gat
ing
tec
hnique
.
Mor
eov
er,
th
e
va
riation
s
in
the
power
c
onsum
pti
on
with
te
m
per
at
u
r
e
is
al
so
rec
ord
ed
for
al
l
thr
ee
rep
orte
d
t
ec
hni
ques
where
power
gating
ag
ai
n
show
opt
imum
var
ia
t
ions
wit
h
least
power
c
o
nsum
pti
on.
Four
m
ore
conve
nti
on
al
compara
tor
c
irc
ui
ts
are
al
so
sim
ula
ted
in
90nm
CMO
S
te
chnol
o
g
y
for
compari
son.
Com
par
iso
n
show
s
bet
te
r
result
s
for
reg
ene
r
at
iv
e
co
m
par
at
or
with
power
gat
ing
t
ec
hniqu
e.
Sim
ula
ti
ons
ar
e
exe
cu
te
d
b
y
emp
lo
y
ing
SP
ICE
b
a
sed
on
90
nm
CMO
S te
chnol
og
y
.
Ke
yw
or
d:
Ca
scod
e
Ca
scod
e
v
oltag
e
s
witc
h
l
ogic
(CVS
L
)
Current
d
ra
w
n
(CD)
Power
c
ons
umpti
on
(
PC)
Pseudo
Re
gen
e
rati
ve
Ultra
d
eep
s
ub
m
ic
ro
n
(UDS
M)
Copyright
©
201
8
Instit
ut
e
o
f Ad
vanc
ed
Engi
n
ee
r
ing
and
S
cienc
e
.
Al
l
rights
reserv
ed
.
Corres
pond
in
g
Aut
h
or
:
An
il
Kh
at
a
k,
Dep
a
rtm
ent o
f B
ME,
GJUS&T
, HIS
AR,
Har
ya
na,
India
.
Em
a
il
: kh
at
ak10an
il
@
gm
ai
l.c
om
1.
INTROD
U
CTION
Com
par
at
or
s
a
re
backb
on
e
of
m
any
i
m
per
ative
A
DC’s
&
DA
C
’s
ci
rc
uits
[1
]
-
[
3].
Co
nve
rter
ci
rc
uits
are
m
ai
nly
st
ud
ie
d
with
pa
ram
et
ers
i.e.
data
co
nversi
on
s
peed,
prec
isi
on
&
powe
r
co
nsum
ption
[4
]
.
Com
apr
at
or
s
a
re
the
m
os
t
essenti
al
com
po
nen
t
of
A
DC’s
as
it
s
pr
eci
sio
n
sig
nifica
ntly
def
i
nes
the
w
orki
ng
genre
of
thes
e
ci
rcu
it
s
[
5].
As
t
o
at
ta
in
t
hese
high
-
pe
rfor
m
ance
pa
ra
m
et
ers
ne
w
te
chn
i
qu
e
s
a
re
e
vo
l
ving
wh
ic
h
le
d
to
m
or
e
com
plex
an
d
hea
vy
arc
hitec
tures.
It
is
f
ound
that
en
erg
y
-
c
ontr
olled
a
pp
li
cat
io
ns
su
c
h
a
s
portable
cel
lp
hone
de
vices,
de
vices
in
healt
h
fiel
d,
wireles
s
netw
orks,
et
c
.,
re
qu
i
re
ef
fec
ti
ve
power
m
anag
i
ng
analo
g
-
to
-
dig
it
al
con
ve
rters
(
AD
Cs
)
f
or
a
longe
r
wor
king
extent
[6
]
,
[
7].
High
sp
ee
d
com
par
at
or
s
c
onsu
m
e
high
power
w
it
h
gr
eat
e
r
c
hip
area
w
hich
le
d
to
degra
de
d
reli
abili
ty
with
inc
rease
d
cost
of
c
ooli
ng
&
pack
a
ging
[
8]
. H
ence
, d
esi
gni
ng
a h
ig
h
-
sp
ee
d
c
om
par
at
or
ha
ving
sm
al
l
power
co
nsum
pti
on
be
com
es
dif
ficult
with least
pow
er supply
.
As
in
t
he
U
DSM
CM
OS
te
ch
no
l
og
y
dev
ic
e
siz
e
&
sup
ply
vo
lt
age
is
dwindlin
g
s
o
it
is
stren
uous
t
o
desig
n
a h
ig
h
-
s
peed
com
par
at
or
ci
rcu
it
f
or
lo
w
vo
lt
ages
le
ve
ls
[9
]
-
[11].
R
egen
e
rati
ve
c
om
par
at
or
is
ve
r
y
high
sp
ee
d
sta
ble
c
om
par
at
or
wit
h
la
r
ge
powe
r
co
nsum
ption
wh
ic
h
restrict
it
s
app
li
cabil
it
y
in
na
no
el
ec
tro
nics
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N: 20
88
-
8708
An
alysis of C
MOS Co
m
pa
r
at
or
in
90
nm Te
chnolo
gy
wi
th
Diff
erent P
ower Reduct
io
n
.
..
(
Anil K
ha
t
ak)
4923
do
m
ai
n
[12]
,
[
13
]
.
S
o,
in o
r
de
r
to
co
unte
r
th
is
reg
e
ner
at
ive co
m
par
at
or
is desig
ne
d
by
in
corp
or
at
in
g
dif
fer
e
nt
powe
r
re
duct
io
n
te
ch
niques
s
uch that i
t ca
n com
ply wit
h
th
e d
e
vice de
sig
ner’s
requisi
te
s.
Thr
ee
dif
fer
e
nt
power
reducti
on
te
c
hn
i
qu
es
are
em
plo
ye
d
on
reg
e
ne
rati
ve
com
par
at
or
[
13
]
in
or
de
r
to
drop
it
s
po
wer
co
nsum
pti
on
le
vel
so
t
ha
t
it
can
be
em
plo
ye
d
for
desi
gnin
g
high
pe
rfor
m
ance
ci
rc
uits
wit
h
m
arg
inal
powe
r
c
on
s
um
ption.
I
niti
al
ly
reg
e
ner
at
ive
com
pa
rator
is
desig
n
with
ps
e
udo
NM
OS
te
c
hnology
with
loa
d
ci
rc
uit
us
e
d
as
gro
unde
d
PMO
S
tra
ns
ist
or
[
14
]
.
A
fter
t
ha
t
CVSL
(casc
od
e
volt
ag
e
s
witc
h
log
ic
)/
DCVS
(
diff
e
re
ntial
cascod
e
volt
age
s
witc
h)
te
ch
no
l
og
y
is
use
d
on
reg
e
ne
rati
ve
com
par
at
or.
CVS
L
te
chn
iq
ue
ce
rt
ai
nly
i
m
pr
oves
par
am
et
ers
li
ke
ci
rcu
it
delay
,
powe
r
dissipa
ti
on
.
O
ne
m
or
e
point
w
hic
h
m
akes
this
te
chn
iq
ue
m
or
e
app
li
ca
ble
is
that
the
dig
it
al
ci
rcu
it
s
can
be
easi
ly
structur
e
d
on
the
basis
of
ta
bu
la
r
m
et
ho
ds
a
nd
Karna
ugh
m
a
ps
(
K
-
m
aps)
[15].
Thir
d
powe
r
re
du
ct
io
n
te
chn
i
qu
e
i
s
power
gati
ng
that
su
bst
antia
ll
y
red
uce
s
the
po
w
er
co
nsum
ption
of
re
generati
ve
com
par
at
or.
Power
gating
te
chn
iq
ue
pro
vi
des
a
m
echan
ism
thro
ug
h
gatin
g
de
vice
that
save
sta
ti
c
le
akag
e
powe
r
[16].
N
ano
scal
e
ci
rcui
ts
are
ver
y
su
btle
to
tem
per
at
ur
e
v
a
riat
ion
s
so t
em
per
at
ur
e
va
riat
ion
s
are
also
no
te
d
f
or these
com
par
at
or
s [1
7].
The
furthe
r
ar
r
ang
em
ent
of
ar
ti
cl
e
is
as
fo
ll
ows.
Segm
ent
2
pro
vid
es
a
has
ty
ov
er
view
of
the
ci
rc
uit
descr
i
ption
al
ong
wit
h
diff
e
re
nt
power
re
duc
ti
on
te
c
hn
i
qu
e
s
.
Sect
io
n
3
pro
vid
es
detai
ls
a
bout
t
he
sim
ulati
on
s,
analy
sis
of
po
wer
dissipati
on
a
nd
te
m
per
at
ur
e
var
ia
ti
on
for
diff
e
re
nt
powe
r
reducti
on
te
chn
i
qu
e
s
f
ol
lowing
com
par
ison wi
th con
ven
ti
on
a
l com
par
at
or
s
in
se
gm
ent 4
with conclu
sio
n & re
fer
e
nces i
n
the
end.
2.
DIFFE
RENT
POWER
R
E
D
UC
TI
ON T
ECH
NIQUES
2.1.
Reg
e
nera
tive
Co
m
pa
r
ator
in CMO
S Tec
hno
l
ogy
Figure
1
de
picts
the
arch
it
ect
ur
e
for
re
ge
nerat
ive
com
par
at
or
in
CM
OS
te
chnolo
gy.
T
he
input
is
a
diff
e
re
ntial
pair
am
plifie
r
whic
h
cosists
m
os
de
vices.
T
he
se
de
vices
are
ke
pt
in
feeb
le
inv
e
rsion
sta
t
e
tha
t
a
m
plifie
s
the
input
dif
fere
nc
e
sign
al
.
P
os
it
ive
fe
e
db
e
ck
&
reg
e
ner
at
io
n
a
re
em
plo
ye
d
in
the
sec
ond
st
age
&
they
becam
e
a
ct
ive
wh
e
n
res
et
po
int
is
init
ia
te
d
with
lo
w
sign
al
an
d
th
us
diff
e
ren
ce
si
gnal
is
transform
ed
to
VDD
&
VS
S
accor
dingly
as
final
dig
it
al
outp
ut
[
13]
.
T
his
com
par
at
or
a
rch
it
ect
ure
eas
il
y
i
nco
r
porate
s
la
rg
e
sign
al
s
due
to
high
in
pu
t
com
m
on
-
m
od
e
range.
A
n
ON
&
OF
F
c
hip
c
urr
ent
source
util
i
zed
in
this
sch
e
m
at
ic
pro
vid
es
ad
diti
on
al
functi
on
al
it
ie
s
with
w
hich
cu
rr
e
nt
can
be
m
od
ifie
d
for
va
rio
us
sam
pling
fr
eq
uen
ci
e
s
.
Inp
uts
are
give
n
th
rou
gh
tra
n
sist
or
T
1,
T
2
an
d
lo
wer
rail
of
M
OS
t
ran
s
ist
or
s
a
re
use
d
for
biasin
g
pu
rpose
.
Tw
o
N
OR
gate
s
are
us
e
d
at
e
nd
of
the
sc
he
m
at
ic
wh
ic
h
a
r
e
la
tc
hed
t
og
et
her
t
o
pro
vid
e
reg
e
ner
at
i
ve
fe
edb
ac
k
at
the output t
ha
t enh
a
nces
the
sp
ee
d of t
his c
om
par
at
or
.
Figure
1
.
Rege
ner
at
ive
Com
par
at
or in CM
O
S Tec
hnology
2
.2.
Reg
e
nera
tive
Co
m
pa
r
ator
in PS
UED
O
N
MOS T
echn
ol
ogy
Figure
2
sho
w
s
the
struct
ur
e
of
re
ge
ner
at
ive
com
par
at
or
e
m
plo
yi
ng
ps
e
udo
NM
OS
te
ch
no
l
og
y
[
14
]
,
[18]. T
his tech
nique e
ng
a
ges
a gro
unde
d P
MOS as
loa
d f
or N
M
OS
stee
r
ing
ci
rcu
it
.
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
8
, N
o.
6
,
Dece
m
ber
2
01
8
:
4922
-
4931
4924
Figure
2. Re
ge
ner
at
ive
co
m
par
at
or in
ps
e
udo NMO
S tec
hnol
og
y
This
ci
rcu
it
use
ov
e
rall
25
MOSFE
T.
Th
e
input
sig
nal
is
feed
thr
ough
T
1
&
T2
MO
SF
ET’s
w
hich
are
ar
range
d
in
fu
ll
y
dif
fer
e
nt
ia
l
pair.
Sec
on
d
sta
ge
is
us
e
d
for
am
plific
at
i
on
t
hen
tw
o
N
OR
gates
t
o
pr
ov
i
de
reg
e
ner
at
i
ve
fe
edb
ac
k
in
t
he
end.
G
1,
G2,
G3,
G
4,
G5
&
G6
MO
SFE
T
’S
are
str
uct
ured
in
pse
ud
o
NMOS
te
chnol
ogy
whic
h
form
s
the
t
wo
NO
R
gates
at
end
of
ci
rc
uit.
Inver
te
d
and
non
-
i
nverte
d
outp
ut
is
then
ta
ke
n
from
V
ou
t
(
-
) & V
out (+
).
2.3.
Reg
e
nera
tive
Co
m
pa
r
ator
in CVSL Tec
h
no
lo
gy
Figure
3
de
pic
ts
the
arch
it
ect
ur
e
f
or
reg
e
ne
rati
ve
com
par
a
tor
desi
gn
e
d
in
CVSL
te
ch
nolo
gy
[15]
[19].
CVSL
te
chnolo
gy
is
nothing
bu
t
casca
de
volt
age
swi
tc
h
log
ic
or
dif
fer
e
ntial
vo
lt
ag
e
switc
h
log
ic
[20].
MOS
tra
ns
ist
ors
G
1
t
o
G
12
are
use
d
t
o
de
sign
N
OR
ga
te
s
in
CVSL
te
chnolo
gy.
Ga
te
vo
lt
age
t
o
t
hese
MOSFET
’S
ar
e
app
li
ed
in
dif
fer
e
ntial
fo
rm
t
hat
al
te
r
the
on
&
of
f
ti
m
e
of
these
MOSF
E
T’S
w
hich
res
ults
in
decr
ease
in
power
dissi
patio
n
[21].
I
nputs
are
giv
e
n
to
th
e
gate
te
r
m
inal
of
T1
&
T
2
MOSFET
’s
&
ou
t
put
from
V
ou
t
(+)
and Vo
ut (
-
)
. T
he nu
m
ber
of
MOSFET
used
for
t
his ar
c
hite
ct
ur
e is
31.
Figure
3. Re
ge
n
erati
ve
Com
par
at
or in C
VSL t
ech
no
l
og
y
2.4.
Reg
e
nera
tive
Co
m
pa
r
ator
in P
ower G
at
i
ng
Tec
hn
olog
y
Fig
ure
4
de
picts
the
arch
it
ect
ur
e
for
re
ge
ne
rati
ve
com
par
a
tor
usi
ng
po
w
er
gatin
g
te
ch
no
l
og
y
[
16]
.
Power
Gati
ng
is
the
m
os
t
effe
ct
ive
te
ch
nique
to
re
duce
t
he
powe
r
dissip
at
ion
es
pecial
ly
le
akag
e
po
w
er
[
22]
[23].
P
re
-
c
harge
an
d
e
valuat
ion
m
od
e
are
f
ollow
e
d
for
th
e
execu
ti
on
of
fu
ll
com
par
is
on
phase.
D
ua
l
inp
ut
fu
ll
y
dif
fer
e
ntial
pair
that
co
m
pr
ise
of
T1
&
T2
M
OS
F
E
T’s
are
us
e
d
f
or
fee
ding
the
i
nput
sig
nal
an
d
outp
ut
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N: 20
88
-
8708
An
alysis of C
MOS Co
m
pa
r
at
or
in
90
nm Te
chnolo
gy
wi
th
Diff
erent P
ower Reduct
io
n
.
..
(
Anil K
ha
t
ak)
4925
from
Vo
ut
(
+
)
and
V
ou
t
(
-
)
.
P1
,
P2,
P3
&
P4
MO
SFET
S
are
us
e
d
to
operate
the
ci
rc
uit
in
Pr
e
-
cha
r
ge
an
d
evaluati
on
m
od
e.
G
1
to
G
8
MOSFET
’S
ar
e
us
e
d
for
des
ign
in
g
t
wo
nor
gates
at
the
e
nd
of
the
sc
he
m
at
ic
.
Substanti
al
power
reducti
on
is
ob
ta
ine
d
wi
th
this
te
chn
i
que.
T
otal
MOS
de
vices
us
e
d
in
this
te
ch
ni
qu
e
i
s
sam
e as CVSL
te
chnolo
gy.
Figure
4. Re
ge
ner
at
ive
Com
par
at
or in
Powe
r
G
at
ing Tec
hn
ology
3.
RESU
LT
S
AND DI
SCUS
S
ION
In
t
his
se
gm
ent
transient
a
na
ly
sis
fo
r
the
a
bove
f
our
ci
rc
uits
are
done
by
va
ryi
ng
c
ha
nn
el
widt
h,
su
pply
vo
lt
age
and tem
per
at
ure. Sim
ulati
on
s
are
perform
ed
us
in
g
s
pice
bas
ed 90 nm
CM
OS
te
c
hnology
.
Figure
5
gi
ves
the
grap
hical
detai
ls
re
gardi
ng
to
the
po
w
er
c
onsu
m
ption
a
nd
c
urren
t
dr
a
w
n
f
or
t
he
reg
e
ner
at
i
ve
c
om
par
at
or
[
8]
and
it
s
va
riat
io
n
with
c
hannel
widt
h
a
nd
s
upply
volt
age
res
pecti
vely
.
I
n
F
igure
5(
a
)
cha
nnel
width
is
va
rie
d
from
1
µ
m
to
5
µm
by
keep
in
g
the
s
upply
volt
age
c
on
sta
nt
(
0.7
V
).
Th
e
per
ce
ntage
inc
rease
due
to
the
var
ia
ti
on
i
n
cha
nnel
wi
dth
is
38.
67.
Fig
ure
5(b)
s
hows
var
ia
ti
ons
i
n
powe
r
dissipati
on
a
nd
cu
rr
e
nt
dra
wn
by
var
yi
ng
t
he
sup
pl
y
volt
age
from
1.
7
V
to
0.7
V
at
c
ha
nn
el
widt
h
of
1µm
.
Du
e
to
t
his
vari
at
ion
po
wer di
ssipati
on d
ec
re
ases b
y
98
per
c
ent.
(a)
(b)
Figure
5. (a
)
P
ow
e
r
c
onsu
m
ption
& c
urren
t
draw
n vs
c
ha
nnel
w
idth
;
(
b)
P
ow
e
r
c
onsu
m
ption
& c
urren
t
draw
n
vs
s
upply
vo
lt
a
ge
Table
1
gi
ves
the
detai
ls
re
ga
rd
i
ng
t
o
the
va
riat
ion
s
i
n
po
w
er
co
nsum
ption
an
d
c
urre
nt
draw
n
by
the
ci
rcu
it
to
the
va
ryi
ng
te
m
per
at
ur
e
in
dif
fer
e
nt
m
od
es
i.e.
le
akag
e
,
sta
ti
c,
dynam
ic
,
total
.
Power
co
nsum
ptio
n
and cu
rr
e
nt
dr
a
wn v
al
ues
s
ho
ws
var
ia
ti
ons
wh
e
n
te
m
per
at
ur
e
v
a
ries f
r
om
-
35
ᵒ t
o 8
0ᵒ
Ce
lsi
us
.
0.
6
0.
8
1.
0
1.
2
1.
4
1.
6
0
500
1000
1500
Sup
pl
y
v
ol
tage
(v
ol
ts)
Pow
e
r c
onsumpt
ion (µW
)
Cu
rre
nt
dra
w
n (µA
)
1
3
5
0
2
4
6
20
40
60
80
100
120
Cha
nn
e
l
w
id
th
(µm
)
Pow
e
r c
onsumpt
ion (µW
)
Cu
rre
nt
dra
w
n (µA
)
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
8
, N
o.
6
,
Dece
m
ber
2
01
8
:
4922
-
4931
4926
Table
1.
T
otal
, st
at
ic
, d
ynam
ic
& leaka
ge
po
wer co
nsum
pti
on of
reg
e
ne
rati
ve
com
par
at
or
at dif
fer
e
nt
tem
per
at
ur
e
Te
m
p
Leakag
e
Static
Dy
n
a
m
i
c
Total
PC (nW
)
CD (nA)
PC (µW
)
CD (µA)
PC (µW
)
CD
(µA)
PC (µW
)
CD (µA)
80ᵒ
1
2
2
.85
1
7
5
.50
1
3
.57
1
9
.39
1
.91
2
.73
1
5
.49
2
2
.12
75ᵒ
1
1
1
.62
1
5
9
.46
1
3
.62
1
9
.45
1
.84
2
.64
1
5
.47
2
2
.10
65ᵒ
0
9
1
.41
1
3
0
.59
1
3
.70
1
9
.58
1
.71
2
.44
1
5
.41
2
2
.02
55ᵒ
0
7
4
.02
1
0
5
.75
1
3
.78
1
9
.69
1
.56
2
.23
1
5
.35
2
1
.93
45ᵒ
0
5
9
.22
0
8
4
.60
1
3
.85
1
9
.79
1
.41
2
.02
1
5
.26
2
1
.81
35ᵒ
0
4
6
.76
0
6
6
.81
1
3
.90
1
9
.86
1
.25
1
.79
1
5
.16
2
1
.65
25ᵒ
0
3
6
.42
0
5
2
.03
1
3
.92
1
9
.89
1
.09
1
.56
1
5
.02
2
1
.46
15ᵒ
0
2
7
.96
0
3
9
.94
1
3
.92
1
9
.89
0
.93
1
.34
1
4
.86
2
1
.23
05ᵒ
0
2
1
.14
0
3
0
.20
1
3
.88
1
9
.83
0
.78
1
.12
1
4
.66
2
0
.95
00ᵒ
0
1
8
.28
0
2
6
.11
1
3
.84
1
9
.77
0
.71
1
.01
1
4
.55
2
0
.79
-
05ᵒ
0
1
5
.74
0
2
2
.49
1
3
.79
1
9
.70
0
.64
0
.91
1
4
.43
2
0
.62
-
15ᵒ
0
1
1
.56
0
1
6
.52
1
3
.64
1
9
.49
0
.51
0
.73
1
4
.16
2
0
.23
-
25ᵒ
0
0
8
.39
0
1
1
.99
1
3
.44
1
9
.20
0
.41
0
.58
1
3
.85
1
9
.79
-
35ᵒ
0
0
6
.04
0
0
8
.63
1
3
.18
1
8
.82
0
.32
0
.46
1
3
.50
1
9
.29
As
the
te
m
per
at
ur
e
in
creases
le
akag
e
&
dynam
ic
po
wer
c
on
s
um
ption
in
creases
on
the
oth
e
r
ha
nd
sta
ti
c
po
wer
c
onsu
m
ption
al
so
sh
ows
this
inc
reasin
g
patte
r
n
bu
t
sli
gh
t
va
riat
ion
at
hig
he
r
tem
per
at
ur
es
.
Th
us
,
the
total
powe
r
co
nsum
ption
al
so
increase
s
from
13
.50
µ
W
at
-
35ᵒ
C
to
15.
49
µ
W
at
80
ᵒ
C.
T
otal
powe
r
consum
ption
i
ncr
ease
s
by 14 perce
nt
du
e
to t
he
va
riat
ion i
n t
e
m
per
at
ure.
Figure
6
gi
ves
the
grap
hical
detai
ls
re
gardi
ng
to
the
po
w
er
c
onsu
m
ption
a
nd
c
urren
t
dr
a
w
n
f
or
t
he
reg
e
ner
at
i
ve
com
par
at
or
de
sign
e
d
us
i
ng
psu
edo
NMOS
te
c
hnology
an
d
it
s
var
ia
ti
on
with
cha
nn
el
wi
dth
an
d
Supp
ly
vo
lt
age
res
pecti
vely
.
I
n
F
i
gure
6
(a
)
c
hannel
width
is
var
ie
d
from
1
µ
m
to
5
µm
by
kee
ping
the
supp
l
y
vo
lt
age
co
ns
ta
nt
(
0.7
V)
.
Th
e
powe
r
c
on
s
um
pt
ion
is
m
axi
m
u
m
at
2.
5
µ
m
that
is
137.13
µ
W
at
t
his
ci
rcu
it
al
so
dra
w
m
a
xim
u
m
cur
ren
t
from
su
pply
vo
lt
age
.
Fi
gur
e
6
(
b)
s
hows
var
ia
ti
ons
in
powe
r
dissipati
on
a
nd
current
draw
n
by
var
yi
ng
the
su
pply
vo
lt
ag
e
fr
om
1.
7
V
to
0.7
V
at
chan
nel
wi
dth
of
1
µ
m
.
Du
e
to
this
var
ia
ti
on
m
inim
u
m
po
wer
co
ns
um
ption
is
a
t
0.7
V
t
hat
is
126.5
3
µ
W
a
n
d
m
axi
m
u
m
at
1.5
V
t
hat
is
1106.
00
µ
W
.
(a)
(b)
Figure
6.
(a
)
P
ow
e
r
c
onsu
m
ption
&
c
urren
t
draw
n vs
c
ha
nnel
w
idth
;
(
b)
P
ow
e
r
c
onsu
m
ption
& c
urren
t
draw
n
vs
s
upply
vo
lt
a
ge
Table
2
gi
ves
the
detai
ls
re
ga
rd
i
ng
t
o
the
va
riat
ion
s
i
n
po
w
er
co
nsum
ption
an
d
c
urre
nt
draw
n
by
the
ci
rcu
it
to
the
va
ryi
ng
te
m
per
at
ur
e
in
dif
fer
e
nt
m
od
es
i.e.
le
akag
e
,
sta
ti
c,
dynam
ic
,
total
.
Power
co
nsum
ptio
n
and
c
urren
t
dr
awn
values
s
hows
var
ia
ti
ons
wh
e
n
tem
perat
ur
e
va
ries
from
-
35
ᵒ
to
80ᵒ
Ce
lsi
us
.
As
the
tem
per
at
ur
e
in
creases lea
kage &
dynam
ic
p
ow
e
r
c
onsu
m
ption
i
ncr
ease
s
with a
ppr
ox
im
at
el
y 30
0%
vari
at
ion
.
0.
6
0.
8
1.
0
1.
2
1.
4
1.
6
0
500
1000
1500
Sup
pl
y
v
ol
tage
(v
ol
ts)
Pow
e
r c
onsumpt
ion (µW
)
Cu
rre
nt
dra
w
n (µA
)
1
3
5
2
4
20
40
60
80
100
120
140
160
180
200
220
Cha
nn
e
l
w
id
th
(µm
)
Pow
e
r c
onsumpt
ion (µW
)
Cu
rre
nt
dra
w
n (µA
)
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N: 20
88
-
8708
An
alysis of C
MOS Co
m
pa
r
at
or
in
90
nm Te
chnolo
gy
wi
th
Diff
erent P
ower Reduct
io
n
.
..
(
Anil K
ha
t
ak)
4927
Table
2.
T
otal
, st
at
ic
, d
ynam
ic
& leaka
ge
po
wer co
nsum
pti
on of
reg
e
ne
rati
ve
com
par
at
or
(psu
e
do
NMO
S)
at
diff
e
re
nt tem
pe
ratur
e
Te
m
p
Leakag
e
Static
Dy
n
a
m
i
c
Total
PC (µW
)
CD (µA)
PC (µW
)
CD (µA)
PC (µW
)
CD (µA)
PC (µW
)
CD (µA)
80ᵒ
1
5
.62
2
2
.31
0
8
0
.43
1
1
4
.90
3
2
6
.61
4
6
6
.59
4
0
7
.04
5
8
1
.49
75ᵒ
1
4
.84
2
1
.20
0
8
1
.36
1
1
6
.23
3
3
0
.40
4
7
2
.00
4
1
1
.77
5
8
8
.24
65ᵒ
1
3
.34
1
9
.06
0
8
3
.29
1
1
8
.99
0
8
0
.38
1
1
4
.83
1
6
3
.67
2
3
3
.82
55ᵒ
1
1
.92
1
7
.03
0
8
5
.31
1
2
1
.87
0
6
3
.16
0
9
0
.23
1
4
8
.47
2
1
2
.11
45ᵒ
1
0
.59
1
5
.13
0
8
7
.42
1
2
4
.89
0
5
1
.38
0
7
3
.40
1
3
8
.81
1
9
8
.30
35ᵒ
0
9
.35
1
3
.36
0
8
9
.64
1
2
8
.06
0
4
2
.21
0
6
0
.30
1
3
1
.85
1
8
8
.37
25ᵒ
0
8
.22
1
1
.74
0
9
1
.97
1
3
1
.39
0
3
4
.55
0
4
9
.36
1
2
6
.53
1
8
0
.75
15ᵒ
0
7
.19
1
0
.28
0
9
4
.42
1
3
4
.89
0
2
7
.51
0
3
9
.30
1
2
1
.93
1
7
4
.19
05ᵒ
0
6
.28
0
8
.98
0
9
7
.01
1
3
8
.58
0
2
0
.06
0
2
8
.66
1
1
7
.07
1
6
7
.25
00ᵒ
0
5
.87
0
8
.39
0
9
8
.35
1
4
0
.50
0
1
6
.10
0
2
3
.00
1
1
4
.46
1
6
3
.51
-
05ᵒ
0
5
.49
0
7
.84
0
9
9
.73
1
4
2
.48
0
1
2
.09
0
1
7
.27
1
1
1
.82
1
5
9
.75
-
15ᵒ
0
4
.82
0
6
.89
1
0
2
.61
1
4
6
.59
0
0
4
.00
0
0
5
.72
1
0
6
.62
1
5
2
.32
-
25ᵒ
0
4
.29
0
6
.13
1
0
5
.66
1
5
0
.95
-
0
0
3
.97
-
0
0
5
.68
1
0
1
.69
1
4
5
.27
-
35ᵒ
0
3
.89
0
5
.56
1
0
8
.90
1
5
5
.57
-
0
1
1
.31
-
0
1
6
.17
0
9
7
.58
1
3
9
.40
On
the
oth
e
r
hand,
sta
ti
c
powe
r
co
ns
um
pt
ion
f
ollows
inv
ers
e
ru
le
with
tem
per
at
ur
e
ha
ving
m
axi
m
u
m
value
of
10
8.90
µ
W
at
-
35ᵒ
C.
T
hu
s
,
the
total
powe
r
c
onsu
m
ption
al
so
inc
rea
ses
f
ro
m
97
.
58
µ
W
a
t
-
35ᵒ
C
to
407.0
4
µ
W
at
80
ᵒ
C.
T
otal
power
co
nsum
ption
increa
ses
by
31
7%
due
to
t
he
var
ia
ti
on
in
tem
per
at
ur
e
.
Figure
7
gi
ves
the
grap
hical
detai
ls
re
gardi
ng
to
the
po
w
er
c
onsu
m
ption
a
nd
c
urren
t
dr
a
w
n
f
or
t
he
reg
e
ner
at
i
ve
c
om
par
at
or
desi
gn
e
d
us
in
g
C
VS
L
te
c
hnol
ogy
an
d
it
s
var
i
at
ion
with
c
ha
nn
el
widt
h
a
nd
supp
ly
vo
lt
age
res
pecti
vely
.
Figure
7(a)
dep
ic
t
the
var
it
io
ns
w
he
n
channel
wi
dth
is
var
ie
d
fro
m
1
µ
m
to
5
µ
m
by
keep
i
ng
t
he
s
upply v
oltage c
on
sta
nt (0.
7
V
)
. Th
e
powe
r
co
ns
um
ption
for t
his circu
it
is m
axi
m
u
m
at 5
µ
m
that
is
256.6
7
µ
W
and
dra
w
m
axim
u
m
cur
ren
t
from
su
pp
ly
v
olt
age.
O
n
the
ot
her
si
de
in
fi
gure
7(b)
t
he
var
i
at
ion
s
in pow
e
r
dissi
pa
ti
on
and curr
e
nt d
ra
w
n
by v
a
ryi
ng
the s
uppl
y vo
lt
age fro
m
1
.
7
V
to
0.7 V at chan
nel
widt
h
of
1
µm
is
sh
own.
D
ue
to
this
var
ia
ti
on
m
ini
m
u
m
po
wer
consum
ption
is
at
0.
7
V
th
at
is
18
.94
µW
a
nd
m
axi
m
u
m
at 1.3
V
that i
s
39
9.59 µ
W.
(a)
(b)
Figure
7. (a
)
P
ow
e
r
c
onsu
m
ption
& c
urren
t
draw
n vs
c
ha
nnel
w
idth
;
(
b)
P
ow
e
r
c
onsu
m
ption
& c
urren
t
draw
n
vs
s
upply
vo
lt
a
ge
Table
3
gi
ves
the
detai
ls
re
ga
rd
i
ng
t
o
the
va
riat
ion
s
i
n
po
w
er
co
nsum
ption
an
d
c
urre
nt
draw
n
by
the
reg
e
ner
at
i
ve
com
par
at
or
design
e
d
usi
ng
CV
SL
te
chnolo
gy
to
the
var
yi
ng
tem
per
at
ur
e
in
dif
fer
e
nt
m
o
des
i.e
.
le
akag
e,
sta
ti
c,
dy
nam
ic
,
total
.
Power
c
onsu
m
ption
an
d
c
urren
t
dr
a
wn
val
ues
s
hows
va
riat
ion
s
w
he
n
tem
per
at
ur
e
va
ries
f
ro
m
-
35ᵒ
to
80ᵒ
Ce
ls
ius.
As
the
te
m
per
at
ur
e
inc
r
eases
le
aka
ge
powe
r
c
on
s
um
pt
ion
increases
with
appr
ox
im
at
ely
300%
per
ce
nt
var
ia
ti
on
m
axi
m
u
m
value
of
2378.
64
nW
at
80
ᵒ
C.
On
t
he
othe
r
hand,
sta
ti
c
po
wer
co
nsum
pti
on
inc
reases
w
it
h
increase
in
tem
per
at
ur
e
ha
ving
m
axi
m
u
m
value
of
0.9
3
µ
W
at
80
ᵒ
C.
Dynam
ic
power
c
ons
um
ption
show
s
abru
pt
natu
r
e
as
it
increases
and
dec
reas
es
with
te
m
per
at
ur
e
var
ia
ti
ons.
Ma
xim
u
m
value
of
dynam
ic
po
w
er
c
on
s
um
ption
i
s
23.
37
µ
W
at
65ᵒ
C
&
m
ini
m
u
m
value
of
16.
34
µ
W
at
-
5°
C
.
Th
us
,
t
he
total
powe
r
c
on
s
um
pt
ion
al
so
f
ol
lows
the
sam
e
abru
pt
nat
ure
as
dy
nam
ic
powe
r
consum
ption
f
ollows.
Total
powe
r
co
ns
um
ption
at
ta
ins
it
s
m
axi
m
u
m
value
of
24.
20µ
W
at
65ᵒ
C
and
m
ini
m
u
m
v
al
ue
of
16.77 µ
W a
t
-
5° C.
1
3
5
2
4
20
40
60
80
100
120
Cha
nn
e
l
w
id
th
(µm
)
Pow
e
r c
onsumpt
ion (µW
)
Cu
rre
nt
dra
w
n (µA
)
0.
6
0.
8
1.
0
1.
2
1.
4
1.
6
0
500
1000
1500
Sup
pl
y
v
ol
tage
(v
ol
ts)
Pow
e
r c
onsumpt
ion (µW
)
Cu
rre
nt
dra
w
n (µA
)
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
8
, N
o.
6
,
Dece
m
ber
2
01
8
:
4922
-
4931
4928
Table
3.
T
otal
, st
at
ic
, d
ynam
ic
& leaka
ge
P
ower
cons
um
pti
on of
reg
e
ne
rati
ve
com
par
at
or
(
CV
SL
Tech
no
l
og
y)
at
d
if
fer
e
nt tem
per
at
ur
e
Te
m
p
Leakag
e
Static
Dy
n
a
m
i
c
Total
PC (nW
)
CD (nA)
PC (µW
)
CD (µA)
PC (µW
)
CD (µA)
PC (µW
)
CD (µA)
80ᵒ
2
3
7
8
.6
4
3
3
9
8
.0
6
0
.93
1
.33
0
0
.00
0
0
.01
0
0
.94
0
1
.34
75ᵒ
2
2
6
1
.2
7
3
2
3
0
.3
8
0
.89
1
.28
0
0
.00
0
0
.01
0
0
.90
0
1
.29
65ᵒ
2
0
3
5
.8
1
2
9
0
8
.3
0
0
.82
1
.18
2
3
.37
3
3
.39
2
4
.20
3
4
.57
55ᵒ
1
8
2
3
.3
5
2
6
0
4
.7
8
0
.76
1
.08
2
1
.25
3
0
.36
2
2
.02
3
1
.45
45ᵒ
1
6
2
4
.5
4
2
3
2
0
.7
7
0
.69
0
.99
1
9
.96
2
8
.52
2
0
.66
2
9
.52
35ᵒ
1
4
3
9
.9
7
2
0
5
7
.0
9
0
.63
0
.91
1
9
.06
2
7
.23
1
9
.70
2
8
.14
25ᵒ
1
2
7
0
.1
3
1
8
1
4
.4
7
0
.58
0
.83
1
8
.36
2
6
.23
1
8
.94
2
7
.06
15ᵒ
1
1
1
5
.4
9
1
5
9
3
.5
5
0
.52
0
.75
1
7
.74
2
5
.34
1
8
.27
2
6
.10
05ᵒ
0
9
7
6
.4
5
1
3
9
4
.9
4
0
.47
0
.68
1
7
.05
2
4
.37
1
7
.53
2
5
.05
00ᵒ
0
9
1
2
.9
2
1
3
0
4
.1
8
0
.45
0
.64
1
6
.69
2
3
.85
1
7
.15
2
4
.50
-
05ᵒ
0
8
5
3
.4
4
1
2
1
9
.2
1
0
.43
0
.61
1
6
.34
2
3
.34
1
6
.77
2
3
.96
-
15ᵒ
0
7
4
6
.8
6
0
1
6
6
.9
5
0
.38
0
.55
2
2
.19
3
1
.71
2
2
.58
3
2
.26
-
25ᵒ
0
6
5
7
.1
7
0
9
3
8
.8
2
0
.34
0
.49
1
9
.33
2
7
.62
1
9
.68
2
8
.12
-
35ᵒ
0
5
8
4
.9
8
0
8
3
5
.6
9
0
.30
0
.44
1
8
.03
2
5
.76
1
8
.34
2
6
.20
Figure
8
giv
e
s
the
detai
ls
reg
a
rd
i
ng
to
t
he
powe
r
co
nsum
ption
a
nd
curre
nt
dra
w
n
f
or
th
e
reg
e
ner
at
i
ve
c
om
par
at
or
desi
gn
e
d
us
in
g
po
wer
gatin
g
te
c
hnology.
Fig
ure
8(
a
)
grp
hical
y
sh
ows
t
he
va
riat
ion
s
wh
e
n
c
hannel
width
is
var
ie
d
from
1
µ
m
to
5
µm
by
keep
ing
t
he
s
upply
vo
lt
age
c
onsta
nt
(
0.7
V
).
T
he
powe
r
consum
ption
i
s
m
axi
m
u
m
at
5
µm
that
is
62
.
20
µ
W
at
this
ci
rcu
it
al
so
draw
m
axi
m
u
m
current
of
88.
86
µ
A
from
su
pp
ly
volt
age.
Fig
ur
e
8(b)
sho
ws
va
riat
ion
s
i
n
po
wer
dissipati
on
an
d
c
urre
nt
dr
a
w
n
by
var
y
ing
t
he
su
pply
volt
age
from
1.
7
V
t
o
0.7
V
at
ch
ann
el
width
of
1
µm
.
Du
e
to
this
va
riat
ion
m
ini
m
u
m
powe
r
consum
ption
is
at 0.7
V
t
hat is
21.66 µ
W
an
d m
axi
m
u
m
at 1.
5 V t
hat is
192.7
5
µ
W.
(a)
(b)
Figure
8. (a
)
P
ow
e
r
c
onsu
m
ption
&
c
urren
t
draw
n vs
c
ha
nnel
w
idth
;
(
b)
P
ow
e
r
c
onsu
m
ption
& c
urren
t
draw
n
vs
s
upply
vo
lt
a
ge
Table
4
gi
ves
the
detai
ls
re
ga
rd
i
ng
t
o
the
va
riat
ion
s
i
n
po
w
er
co
nsum
ption
an
d
c
urre
nt
draw
n
by
the
reg
e
ner
at
i
ve
c
om
par
at
or
des
ign
e
d
us
in
g
powe
r
gatin
g
te
chnolo
gy
to
the
var
yi
ng
te
m
per
at
ur
e
in
diff
e
re
nt
m
od
es
i.e.
le
ak
age,
sta
ti
c,
dynam
ic
,
total
.
As
the
te
m
per
at
ure
increase
s
le
a
kag
e
po
wer
c
onsu
m
ption
incr
eases
with
a
pproxim
at
el
y
30
0%
pe
rcen
t
var
ia
ti
on
m
axi
m
u
m
val
ue
of
33.
53
nW
at
80ᵒ
C
.
O
n
the
ot
her
ha
nd,
sta
ti
c
powe
r
c
on
s
umpti
on inc
reases
w
it
h
i
ncr
ease
i
n
te
m
per
at
ur
e
hav
i
ng m
axi
m
um
v
al
ue
of
46
.53 n
W
at
80ᵒ
C.
1
3
5
0
2
4
6
20
40
60
80
100
Cha
nn
e
l
w
id
th
(µm
)
Pow
e
r c
onsumpt
ion (nW
)
Cu
rre
nt
dra
w
n (n
A
)
0.
6
0.
8
1.
0
1.
2
1.
4
1.
6
0
20
40
60
80
100
120
140
160
180
200
220
Sup
pl
y
v
ol
tage
(v
ol
ts)
Pow
e
r c
onsumpt
ion (nW
)
Cu
rre
nt
dra
w
n (n
A
)
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N: 20
88
-
8708
An
alysis of C
MOS Co
m
pa
r
at
or
in
90
nm Te
chnolo
gy
wi
th
Diff
erent P
ower Reduct
io
n
.
..
(
Anil K
ha
t
ak)
4929
Table
4.
T
otal
, st
at
ic
, d
ynam
ic
& leaka
ge
P
ower
cons
um
pti
on of
reg
e
ne
rati
ve
com
par
at
or
(gati
ng
Tech
no
l
og
y)
at
d
if
fer
e
nt tem
per
at
ur
e
Te
m
p
Leakag
e
Static
Dy
n
a
m
i
c
Total
PC (nW
)
CD (nA)
PC (nW
)
CD (nA)
PC (nW
)
CD (nA)
PC (nW
)
CD (nA)
80ᵒ
3
3
.53
4
7
.91
4
6
.53
6
6
.47
0
.00
7
6
0
0
.01
0
8
6
4
6
.52
6
6
.46
75ᵒ
3
0
.36
4
3
.37
4
3
.74
6
2
.48
0
.00
5
1
1
0
.00
7
3
0
4
3
.73
6
2
.48
65ᵒ
2
4
.66
3
5
.24
3
8
.48
5
4
.97
0
.00
2
3
8
0
.00
3
3
9
3
8
.48
5
4
.97
55ᵒ
1
9
.80
2
8
.29
3
3
.66
4
8
.09
0
.00
1
1
8
0
.00
1
6
7
3
3
.66
4
8
.08
45ᵒ
1
5
.69
2
2
.41
2
9
.26
4
1
.81
0
.00
0
6
2
0
.00
0
8
7
2
9
.26
4
1
.80
35ᵒ
1
2
.25
1
7
.51
2
5
.26
3
6
.09
0
.00
0
3
4
0
.00
0
4
8
2
5
.26
3
6
.09
25ᵒ
0
9
.43
1
3
.47
2
1
.66
3
0
.95
0
.00
0
1
9
0
.00
0
2
6
2
1
.66
3
0
.95
15ᵒ
0
7
.13
1
0
.19
1
8
.46
2
6
.38
0
.00
0
0
9
0
.00
0
1
3
1
8
.46
2
6
.38
05ᵒ
0
5
.30
0
7
.58
1
5
.68
2
2
.41
0
.00
0
0
3
0
.00
0
0
3
1
5
.68
2
2
.41
00ᵒ
0
4
.54
0
6
.49
1
4
.45
2
0
.65
0
.00
0
0
0
0
.00
0
0
0
1
4
.45
2
0
.65
-
05ᵒ
0
3
.58
0
5
.11
1
3
.32
1
9
.03
0
.00
0
0
3
0
.00
0
0
3
1
3
.32
1
9
.03
-
15ᵒ
0
2
.58
0
3
.68
1
1
.37
1
6
.25
0
.00
0
0
7
0
.00
0
0
9
1
1
.37
1
6
.25
-
25ᵒ
0
1
.83
0
2
.62
0
9
.82
1
4
.03
0
.00
0
1
1
0
.00
0
1
5
0
9
.82
1
4
.03
-
35ᵒ
0
1
.41
0
2
.01
0
8
.92
1
2
.95
0
.28
3
7
6
0
.61
1
6
4
0
8
.64
1
2
.34
Dynam
ic
po
w
er
co
nsum
ption
s
hows
ve
ry
sm
a
ll
var
ia
ti
on
s
of
few
nW.
T
hus,
t
he
total
powe
r
consum
ption
a
lso
at
ta
ins
m
arg
inal
value
as
com
par
ed
to
e
arli
er
discuss
ci
rcu
it
s.
Total
powe
r
consum
ption
at
ta
ins its m
axi
m
u
m
v
al
ue
of
46.52 n
W
at
80ᵒ C a
nd m
ini
m
u
m
v
al
ue
of
8.64 nW at
-
35° C.
Table
5
s
how
s
the
bri
ef
c
omparis
on
of
the
resu
lt
s
f
or
al
l
f
our
ty
pes
of
sc
hem
atics
discu
ssed
a
bove.
The
re
gen
e
rati
ve
com
par
at
or
us
in
g
powe
r
ga
ti
ng
te
chn
i
que
sh
ows
m
ini
m
um
po
wer
co
nsum
ption
of
21.66
61
4
nW as c
om
par
ed
to
o
t
her
s
wh
il
e d
ra
wing a c
urren
t
of
30.
95
163 n
A
f
r
om
0
.7
V
s
upply
vo
lt
age.
Table
5.
C
om
par
iso
n of res
ults
Reg
en
erative
co
m
p
a
rator
Reg
en
erative
co
m
p
a
rator
with
p
seu
d
o
NM
OS
Reg
en
erative
co
m
p
a
rator
with
CVSL
Reg
en
erative
co
m
p
a
rator
with
p
o
wer
g
atin
g
Te
m
p
e
rature
2
5
ᵒC
2
5
ᵒC
2
5
ᵒC
2
5
ᵒC
No
.
Of
MO
SFE
TS
27
25
31
31
Ch
an
n
el L
en
g
th
9
0
n
m
9
0
n
m
9
0
n
m
9
0
n
m
Ch
an
n
el W
id
th
1
µ
m
1
µ
m
1
µ
m
1
µ
m
Su
p
p
ly
vo
ltag
e
0
.7 V
0
.7 V
0
.7 V
0
.7 V
Total Po
wer
con
su
m
p
tio
n
1
5
.02
6
9
6
µW
1
2
6
.5312
8
µW
1
8
.94
7
4
5
µW
2
1
.66
6
1
4
n
W
Total Cu
rr
en
t dr
aw
n
2
1
.46
7
0
8
µA
1
8
0
.7589
7
µA
2
7
.06
7
7
9
µA
3
0
.95
1
6
3
n
A
Table
6
s
how
s
the
va
riat
ion
s
in
the
po
wer
c
onsu
m
ption
s
for
al
l
f
our
sc
hem
atic
s
wh
e
n
th
e
tem
per
at
ur
e
is
var
ie
d
f
r
om
-
35
ᵒ to
+80ᵒ Cel
s
ius.
Table
6.
Perce
ntage va
riat
ion Po
wer c
on
s
um
pt
ion
(P
C)
wi
th te
m
per
at
ure
Reg
en
erative
co
m
p
a
rator
Reg
en
erative co
m
p
arator
with
ps
eu
d
o
NM
O
S
Reg
en
erative co
m
p
arator
with
CVSL
Reg
en
erative co
m
p
arator
with
po
wer
g
atin
g
Te
m
p
e
rature
-
3
5
ᵒ
to
+80
ᵒ
-
3
5
ᵒ
to
+80
ᵒ
-
3
5
ᵒ
to
+80
ᵒ
-
3
5
ᵒ
to
+80
ᵒ
Leakag
e PC
8
3
.39
to
237.2
7
5
2
.63
to 8
9
.9
3
4
5
9
5
7
.2 to
87
.27
8
5
.02
to
255.6
5
Static PC
5
.37
to 2
.51
1
8
.40
to 1
2
.5
5
4
6
.79
to 6
0
.5
7
5
8
.80
to 1
1
4
.76
Dy
n
a
m
i
c PC
7
0
.19
to 7
4
.1
6
1
3
2
.75
to 8
4
5
.24
1
.79
to 9
9
.95
1
4
9
2
4
7
to
3
9
0
0
Total PC
1
0
.10
to 3
.08
2
2
.87
to 2
2
1
.69
3
.17
to
9
5
.02
6
0
.11
to
114.7
2
4.
COMP
AR
I
S
ON
In
orde
r
to
c
om
par
e
si
m
ulatio
n
res
ults
f
our
m
or
e
com
par
at
or
st
ru
ct
ur
es
ar
e
again
sim
ulated
in
90nm
CM
OS
te
ch
no
l
og
y.
T
hese
fou
r
struct
ur
es
a
re
dynam
ic
co
m
par
at
or
(c
onve
ntion
al
)
[
24
]
,
dynam
ic
co
m
par
at
or
(do
ub
le
ta
il
)
[2
4],
dy
nam
ic
com
par
at
or
(
m
od
ifie
d
doub
le
ta
i
l)
[2
4],
c
om
par
at
or
wit
h
two
cr
os
s
-
c
oupled
inv
e
rters
[
25
]
.
Re
su
lt
s
that
a
re
obta
ined
a
fter
si
m
ulati
on
al
ong
with
f
ou
r
recent
c
om
par
at
or
str
uct
ures
[26]
[27
]
-
[
29]
w
hich
are
al
s
o
incl
ud
e
d
for
c
om
par
iso
n
are
sho
wn
i
n
T
a
ble
7.
Re
ge
ner
at
ive
com
par
at
or
(
Power
gating) s
h
ows
op
ti
m
u
m
r
esults i
n
c
om
par
iso
n.
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
8
, N
o.
6
,
Dece
m
ber
2
01
8
:
4922
-
4931
4930
Table
7.
C
om
par
iso
n of res
ults at
25ᵒ Cel
sius
No
.
o
f
MOSF
ET
S
Ch
an
n
el
Leng
th
Ch
an
n
el
W
id
th
Su
p
p
ly
v
o
ltag
e
Total Po
wer
co
n
su
m
p
tio
n
Dy
n
a
m
i
c Co
m
p
a
ra
to
r
(Co
n
v
en
tio
n
al)
[
2
4
]
09
9
0
n
m
1
µ
m
0
.7 V
0
2
.77
nW
Dy
n
a
m
i
c Co
m
p
a
ra
to
r
(Dou
b
le T
ail
)
[
2
4
]
14
9
0
n
m
1
µ
m
0
.7 V
1
0
.60
nW
Dy
n
a
m
i
c Co
m
p
a
ra
to
r
(M
o
d
if
ied
Dou
b
le T
ail)
[
2
4
]
18
9
0
n
m
1
µ
m
0
.7 V
1
4
.19
nW
Co
m
p
a
rator wi
th
t
wo
cr
o
ss
-
co
u
p
led
inv
erter
s [
2
5
]
15
9
0
n
m
1
µ
m
0
.7 V
1
3
.84
µW
[
2
6
]
14
9
0
n
m
-
1
.0 V
8
2
.00
µW
[
2
7
]
-
40
nm
-
0
.6
V
0
1
.50
µW
[
2
8
]
16
180
nm
-
1
.6 V
1
7
.00
µW
[
2
9
]
06
65
nm
-
1
.2 V
7
5
5
.00
µW
Reg
en
erative co
m
p
arator
[
1
3
]
27
9
0
n
m
1
µ
m
0
.7 V
1
5
.02
µW
Reg
en
erative co
m
p
arator
with
po
wer
gatin
g
31
9
0
n
m
1
µ
m
0
.7 V
2
1
.66
nW
So
,
by
com
pari
ng
on
t
he
basis
of
powe
r
c
onsu
m
ption
with
it
s
var
ia
ti
on
wi
th
cha
nn
el
widt
h,
s
upply
vo
lt
age
&
te
m
per
at
ur
es
it
is
cl
early
ob
se
rved
that
r
ege
ne
rati
ve
com
par
a
tor
with
po
we
r
gatin
g
s
how
good
perform
ance wi
th r
es
pect to
ot
her
s.
5.
CONCL
US
I
O
N
Thr
ee
di
ff
e
rent
power
re
duct
ion
te
c
hn
i
qu
e
s
are
im
ple
m
ented
on
reg
e
ne
r
at
ive
com
par
at
or
ci
rcu
it
s.
Ou
t
of
the
se
th
ree
te
ch
niques
powe
r
gatin
g
t
echn
i
qu
e
sho
w
s
substanti
al
de
crease
in
total
power
c
ons
um
pt
ion
al
ong
with
the
reducti
on
i
n
le
akag
e
c
urre
nt.
Total
po
we
r
consum
ption
of
15.
026
µ
W
r
edu
ce
d
to
21.66
6
nW
for
re
ge
ner
at
iv
e
com
par
at
or
with
po
wer
ga
ti
ng
te
ch
nique
wh
ic
h
certai
nl
y
i
m
pr
ov
es
it
s
perf
or
m
ance.
Four
conve
ntion
al
&
f
our
rece
nt
c
om
par
at
or
str
uc
tures
a
re
com
par
e
d
with
re
ge
ner
at
ive
c
omparat
or
(
po
wer
gatin
g).
Hen
ce
it
is
observ
e
d
that
the
reg
e
ner
at
ive
c
om
par
at
or
wit
h
powe
r
gatin
g
te
chn
i
que
shows
optim
u
m
powe
r
consum
ption
with
hi
gh
sp
ee
d
of
op
e
rati
on
du
e
t
o
re
ge
nerat
ive
la
tc
h
at
it
s
en
d.
T
his
de
f
init
el
y
will
inc
rease
it
s
app
li
cabil
ti
es in
high s
pee
d
a
nd lo
w
powe
r UD
SM data
conv
e
rter ci
rcu
it
s
d
esi
gn
s
.
REFERE
NCE
S
[1]
B.
Ra
za
v
i, “P
rinc
iples of
Da
ta C
onver
sion S
y
st
e
m
Design
,
”
AT
&
T B
ell
laboratori
es
,
IE
EE Press, 1995.
[2]
P.
E. Allen and
D.
R.
Holber
g
,
“
CMO
S Anal
og
Circ
uit Design,
”
Oxford Unive
rs
i
t
y
Press
,
Se
cond
Edition, 2002
.
[3]
C.
Fa
y
om
i,
e
t
al
.
,
“
Low
power/
l
ow
volt
age
high
-
spee
d
CMO
S
diffe
ren
ti
a
l
tr
ac
k
and
latch
compara
tor
with
rail
-
to
-
rai
l
input,”
Proc
.
ISCAS5
,
pp
.
653
–
656
,
2000
.
[4]
A.
Al
,
et
al.
,
“
An
Im
prove
d
A
Low
Pow
er
C
MO
S
TIQ
Compa
rat
or
Flash
A
DC,
”
TEL
KOMNIKA
Indone
sia
n
Journal
of
Elec
t
rical
Engi
ne
erin
g,
v
ol
/i
ss
ue:
12
(
7
)
,
pp
.
5204
-
521
0,
2014
.
[5]
M
.
Marufuz
za
m
an,
et
a
l.
,
“
Desig
n
of
3
-
Bit
AD
C
i
n
0.
18μm
CMOS Proc
ess,”
TEL
KOMNIKA
Indone
sian J
ournal
o
f
El
e
ct
rica
l
Eng
in
ee
ring,
v
ol
.
12
,
pp.
5197
-
5203
,
2014.
[6]
M
.
Muham
ad
,
e
t
al
.
,
“
Design
o
f
Low
Pow
er
L
ow
Noise
Am
pli
fie
r
using
Gm
-
boosted
T
ec
hniq
ue,
”
Indone
sian
Journal
of
Elec
t
rical
Engi
ne
erin
g
and
Computer
Sci
en
ce
(
IJE
ECS
)
,
v
ol.
9
,
pp
.
685
-
689,
2018
.
[7]
B.
Goll
and
H.
Zi
m
m
ermann,
“
Low
power
600
MH
z
compara
to
r
for
0.
5
V
suppl
y
vo
lt
ag
e
in
0.
12
µm
CM
OS
,
”
IE
EE
El
e
ct
ron.
Lett.
V
ol/
issue
:
43
(
7
)
,
p
p.
388
-
390
,
200
7.
[8]
S
.
Go
y
al
and
V
.
Sulocha
n
a,
“
De
sign
of
Low
Leaka
ge
Multi
Thr
eshold
(V
th
)
CM
OS
Le
vel
Shifter,”
In
te
rnationa
l
Journal
of
Elec
t
rical
and
Computer
Eng
ine
ering
(
IJE
CE)
,
v
ol. 3,
pp.
584
-
592
,
20
13.
[9]
N
.
Mukaha
r
and
S
.
H
.
Ruslan,
“
A
93.
36
dB,
161
MH
z
CMO
S
O
per
ational
Tra
ns
conduc
t
anc
e
Am
pli
fier
(OTA)
for
a
16
Bit
Pip
el
i
ne
Analog
-
to
-
D
igi
tal
Conver
te
r
(AD
C),
”
Inte
r
nati
onal
Journal
of
E
le
c
tric
al
and
Computer
Engi
ne
ering
(
IJ
ECE
)
,
v
o
l
.
2
,
pp
.
106
-
111,
2012.
[10]
S.
K.
Patna
ik
a
nd
S.
Bane
rj
ee,
“
Noise
and
Err
or
Anal
y
sis
and
Optimiza
ti
on
o
f
a
CMO
S
La
tched
Com
par
at
or
,
Inte
rna
ti
ona
l
Co
nfe
ren
c
e
on
Co
m
m
unic
at
ion
T
e
chnol
og
y
and
S
y
stem
Design
2
011,
”
Proce
d
ia
Engi
ne
ering
,
vo
l.
30
,
pp
.
210
–
217
,
2012.
[11]
G.
M.
Yin,
et
a
l.
,
“
A
high
-
spee
d
CMO
S
compara
tor
with
8
-
bi
t
re
soluti
on,
”
IE
EE
J.
Sol
id
Sta
te
C
i
rcuit
s
,
vol
/i
ss
ue:
27(2),
pp
.
208
–
2
11,
1992
.
[12]
J
.
N
.
Redd
y
,
e
t
al.
,
“
Subthreshold
Dual
Mode
Logi
c,”
Bul
l
et
in
o
f
El
ectric
a
l
Enginee
ring
and
Informatic
s
(
BE
EI)
,
v
ol
.
3
,
pp
.
141
-
1
48,
2014
.
[13]
M.
Saber
i
and
R
.
Lot
fi
,
“
Segme
nte
d
Archi
te
c
tu
re
for
Succe
ss
ive
Approxim
at
ion
Analog
-
to
-
Digit
al
Conver
te
rs,
”
IEE
E
Tr
ansacti
o
ns on
Ve
ry
Lar
ge
-
Scale
In
te
grati
on
(
VLSI)
Syste
ms
,
vol
/i
ss
ue:
22
(
3
)
,
2014
.
[14]
H
.
Jeong,
et
a
l.
,
“
Ps
eudo
N
MOS
Based
Sense
Am
pli
fie
r
for
Hi
gh
Speed
Single
-
Ende
d
SR
AM
,
”
2014
21st
IEEE
Inte
rnational
Co
nfe
renc
e
on
Elec
tronic
s,
C
ircui
ts
and
Syste
ms
(
ICECS)
,
Mars
ei
lle,
pp.
331
-
334,
20
14.
[15]
K
.
M.
Chu
and
D
.
L.
Pulfre
y
,
“
A
Com
par
ison
o
f
CMO
S
Circ
uit
Te
chni
qu
es:
Diffe
ren
t
ia
l
C
asc
o
de
Volta
ge
Sw
itch
Logi
c
Versus
Co
nvent
ion
al
Logi
c
,
”
I
EEE
Journal of
Soli
d
-
State
Ci
rcuit
s
,
vo
l/
issue:
SC
-
22
(
4
)
,
1987
.
[16]
K
.
Lee
and
S.
S
.
W
ong,
“
Fault
-
T
ole
ran
t
FP
GA
w
it
h
Colum
n
-
Based
Redunda
nc
y
a
nd
Pow
er
Gati
ng
Us
ing
RR
AM
,
”
IEE
E
Tr
ansacti
o
ns on
Computers
,
vol
/i
ss
ue:
66
(
6
)
,
2017
.
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N: 20
88
-
8708
An
alysis of C
MOS Co
m
pa
r
at
or
in
90
nm Te
chnolo
gy
wi
th
Diff
erent P
ower Reduct
io
n
.
..
(
Anil K
ha
t
ak)
4931
[17]
B
.
H.
Nagpa
ra,
“
A
45
nm
6
Bit
Low
Pow
er
Curre
nt
Ste
eri
ng
D
igi
tal
to
Ana
log
Convert
er
Us
ing
GD
I
Logi
c,”
TEL
KOMNIKA
Indone
sian J
ourn
al
of
Elec
tric
al
Engi
ne
ering,
v
ol
/i
ss
ue:
16
(
1
)
,
pp
.
46
-
51
,
2015.
[18]
P.
Zha
o
,
et
a
l
.
,
“
Low
-
Pow
er
Cloc
ke
d
-
Ps
eudo
-
NM
OS
Flip
-
Flop
for
Le
v
el
Con
v
e
rsion
in
Dua
l
S
uppl
y
S
y
st
ems
,
”
IEE
E
Tr
ansacti
o
ns on
Ve
ry
Lar
ge
-
Scale
In
te
grati
on
(
VLSI)
Syste
ms
,
vol
/i
ss
ue:
17
(
9
)
,
pp
.
1196
-
12
02,
2009
.
[19]
R.
F
.
Mir
zaee,
et
al
.
,
“
Diffe
ren
ti
al
Cascod
e
V
olt
ag
e
Sw
it
ch
(
DCV
S)
Strat
egies
b
y
CNTFET
Technol
og
y
fo
r
Standa
rd Tern
ar
y
Logic,
”
Mi
cro
el
e
ct
ronics
Jour
nal
,
2013
.
h
tt
p:
//dx.doi.
org
/10.
10
16/j
.
m
ej
o
.
2013.
0
8.
010i
[20]
M.
C.
C
ase
y
,
et
al.
,
“
HBD
Us
in
g
Cascode
-
Vol
tage
Sw
it
ch
Logi
c
Gate
s
for
SE
T
Tol
er
ant
Dig
it
a
l
De
signs,”
I
EEE
Tr
ansacti
ons on Nucle
ar S
ci
en
ce
,
vol
/i
ss
ue:
52
(
6
)
,
2005.
[21]
M.
C.
Case
y
,
et
al.
,
“
Single
-
Event
Tol
e
ran
t
La
t
c
h
Us
ing
Cascode
-
Volta
ge
Sw
itch
Logi
c
Ga
te
s,
”
IE
EE
Tr
ansacti
ons
on
Nuclear Scie
nce
,
v
ol
/i
ss
ue:
53
(
6
)
,
2006
.
[22]
J.
J
.
Johanna
ha
,
et
al
.
,
“
Stand
b
y
and
d
y
n
amic
po
wer
m
ini
m
iz
at
io
n
using e
nhance
d
h
y
brid
powe
r
gat
ing
stru
ct
ure
for
dee
p
-
subm
ic
ron C
MO
S VLSI,”
Mic
roel
ec
troni
c
s J
ournal
,
2017
.
htt
p://dx.doi.org/10.1016/j.m
ej
o.
2
017.
02.
003
[23]
A
.
R
.
Tri
v
edi
,
e
t
al
.
,
“
In
Situ
Pow
er
Gati
ng
Eff
ic
i
ency
Learne
r
for
Fine
-
Grain
e
d
Self
-
Adapt
ive
Pow
er
Gati
ng
,
”
IEE
E
Tr
ansacti
o
ns on
Circuits a
nd
Syste
ms
,
vol/i
ss
ue:
61
(
5
)
,
201
4.
[24]
S
.
B
.
Mashhadi
and
R
.
Lot
f
i,
“
Anal
y
s
is
and
Desi
gn
of
a
Low
-
Volta
ge
Low
-
Pow
er
Double
-
Tail
Co
m
par
at
or,
”
IEEE
Tr
ansacti
ons onV
ery
Lar
ge Scale
Int
egrati
on
(
VLSI)
Syste
ms
,
vol
/i
ss
ue:
22
(
2
)
,
20
14.
[25]
J
.
I
.
L
ee
and
J
.
I
.
Song,
“
Flash
AD
C
Archi
te
ct
u
re
using
Multi
p
l
exe
rs
to
Redu
ce
a
Prea
m
pli
f
ie
r
and
Com
par
at
o
r
Count,
”
IE
EE
,
2
013.
[26]
A
.
Rez
apour
,
et
al
.
,
“
Low
Powe
r
High
Speed
D
y
namic
Com
par
at
or
,”
2018
I
EE
E
Int
ernati
on
al
Symposium
on
Circui
ts and
Sys
te
ms
(
ISCAS)
,
Fl
ore
nce,
I
ta
l
y
,
pp
.
1
-
5,
2018
.
[27]
O.
Aiello,
et
al.
,
“
Full
y
S
y
nth
esi
za
bl
e,
R
ai
l
-
to
-
R
ai
l
D
y
n
amic
Vo
lt
ag
e
Com
par
at
o
r
for
Opera
ti
on
down
to
0.
3
V,
”
2018
IEEE
In
te
r
nat
ional
S
ymposium on
Circu
it
s
and
Syste
ms
(
ISCAS)
,
Florence,
Ita
l
y
,
pp
.
1
-
5
,
20
18
.
[28]
A.
Khoram
i
and
M.
Sharifkhani,
“
A
Low
-
Po
wer
High
-
Spee
d
Com
pa
rat
or
for
Prec
ise
Applicati
ons,
”
IEEE
Tr
ansacti
ons on Very
Lar
ge
-
Sca
l
e
Int
egrati
on
(
VLSI)
Syste
ms
,
pp.
1
–
12
,
2018
.
[29]
M.
Nasrolla
hpo
ur,
et
a
l.
,
“
A
high
-
spee
d,
low
-
o
ffset
and
low
-
p
ower
diffe
r
ent
i
a
l
compara
tor
for
ana
log
to
digit
al
conve
rt
ers,
”
201
7
Inte
rnat
ional
S
oC
Design
Conf
ere
nce (
ISOCC)
,
Seoul, Korea (S
outh),
pp
.
220
-
2
21
,
2017
.
Evaluation Warning : The document was created with Spire.PDF for Python.