Indonesian Journal of
Electrical
Engineer
ing and
Computer Science
V
o
l. 10
, No
. 3, Jun
e
20
18
, pp
. 10
13
~
1
022
ISSN: 2502-4752,
DOI: 10.115
91/ijeecs
.v10.i
3.pp1013-1022
1
013
Jo
urn
a
l
h
o
me
pa
ge
: http://iaescore.c
om/jo
urnals/index.php/ijeecs
The Analysis of Soft
Error in C-elements
Nor
huz
aimin Julai,
Ahmed
M. A.
H
a
idar
,
Ab
dul Rahm
an Kram
Departem
ent
of
Ele
c
tri
cal
and
E
l
ectron
i
cs
Eng
i
ne
ering,
F
acu
lt
y of
Engin
eering
,
U
n
ivers
iti
M
a
l
a
y
s
i
a
S
a
raw
a
k,
M
a
la
ys
ia
Article Info
A
B
STRAC
T
Article histo
r
y:
Received Ja
n
2, 2018
Rev
i
sed
Mar
12
, 20
18
Accepted
Mar 28, 2018
S
o
ft errors
ar
e
a
s
e
rious
con
cern
in s
t
ate
holde
rs
as
i
t
c
a
n
caus
e
tem
porari
l
y
m
a
lfunction
of
t
h
e c
i
rcui
t.
C-e
l
e
m
ent is
one of
the state holders
that is used
widel
y
in
the as
y
n
chronous
cir
c
uit. In
this p
a
per
,
th
e inv
e
stigatio
n will fo
cus
on the vuln
e
ra
bilit
y of
two t
y
p
e
s of C-el
e
m
ent towards soft errors.
A framework has been proposed for the rate of er
ror due to neutro
n spectrum
energ
y
th
at
can
caus
e
f
a
ilur
e
in
the s
t
a
t
e ho
lder
.
Effec
tiv
e an
al
ys
is
has
been
conducted on two differen
t
C-elements
at differ
e
nt nodes b
y
using UMC90
nm
technolog
y
and 180nm
tech
nolog
y
.
Bas
e
d
on the vu
lnerab
ilit
y
dat
a
,
a
m
e
thod for assessing vulnerab
ilit
y
on
a diff
erent im
plem
ent
a
tion of C-
elements has been develop
e
d. From th
e obtained data, it can b
e
conclud
e
d
that SIL is
more
resistant towards
soft er
rors.
K
eyw
ords
:
Soft
Er
ro
r
C-Elem
ents
Asy
n
c
h
r
o
no
us ci
rcui
t
S
i
n
g
l
e
ev
en
t
up
s
e
t
Lo
w po
wer
Copyright ©
201
8 Institut
e
o
f
Ad
vanced
Engin
eer
ing and S
c
i
e
nce.
All rights re
se
rve
d
.
Co
rresp
ond
i
ng
Autho
r
:
Norhu
zaim
i
n
Ju
lai,
Depa
rt
m
e
nt
of
El
ect
ri
cal
and
El
ect
roni
cs
E
n
gi
nee
r
i
n
g,
Fac
u
l
t
y
of
En
gi
n
e
eri
n
g,
Un
i
v
ersiti Malaysia Sarawak
,
9
430
0 Ko
ta Samar
a
h
a
n, Sar
a
w
a
k, Malaysia.
E-m
a
i
l: j
n
o
r
h
u
za@un
im
as.
m
y
1.
INTRODUCTION
Asy
n
c
h
r
o
no
us
ci
rcui
t
s
ope
r
a
t
e
wi
t
h
o
u
t
a
cl
ock
an
d t
h
ere are
n
u
m
e
ro
us a
d
vant
a
g
es o
f
usi
n
g
asyn
chr
ono
us cir
c
u
its su
ch
as n
o
cl
o
c
k
sk
ew
pr
ob
lem
and no
global tim
i
ng iss
u
es.
Asy
n
chronous circ
uits are
al
so l
e
ss a
ffect
ed
by
t
h
e
t
ech
nol
ogy
a
n
d
pr
o
cess [
1
]
.
T
h
e
r
e
is also power
issue in
syn
c
hro
nou
s
d
e
sign sin
ce it
u
tilizes clo
c
k
s
to
m
a
k
e
an
y tran
sitio
n
at th
e lo
g
i
c. On
th
e o
t
h
e
r h
a
n
d
, th
e p
o
wer in
asynch
r
on
ou
s d
e
sign
will
be l
e
ss com
p
ar
ed wi
t
h
sy
nch
r
on
o
u
s de
si
g
n
.
Ho
we
ver
,
o
n
e
of t
h
e
di
sa
dva
nt
ages
of asy
n
chr
o
no
us ci
rc
u
i
t
i
s
t
h
e
circu
it failu
re
d
u
e
to
d
ead
l
o
ck
: A state wh
ere th
e syste
m
will b
e
d
i
sab
l
ed
in
d
e
fin
itely u
n
til th
e system h
a
s
b
een reset
or t
h
e erro
r is filtered
or co
rrected
fro
m
th
e syste
m
. Th
at m
e
a
n
s th
e circu
it will b
e
in th
e
waiting
st
at
e unl
ess t
h
ere i
s
a
feed
ba
ck
or s
o
m
e
ki
nd
o
f
ac
kn
o
w
l
e
dgem
e
nt
si
g
n
a
l
si
nce i
t
i
s
d
e
pen
d
e
d
on t
h
e dat
a
itself rath
er than
clo
c
k
to
fun
c
tio
n. Sing
le ev
en
t up
se
t (SEU)
h
a
s b
e
en
id
en
tified
as a p
o
ssi
b
l
e reason
th
at
cause
d data c
o
rr
uptio
n.
The t
e
rm
soft erro
r
refe
rs to th
e tem
p
o
r
arily error th
at is du
e to
th
e
p
a
rticle strik
e
pr
o
v
i
d
e
d
t
h
e
s
u
f
f
i
c
i
e
nt
c
u
r
r
e
n
t
an
d
wi
t
h
c
e
r
t
ai
n wi
dt
h
of
c
u
rrent
pulse is
neede
d
t
o
ca
us
e the state c
h
ange.
Th
e PM
OS
or NMOS will
b
e
th
e m
o
st sen
s
itiv
e toward
s SEU
wh
en it is in
th
e
OFF m
o
d
e
, in
part
i
c
ul
a
r
at
t
h
e drai
n
regi
o
n
.
Fi
gu
re 1 s
h
o
w
s t
h
e si
ng
le
ev
en
t tran
sien
t
(SET)
produc
e
d after a
n
energetic
io
n
i
zing
p
a
rticle h
a
s
b
e
en
b
r
ou
gh
t to th
e silico
n
n
e
ar
sen
s
itiv
e
d
e
v
i
ce [2
]. Th
e d
e
n
s
ity
o
f
electro
n-ho
le pairs
i
s
p
r
od
u
c
ed
b
y
particle as sho
w
n
in
Figu
re
1
(
a). Th
e car
riers are co
llected
b
y
electric field
and
will cau
se the
ch
arg
e
co
llectio
n to
ex
p
a
nd
du
e to
drift curren
t (Fi
g
ur
e
1(b)) a
n
d res
u
lt in the s
u
dde
n
current
pulse
. T
h
e
n
, t
h
e
d
i
ffu
s
i
o
n curren
t
will
d
o
m
in
ate u
n
til all th
e
ex
cess carrier
h
a
v
e
been co
ll
ected
,
reco
m
b
in
ed or
d
i
ffused away
fr
om
junct
i
on
area (
F
i
g
ure
1
(
c)).
T
h
e
size
o
f
fu
nn
el as
show
n in
Figu
r
e
1(
b)
an
d co
llectio
n ti
m
e
is v
e
r
y
m
u
ch
i
nve
rsel
y
pr
o
p
o
rt
i
o
n wi
t
h
t
h
e
sub
s
t
r
at
e d
opi
ng
. The c
o
l
l
e
ctio
n
ti
m
e
is u
s
u
a
lly co
m
p
lete
with
in
p
i
co
seco
nd
s
an
d th
e
d
i
ffusio
n
s
cu
rren
t
b
e
g
i
n
t
o
d
o
m
in
ate un
til all th
e ex
cess carriers
h
a
v
e
b
e
en
co
ll
ected
[3
].
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
502
-47
52
I
ndo
n
e
sian
J Elec Eng
& Com
p
Sci, V
o
l. 10
,
No
.
3
,
Jun
e
2
018
:
10
13
–
1
022
1
014
Fi
gu
re 1.
SET
pr
o
duce
d
[2]
2.
CO
MP
ARI
N
G
SEU
WIT
H
DIF
F
ERENT
TECH
N
OLO
G
Y
In orde
r t
o
c
o
m
p
are C-element wit
h
diffe
rent techno
logy agai
nst SE
U, the circ
uits c
h
osen a
r
e
Single
Rail w
ith
In
v
e
r
t
er
Latch
(SIL)
as
show
n in
Fi
g
u
r
e
3
(
a)
an
d th
e cor
r
e
sp
ond
ing
layout in
Figur
e
3
(
b
)
. The
circu
it is
m
o
delled
to
h
a
v
e
th
e sam
e
wid
t
h
of th
e m
a
in
transistors a
n
d the feedbac
k
tra
n
sistors.
For t
h
is
pu
r
pose
,
t
w
o
d
i
ffere
nt
t
y
pes
o
f
C
a
de
nce
Tec
h
n
o
l
o
gy
a
r
e
us
ed i
n
t
h
e
si
m
u
lat
i
on:
UM
C
9
0
n
m
and
18
0
n
m
.
A cu
rre
nt
p
u
l
s
e can be
rep
r
es
ent
e
d as
havi
n
g
fast
ri
si
ng t
i
m
e and sl
o
w
f
a
l
l
i
ng t
i
m
e
. The am
pl
i
t
ude,
ri
si
ng
t
i
m
e
and
fal
l
i
ng t
i
m
e of t
h
e cu
rre
nt
pu
l
s
e depe
n
d
on
fact
or
s s
u
ch a
s
t
h
e t
y
pe
of
pa
rt
i
c
l
e
, t
h
e ene
r
gy
o
f
th
e p
a
rticle and
th
e ang
l
e of
th
e strik
e
. Th
ese facto
r
s can
ad
d
co
m
p
lex
iti
es in
m
o
d
e
lling
curren
t
pu
lse. Th
e
m
o
d
e
l sh
ow
n
i
n
Figur
e 2 is used
as a cur
r
e
nt in
j
ection
to c
o
m
p
are the crit
ical ch
arge
s be
tween t
h
e node
s
and
C-ele
m
en
ts. Th
e m
o
d
e
l in [4
] stated
t
h
e
risin
g
and
fallin
g tim
es o
f
curren
t
p
u
l
se to
b
e
5
0
p
s
and
1
6
4
p
s
respectively.
The curre
nt pulses are injec
t
ed at
th
e
m
a
i
n
tran
sist
o
r
s an
d
th
e
ou
tpu
t
o
f
th
e circu
it as sh
own
in
Fig
u
re
3
.
Th
e
risin
g
and
fallin
g ti
m
e
s o
f
the cu
rren
t pu
lse are fi
x
e
d.
Howev
e
r, i
n
o
r
d
e
r to
ch
an
g
e
the area
u
n
d
e
r th
e curve, th
e am
p
litu
d
e
is
v
a
ried
un
til th
e ou
tpu
t
is flipp
e
d. The si
m
u
latio
n
i
s
do
n
e
u
s
i
n
g
circu
i
t
an
alyser (sp
ect
re). Th
e am
p
lit
u
d
e
o
f
th
e cu
rren
t
p
u
l
se is
increased
u
n
til th
e ou
tpu
t
is flip
p
e
d
at no
d
e
s
(i) (ii)
an
d
no
d
e
s (iii) o
f
Figu
re
3
(
a). Th
e critical ch
arg
e
s
wh
ich
co
rresp
ond
s to th
e am
p
litu
d
e
o
f
t
h
e cu
rren
t
p
u
l
se
th
at cau
ses th
e state to
ch
an
ge ar
e ob
tain
ed
an
d th
e exp
e
r
i
men
t
s ar
e
r
e
p
e
ated
w
ith
d
i
f
f
e
r
e
n
t
tech
no
logy and
wi
t
h
di
ffe
re
nt
t
e
m
p
erat
ur
e.
Fig
u
re
2
.
Model o
f
curren
t
p
u
lse [4
] con
d
ition
Fi
gu
re
3.
(a
) S
c
hem
a
t
i
c
SIL;
(b
) Lay
out
of
S
I
L
P
eak current
Time
tau=50ps
pw=10ps
tau=164ps
Actual SE
U
A
P1
P2
N1
N2
(ii)
(i)
(iii)
Iseu
N3
P3
N4
B
O
ut
P4
W(N1,N2,N3)=1.6u
W(P1,P2,P3)=1.8u
W(P4)=450n
W(N4)=400n
Out’
(b
)
(a)
++
++
++
++
++
++
++
++
+
++
−+
− − − − − − − −
− − − − − − − − −
− − − −
− − − − −
+
−
−
+
+
+
−
−
+
−
+
−
+ − + − + − + − + − +
−
+ − + − + − + − + − +
−
+ − + − + − + − + − +
−
+ −
+ −
(
a
)
(
b
)
(
c
)
Evaluation Warning : The document was created with Spire.PDF for Python.
In
d
onesi
a
n
J
E
l
ec En
g &
C
o
m
p
Sci
ISS
N
:
2
5
0
2
-
47
52
Th
e
Ana
l
ysis
of S
o
ft Erro
r in
C-Elemen
ts (No
r
hu
za
imi
n
Ju
la
i)
1
015
Th
e
work in
itially in
v
e
stig
ated
th
e effect
of soft
errors
o
n
two
d
i
fferen
t
C-ele
m
en
ts b
y
v
a
rying
the
wi
dt
h
o
f
t
h
e i
n
ject
e
d
c
u
r
r
ent
pul
se i
n
t
o
var
i
ous l
o
cat
i
o
n
s
on t
h
e ci
rc
ui
t
.
Aut
h
o
r
s i
n
[
5
]
hav
e
de
fi
ne
d
val
i
d
out
put
s
u
ch t
h
at
Out
(
t
)
≤
0.
2. T
h
e res
p
ons
es of t
h
e st
at
e hol
der
s
are chara
c
terized into three pos
sible
cat
ego
r
i
e
s as s
h
o
w
n i
n
Fi
g
u
r
e
4
(a,
b
)
, a
n
d e
xpl
ai
ne
d i
n
t
h
e
f
o
l
l
o
wi
ng:
a)
No ch
ang
e
to th
e state ho
l
d
er
–
Th
ere is no
sig
n
i
fica
n
t
ou
tpu
t
pu
lse that h
a
s b
e
en
g
e
n
e
rat
e
d
an
d will not
cause a
n
y state cha
nge
. It is a
ssum
e
d that if
the ge
ne
rated
p
u
l
se is less t
h
an
0.2 of th
e i
n
pu
t pu
lse su
ch
p
u
l
se
will b
e
fu
rt
h
e
r attenu
at
ed
in
t
h
e fo
llowing
gates and cau
sed
n
o
fu
rt
h
e
r
d
a
m
a
g
e
as
seen
fro
m
circl
e
s
“a” in Fi
gure
4 (a, b).
b)
Pul
s
e
out
put
-
Ove
r
a
sm
all
range
o
f
i
n
p
u
t
p
u
l
s
e wi
dt
h
,
t
h
e
pul
se
o
u
t
p
ut
i
s
gene
rat
e
d
.
It
i
s
assum
e
d t
h
at
i
f
th
e g
e
n
e
rated
pu
lse is 0.2
or
m
o
re o
f
th
e inp
u
t
pu
lse,
su
ch p
u
l
se
will b
e
v
e
ry lik
ely to
cau
s
e th
e prob
lem
as seen from
circles “b”
in
Figure
4 (a, b).
c)
St
at
e chan
ge
–
At
cert
a
i
n
wi
dt
h
wi
t
h
fi
xed
m
a
gni
t
ude
of c
u
rrent
pulse
, the stat
e holde
r
m
a
y change it
s
state as as see
n
from
circles “c” in Fi
gure
4
(a, b).
Fi
gu
re
4.
(a
) St
at
e hol
der
cha
n
ge
fr
om
l
o
w t
o
hi
g
h
(
0
-
1
);
(b
)
St
at
e h
o
l
d
e
r
c
h
an
ge
fr
om
hi
g
h
t
o
l
o
w
(1
-0
)
G
e
n
e
r
a
lly, as th
e tech
no
logy is scaled
d
o
w
n
th
e tr
an
sisto
r
s ar
e
v
e
ry v
u
l
n
e
r
a
b
l
e t
o
sof
t
er
ro
r.
Fig
u
r
e
s
5
(
a
,
b
)
and
6
show
th
e cr
itical ch
arg
e
o
f
t
h
e
in
j
ected
sof
t
er
ro
r
w
ith
d
i
f
f
e
r
e
n
t
techno
logy an
d
te
m
p
eratu
r
e at
d
i
fferen
t
n
o
d
e
s. As tem
p
eratu
r
e increas
es,
it d
e
g
r
ad
es th
e th
resh
o
l
d
vo
ltag
e
, carrier m
o
b
ility
an
d
sat
u
ration
v
e
lo
city [6, 7
]
. Th
erefore, t
h
e carrier
m
o
b
ility d
e
g
r
ad
es and
th
e drain
cu
rren
t b
e
co
m
e
s l
o
wer
resu
lt in
th
e sen
s
itiv
ity o
f
th
e n
o
d
e
toward
s
SEU is in
cr
eased
. Hen
c
e, th
e critical ch
arg
e
n
eed
ed
to
flip th
e
out
put
i
s
decre
a
sed.
To
o
b
ser
v
e t
h
e c
h
a
nge
i
n
t
e
m
p
erat
ur
e
v
a
riation
s
, t
h
e
p
r
o
cess corn
er
is set to
TT
with
the
width of t
h
e tra
n
sistors a
r
e ide
n
tical.
Figure
5(a
)
s
h
ows
the c
r
itical charge
with
respect
t
o
temperat
ure
va
riation
when t
h
e
soft e
r
ror is
injecte
d
at node (i). The c
r
itical ch
arges
red
u
ce by
38% f
o
r (1
- 0
)
cha
nge a
nd
by
51
% fo
r
(0
-1
) cha
n
ge as t
h
e
t
echn
o
l
o
gy
ch
ange
fr
om
180
nm
t
o
90 nm
. The cri
t
i
cal
char
ge al
so
dec
r
eases by
29
.2
% fo
r 1
-
0 c
h
a
nge a
n
d
8.
2%
f
o
r
0
-
1 c
h
an
ge
as t
h
e
t
e
m
p
erature inc
r
eases
from
40
to
100
f
o
r
18
0n
m
tech
no
log
y
. Similar
l
y,
for
90nm
technol
ogy t
h
e c
r
itical char
ges de
crease by 21.5%
for
1-0
c
h
a
nge
a
n
d 9.2%
for
0-1 c
h
a
n
ge
on t
h
e
sam
e
te
m
p
erat
u
r
e in
crem
en
t.
Th
e exp
e
rim
e
n
t
s are
rep
eated
at n
o
d
e
(ii) and (iii).
Fig
u
re
5(b) sho
w
s th
e critical ch
arg
e
with resp
ect
to tem
p
eratu
r
e
v
a
riation
when th
e th
e soft erro
r is
injecte
d
at node (ii) The
c
r
itical charges re
duces by 39% for (1-0) ch
a
n
ge a
nd
by
4
9
% f
o
r (0
-1
) cha
n
ge a
s
t
h
e
t
echn
o
l
o
gy
ch
ange
fr
om
180
nm
t
o
90 nm
. The cri
t
i
cal
char
ge al
so
dec
r
eases by
27
.3
% fo
r 1
-
0 c
h
a
nge a
n
d
10.1% for 0-1 change
as
t
h
e
t
e
m
p
erature inc
r
eases
from
40
to
100
fo
r 180
n
m
tech
no
log
y
. Sim
i
l
a
r
l
y,
fo
r 9
0
n
m
t
echnol
ogy
t
h
e cri
t
i
cal
charge
s de
crease
by
2
4
.
5
% fo
r
1-
0 c
h
an
ge an
d
1
2
.
9
%
fo
r 0
-
1 cha
n
ge
on
t
h
e
sam
e
te
m
p
erature
increm
ent.
Fig
u
re 6
sh
ows th
e so
ft erro
r is in
j
ected
at
n
o
d
e
(iii). The critical ch
arg
e
s redu
ce b
y
3
9
% fo
r (1-0)
chan
ge a
n
d by
51
% f
o
r
(
0
-
1
)
cha
nge a
s
t
h
e
t
echn
o
l
o
gy
ch
ange
f
r
om
18
0
nm
t
o
90
nm
. The c
r
i
t
i
cal
char
ge
also
decrea
ses
by
27.4%
for
1-0 cha
n
ge
and
8.2%
fo
r 0-
1 ch
an
g
e
a
s
th
e te
m
p
erature
increase
s
from
40
to
100
for
1
80n
m tech
no
log
y
. Si
m
i
larly, for
90
n
m
tech
no
logy th
e critical ch
arg
e
s
d
e
crease b
y
23
.7
% fo
r 1
-
0
chan
ge an
d
10
.3% f
o
r 0-
1 ch
ange
on t
h
e sa
me te
m
p
erature increm
ent. It is concluded t
h
at as
the technology is scaled down the tran
si
st
o
r
s are very
vul
n
e
rabl
e t
o
so
ft
err
o
r by
3
8
% f
o
r (1
-0
) cha
nge
and
by
51
% f
o
r (
0
-1
).
5.0
5.1
5.2
5.3
1.0
0.75
V (V)
0.5
(b)
(c)
(a)
0.25
time (ns)
(a)
2.0
(a)
(b)
(c)
2.1
2.2
2.3
1.0
0.75
0.5
0.25
V (V)
time (ns)
(b
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
502
-47
52
I
ndo
n
e
sian
J Elec Eng
& Com
p
Sci, V
o
l. 10
,
No
.
3
,
Jun
e
2
018
:
10
13
–
1
022
1
016
Fi
gu
re
5.
(a
) T
e
m
p
erat
ur
e
var
i
at
i
on f
o
r
S
I
L c
o
n
f
i
g
urat
i
o
n
at
n
ode
(i
)
wi
t
h
di
ffe
re
nt
t
ech
n
o
l
o
gy
;
(
b
)
t
e
m
p
erat
ure
va
ri
at
i
on fo
r SIL
con
f
i
g
urat
i
o
n
at
n
o
d
e
(ii) with
d
i
fferen
t techno
log
y
Fig
u
re
6
.
Temp
erat
u
r
e
v
a
riatio
n fo
r SIL co
nfigu
r
ation
at no
d
e
(iii) with
differen
t
techno
lo
g
y
3.
CO
MP
ARI
N
G
SEU
WIT
H
DIF
F
ERENT
C-ELE
MEN
T
S
I
n
pa
rt
3
,
t
h
e ex
peri
m
e
nt
s i
n
pa
rt
2 are
repeat
e
d
f
o
r si
ngl
e rai
l
wi
t
h
con
v
e
n
t
i
onal
(
S
C
)
p
u
l
l
u
p
an
d
p
u
ll down
as
sh
own
b
y
Fi
g
u
re 7
(
a) and
th
e l
a
yo
u
t
in
Fi
g
u
re 7
(
b
)
. Fo
r inpu
t A=1
B=0
,
on
ly n
o
d
e
(ii) an
d
(iii)
are vulne
r
a
b
le to soft error as
othe
r nodes a
r
e connecte
d
wi
th supply volta
ge
a
nd he
nce not
affected with
s
o
ft
error for both
SIL and SC. T
h
e critical
charges are
obtaine
d for bot
h
nodes
with the tem
p
eratures are
varie
d
fr
om
40
to
100
.
Fig
u
re
7
.
(a) Si
n
g
l
e Rail
with
co
nv
en
tio
n
a
l
pu
ll up
an
d
pu
ll
d
o
wn
; (b
) Lay
o
u
t
of sing
le
rail with
conv
en
tio
n
a
l
pul
l
up
an
d p
u
l
l
do
w
n
O
ut
B
B
A
A
B
B
A
P1
P2
N2
N1
(i)
(ii)
(iii)
P3
N3
P4
P5
P6
N4
N5
N6
W(P1,P2,P3)=1.8u
W(N1,N2,N3)=1.6u
W(P4,P5,P6)=450n
W(N4,N5,N6)=400
n
(iv)
(v)
A
(a)
(b
)
Lege
nd
(a)
(b
)
Evaluation Warning : The document was created with Spire.PDF for Python.
In
d
onesi
a
n
J
E
l
ec En
g &
C
o
m
p
Sci
ISS
N
:
2
5
0
2
-
47
52
Th
e
Ana
l
ysis
of S
o
ft Erro
r in
C-Elemen
ts (No
r
hu
za
imi
n
Ju
la
i)
1
017
In
or
der t
o
cal
c
u
l
a
t
e
t
h
e soft
e
r
r
o
r a m
e
t
hod i
s
pr
op
ose
d
as bel
o
w. F
o
r si
m
p
l
i
c
i
t
y
, an i
n
je
ct
ed cur
r
e
n
t
th
at resem
b
le
SEU is assu
m
e
d
to
h
a
v
e
trap
ezo
i
d
a
l sh
ap
e
with
wid
t
h
(t
w
),
f
a
st rising tim
e
(t
r
), slow
fallin
g
ti
m
e
(t
f
) an
d a
n
am
pl
i
t
ude
Am
p
i
. Let Am
p
1
be an am
pl
i
t
ude of
i
n
ject
e
d
p
u
l
s
e
in suc
h
a
way that produce
Out
(
t)
≥
0.
2 an
d Am
p
2
is the am
plitude
of injected
pul
se that cause th
e out
put to c
o
rrupt or cha
nge
the state to cha
nge
.
Si
nce t
h
e
gen
e
rat
e
d
pul
se
di
r
ect
l
y
pro
p
o
r
t
i
o
n wi
t
h
t
h
e
rati
o of the i
n
jecte
d
cha
r
g
e
th
at produ
ced
Ou
t(t)
≥
0.2,
(Q
injected
), wit
h
the injecte
d
charge
that cause
the state to change
Q
state-change
, the m
a
the
m
atical expressi
ons are
d
e
ri
v
e
d to d
e
scrib
e
t
h
e
resp
on
se
o
f
th
e state ho
ld
ers as illu
strated above, an
d to show wh
eth
e
r t
h
e state is
co
rrup
ted
o
r
no
t d
e
p
e
nd
ed
on
th
e po
larity o
f
th
e cu
rren
t so
urce. Th
e
p
o
sitiv
e p
o
l
arity o
f
cu
rren
t
o
n
n
-
type
drai
n ca
n ca
us
e t
h
e st
at
e t
o
c
h
an
ge
fr
om
1-
0-
1.
The
negat
i
ve
pol
a
r
i
t
y
on
n-t
y
pe
d
r
ai
n
ca
n
onl
y
rei
n
fo
rc
e l
ogi
c
st
at
e 1 [
8
]
.
T
h
eref
ore
,
by
t
a
k
i
ng i
n
t
o
c
o
nsi
d
erat
i
o
n t
h
e
p
o
l
a
ri
t
y
of
cu
rr
ent
,
a c
o
nst
a
nt
1/
2
i
s
a
dde
d
t
o
t
h
e
response e
quat
i
on t
o
indicate that there
are
50% cha
n
ces
of curre
nt to ca
use SEU. T
h
e
response
of t
h
e state
hol
der
eq
uat
i
o
ns i
s
gi
ve
n
by
(
1
)
–
(
3
)
.
0
0.5
(1)
0
.∗
∗
∗
.∗
∗
∗
0.5
(
2
)
0
0.5
(
3
)
We d
e
fin
e
d
the fo
llowing
term
s
to
illu
strate th
e sen
s
itiv
e area o
f
n
/
p
-
t
y
p
e
d
r
ai
n
o
f
differen
t
C-elemen
ts
i
m
p
l
e
m
en
tatio
n
s
:
(a)
,
Th
e area
o
f
sensitiv
e n
-
type
d
r
ain
area of SIL at no
d
e
(i)
(b
)
,
Th
e area
o
f
sensitiv
e p
-
type
d
r
ain
area of SIL at no
d
e
(ii)
(c)
,
Th
e area
o
f
sensitiv
e p
-
type
d
r
ai
n
area of SIL at no
d
e
(iii)
(d
)
,
Th
e area
o
f
sensitiv
e n
-
type
d
r
ain
area of SIL at no
d
e
(iii)
(e)
,
Th
e area
o
f
sensitiv
e n
-
type
d
r
ain
area of SC
at n
o
d
e
(i)
(
f)
,
Th
e area
o
f
sensitiv
e p
-
type
d
r
ain
area of SC
at n
o
d
e
(ii)
(g
)
,
Th
e area
o
f
sensitiv
e p
-
type
d
r
ain
area of SC
at n
o
d
e
(iii)
(h
)
,
Th
e area
o
f
sensitiv
e n
-
type
d
r
ain
area of SC
at n
o
d
e
(iii)
The t
o
t
a
l
areas
of
vul
nera
bl
e
no
des of di
f
f
er
ent
co
nfi
g
u
r
at
i
ons
of C-elements ar
e the
sum of the
drai
n of p
-
t
y
pe
an
d n-t
y
pe
, w
h
i
c
h
a
r
e gi
ve
n by
(
4
)
an
d (
5
).
,
,
,
,
(
4
)
,
,
,
,
,
,
(
5
)
Th
erefo
r
e, th
e
p
r
ob
ab
ility o
f
cu
rren
t t
h
at can
h
it th
e
d
r
ai
n
for an
y g
i
v
e
n no
d
e
s is
g
i
v
e
n
by Equ
a
tio
n (6
).
,
,
node
nodenumberitoiii
n,
p
drainofNMOSor
PMOS
circuit
SIL,
S
C
(
6
)
Th
erefo
r
e, we
can
ex
tend
th
e ab
ov
e
p
r
ob
abilit
y to
fin
d
the p
r
ob
ab
ility o
f
curren
t
th
at
can
h
i
t
for
d
i
ff
er
en
t
im
pl
em
ent
a
t
i
o
n
of
C
-
el
em
ents i
s
gi
ven
by
E
quat
i
o
n
(
7
) a
n
d
(
8
).
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
502
-47
52
I
ndo
n
e
sian
J Elec Eng
& Com
p
Sci, V
o
l. 10
,
No
.
3
,
Jun
e
2
018
:
10
13
–
1
022
1
018
,
,
,
,
(7)
,
,
,
,
,
,
(
8
)
whe
r
e,
,
are t
h
e
total area
of SIL, SC
res
p
ectively. T
h
e
nu
mb
er of
ev
en
ts is red
u
c
ed in quad
r
atic
with
neut
ro
n e
n
er
gy
. O
n
a l
o
g
-
l
o
g
pl
ot
of t
h
e n
u
m
ber of e
v
ent
per e
n
e
r
gy
,
ve
rsus
En
er
gy
(
M
eV),
as s
h
o
w
n i
n
Fi
gu
re
8 [
9
]
can be
ap
pr
o
x
i
m
at
ed by a strai
g
ht line for t
h
e interval of
“1
–
100” Me
V. T
h
e line can
be used t
o
pre
d
i
c
t
t
h
e e
r
r
o
r
rat
e
of
t
h
e
st
at
e hol
ders
b
y
neut
ro
n e
n
er
gy
. T
w
o
para
m
e
t
e
rs
a
n
d
can be extract
e
d
fro
m
th
e g
r
aph as fo
llows:
1)
C
o
nst
a
nt
equ
a
ls to th
e y-i
n
tersect of th
e st
raigh
t
lin
e segmen
t o
f
th
e
p
l
ot.
2)
C
o
nst
a
nt
is
th
e slop
e
o
f
th
e straigh
t
lin
e seg
m
en
t o
f
t
h
e
p
l
o
t
.
The st
rai
g
ht
l
i
n
e
of
s
p
ect
r
u
m
de
nsi
t
y
o
f
ne
ut
r
o
n
t
h
at
i
s
l
a
rge
r
or
eq
ual
t
o
1 M
e
V,
,
can
be m
odel
l
e
d as
i
n
(9
),
∗
MeV/
/
(
9
)
Fi
gu
re
8.
Ne
ut
r
o
n
ene
r
gy
spec
t
r
um
[9]
For
s
p
ect
r
u
m
den
s
i
t
y
of
ne
u
t
ro
n t
h
at
i
s
eq
ual
or
sm
al
l
e
r t
o
1
M
e
V,
,
the equati
on
can be ap
pr
o
x
i
m
a
t
e
d from
Fi
gu
re 9
(
a) b
y
[10]
. The si
m
u
l
a
t
i
ons o
n
fo
ur
di
ffe
re
nt
con
f
i
g
urat
i
o
ns
of C
-
ele
m
ents show that the c
r
itical energy
nee
d
e
d
t
o
cause
0
.
2
of i
n
p
u
t
pul
se
or
ca
usi
n
g the
state to cha
n
ge
is lies
bet
w
ee
n
0
.
1
5
M
e
V
t
o
0.
9
M
e
V. In ot
he
r wo
rd
s,
e
n
er
gy
th
at is less th
an
1
MeV is su
fficien
t
to
cau
s
e th
e
o
u
t
p
u
t
of
C-
el
e
m
en
t to
ch
ang
e
.
Th
is r
a
n
g
e is as sho
w
n by th
e r
e
d cir
c
l
e
in
Figu
r
e
9(a)
. Th
e con
s
tan
t
360
0
refe
rs t
o
t
h
e c
o
nve
rsi
o
n
o
f
sec
o
n
d
t
o
ho
u
r
. T
h
e a
p
p
r
oxi
m
a
te eq
uat
i
o
n
o
f
t
h
e l
i
n
e i
s
gi
ven
as,
1)
C
o
nst
a
nt
eq
u
a
ls to th
e sl
o
p
e of th
e straight lin
e seg
m
en
t.
2)
C
o
nst
a
nt
equ
a
ls th
e y-in
tersect o
f
th
e
strai
g
h
t
lin
e seg
m
e
n
t.
3600
∗
1
0
MeV/
/hr
(
1
0)
The ratio of norm
a
lized atm
o
spheri
c n
e
u
t
ron
cross sectio
n with
th
e drain area of PMOS and NM
OS
with
0.1
t
echnol
o
g
i
e
s f
o
r
neut
ro
n ene
r
gy
i
s
s
h
o
w
n i
n
Fi
g
u
re
9
(b
) [
11]
.
As t
h
e 90
nm
-t
echn
o
l
o
gy
i
s
used
i
n
t
h
e
si
m
u
l
a
tion
,
t
h
e
rat
i
o
ca
n
be a
p
p
r
oxi
m
a
t
e
d by
ne
ut
r
o
n c
r
oss
sect
i
o
n
an
d
drai
n cr
os
s sect
i
o
n
wi
t
h
0.
1
t
echn
o
l
o
gy
. It
i
s
very
o
b
v
i
ous
fr
om
t
h
e gra
p
h t
h
at
N
M
OS t
r
a
n
si
st
o
r
i
s
m
o
re vul
nera
bl
e t
o
war
d
s SE
U
com
p
ared wit
h
PMOS tra
n
sistor. Aut
h
ors
in [12] su
ggested that for the sa
m
e
tra
n
sistor widt
h, NMOS
tran
sistor is 2
.
2
ti
m
e
s
m
o
re s
e
n
s
itiv
e co
m
p
ared
with
PM
OS
tran
sistor. Th
is is d
u
e
to
the co
llected
ch
arg
e
fo
r
drai
n NM
OS i
s
hi
g
h
er
f
o
r
d
r
ai
n PM
O
S
. T
h
e equ
a
t
i
ons
for norm
alized cross
section of PMOS a
n
d NMOS
straight
-line fo
r 0.
1
t
ech
nol
o
g
y
are
gi
ven
b
y
(1
1)
an
d
(
1
2
)
.
1
0
(
1
1)
1
0
(
1
2)
Time
10
100
1000
10
0.1
1
0.01
0.001
dN/dE [n/MeV/cm2/hr]
neutron energy [MeV]
Evaluation Warning : The document was created with Spire.PDF for Python.
In
d
onesi
a
n
J
E
l
ec En
g &
C
o
m
p
Sci
ISS
N
:
2
5
0
2
-
47
52
Th
e
Ana
l
ysis
of S
o
ft Erro
r in
C-Elemen
ts (No
r
hu
za
imi
n
Ju
la
i)
1
019
whe
r
e,
,
,
and
are constants. Four
param
e
ters
,
,
and
can be
extracted
from the gra
ph
of
Fi
gu
re
9(
b
)
as
fol
l
o
ws:
1)
C
o
nst
a
nt
and
equal
t
o
t
h
e s
l
ope
o
f
t
h
e
st
ra
i
ght
l
i
n
e
segm
ent
o
f
nm
os an
d
pm
os g
r
ap
h
re
spect
i
v
el
y
.
2)
C
o
nst
a
nt
and
eq
ual
t
h
e y
-
i
n
t
e
rsect
of
t
h
e
st
rai
ght
l
i
n
e se
gm
ent
of
nm
os an
d
pm
os gra
p
h
respect
i
v
el
y
.
Fi
gu
re
9.
(a
)
N
e
ut
r
o
n
spect
ru
m
bel
o
w 1
M
e
V i
n
cl
udi
ng
t
h
erm
a
l
-
energy
[
10]
;
(
b
)
No
rm
al
i
zed at
m
o
sphe
ri
c
neut
ro
n c
r
o
ss s
ect
i
on
[1
1]
The er
ro
r rat
e
of
neut
ro
n spe
c
t
r
um
energy that causes
failure for any nod
e of NM
OS transist
or for spectrum
of
ene
r
gy
m
o
re
t
h
an
1
M
e
V
i
n
st
at
e h
o
l
d
er
i
s
gi
ve
n
by
(
1
3) a
n
d
(
1
4).
,
∗
∗
(
1
3)
,
∗
∗1
0
∗
∗
(
1
4)
Si
m
ilarly,
th
e erro
r rate
o
f
neu
t
ro
n
s
p
ectrum energy that cause failure fo
r a
n
y
no
de o
f
NM
OS t
r
a
n
si
st
or f
o
r
spect
r
u
m
of e
n
ergy
l
e
ss
t
h
a
n
1 M
e
V
i
n
st
at
e h
o
l
d
e
r
i
s
gi
ve
n
by
eq
uat
i
o
ns
(1
5)
an
d
(
1
6
)
.
,
∗
∗
∗
(15)
,
∗
∗1
0
∗
3600
∗
1
0
∗
(16)
Eq
uat
i
ons
(
1
4)
an
d
(1
6
)
a
r
e
adde
d t
o
cal
c
u
l
a
t
e
t
h
e t
o
t
a
l
e
r
r
o
r
rat
e
of
ne
ut
r
o
n
spe
c
t
r
u
m
energy
t
h
at
cause
fai
l
u
re f
o
r
a
n
y
no
de o
f
NM
O
S
t
r
a
n
si
st
or
i
s
gi
ve
n by
(
1
7)
a
n
d
(
1
8).
,
,
(17)
∗
∗1
0
∗
∗
∗
(18)
∗
∗1
0
∗
3600
∗
1
0
∗
The
e
r
r
o
r rat
e
of ne
ut
r
on spe
c
t
r
um
energy
t
h
at
cause
s f
a
il
u
r
e
fo
r
an
y nod
e
o
f
PM
OS t
r
ansistor for spectrum
of
ene
r
gy
m
o
re
t
h
an
1
M
e
V
i
n
st
at
e h
o
l
d
er
i
s
gi
ve
n
by
E
quat
i
on
(
1
9
)
.
,
∗
∗
∗
(
1
9)
Equ
a
tio
n (19
)
can
b
e
written as sh
own
b
y
(2
0). Howev
e
r,
sin
ce en
erg
y
t
h
at is less t
h
an 1
MeV
on
ly affect
NM
OS t
r
a
n
si
s
t
or [
13]
, t
h
e t
o
t
a
l
erro
r rat
e
of ne
ut
r
on s
p
ect
rum
energy
t
h
at
cause fai
l
u
re f
o
r any
n
ode
of
PMOS t
r
an
sist
o
r
is equ
a
l to Eq
u
a
tion
(2
1).
(a)
(b
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
502
-47
52
I
ndo
n
e
sian
J Elec Eng
& Com
p
Sci, V
o
l. 10
,
No
.
3
,
Jun
e
2
018
:
10
13
–
1
022
1
020
,
∗
∗1
0
∗
∗
(
2
0)
,
0
∗
∗1
0
∗
∗
(
2
1)
Th
e
p
r
o
b
a
b
ility can
b
e
ex
tend
ed
i
n
ord
e
r t
o
fi
n
d
t
h
e to
tal p
r
o
b
a
b
ility d
u
e
to
th
e
drain o
f
NMOS or
PMOS
t
r
ansi
st
o
r
of
an
y
gi
ve
n C
-
el
e
m
ent
ci
rcui
t
as
sho
w
n
by
(
2
2)
–
(2
5)
.
,
,
,
(
2
2)
,
,
,
(
2
3)
,
,
,
(24)
,
,
,
(25)
whe
r
e,
,
are th
e
to
tal area o
f
SIL, SC resp
ectiv
ely. Th
e t
o
tal p
r
o
b
a
b
ility o
f
cu
rren
t
p
u
l
se
th
at can
h
i
t
fo
r
NM
OS
an
d
PM
OS
t
r
a
n
si
st
or
i
n
ci
rc
ui
t
are
gi
ve
n
by
(2
6)
and
(
2
7).
,
∗
,
(
2
6)
,
∗
,
(
2
7)
Th
e t
o
tal errors rate
d
u
e
to
SEU
o
f
an
y co
nfigurations
of C
-
elem
ents are given
by,
(
2
8, a)
,
∗1
0
∗
∗
,
∗1
0
(
2
8,
b)
∗
3600
∗
1
0
∗
∗
,
∗1
0
∗
∗
∗
∗1
0
∗
∗
(
2
9)
In
o
r
de
r t
o
ve
ri
fy
t
h
e
pr
o
pose
d
t
ech
ni
q
u
e,
w
e
ha
ve
gene
rat
e
d a
ran
d
o
m
wi
dt
h
of c
u
rre
nt
i
n
ject
e
d
i
n
t
o
di
ffe
re
nt
t
y
pes of C
-
el
em
ent
.
A M
ont
e-C
a
rl
o anal
y
s
i
s
was
used t
o
ge
ne
r
a
t
e
rand
om
sam
p
l
e
wi
t
h
t
h
e num
ber
of sam
p
les were fixed at
1000. The error
rate for bo
th S
I
L a
nd
SC are calculated by
using the m
e
thod s
h
own
above.
As sh
o
w
n in F
i
gu
re 1
0
, the err
o
r rates f
o
r S
I
L in
crease
with the increase
of
tem
p
eratu
r
e d
u
e
to
the
d
e
grad
atio
n of th
e m
o
b
ility carrier. Th
erefore, t
h
ese
n
o
d
e
s are m
o
re
v
u
l
nerab
l
e t
o
SEU
at h
i
gh
tem
p
eratu
r
e.
T
h
e
er
ro
r r
a
tes
of
0
-
1
ch
ang
e
in
cr
e
a
s
e b
y
29
.1
%
and
the error rates
of
1-0 c
h
ange
increase
by
132%
by
in
creasing th
e
te
m
p
erature
from
40
to
100
. From
the i
n
crem
ent of c
r
itical cha
r
ge, it is
concl
u
ded t
h
a
t
the PM
OS tra
n
sistors
had greater effect
on te
m
p
erat
ure
variatio
n
co
m
p
ared with NM
OS. Sim
ilarly fo
r SC
,
t
h
e err
o
r
rat
e
s
of
0-
1 cha
n
ge
i
n
crease
by
1
4
.
2% a
nd t
h
e er
ro
r rat
e
s o
f
1-
0 cha
n
ge i
n
cre
a
se by
7
3
.
9
%
whe
n
increasing t
h
e t
e
m
p
erature from
40
to
100
as see
n
i
n
Fi
gure
11.
To
id
en
tify which
i
m
p
l
e
m
en
t
a
tio
n
s
are m
o
re resistan
t to
ward
s soft erro
r
at d
i
fferen
t
tran
sitio
n, th
e
error
rates from
Figures 10
and 11 a
r
e added for the sam
e
C-elem
ent. Figu
re 12
(a,
b
)
sho
w
s th
e to
t
a
l erro
r
rat
e
o
f
neut
ro
n
spect
r
u
m
ener
gy
f
o
r
1
-
0 a
n
d
0-
1
o
n
di
ffe
ren
t
C
-
el
em
ent
s
re
spect
i
v
el
y
.
We
co
ul
d c
o
ncl
u
d
e
t
h
at
SIL im
p
l
e
m
en
tatio
n
s
are m
o
re resistan
t t
o
w
a
rd
so
ft er
r
o
r c
o
m
p
ared
with
SC.
Evaluation Warning : The document was created with Spire.PDF for Python.
In
d
onesi
a
n
J
E
l
ec En
g &
C
o
m
p
Sci
ISS
N
:
2
5
0
2
-
47
52
Th
e
Ana
l
ysis
of S
o
ft Erro
r in
C-Elemen
ts (No
r
hu
za
imi
n
Ju
la
i)
1
021
Fig
u
r
e
10
. Er
ror
r
a
te
du
e to neu
t
ro
n en
er
g
y
s
p
ectrum
with respect
to tem
p
eratu
r
e SIL
Fig
u
r
e
11
. Er
ror
r
a
te
du
e to neu
t
ro
n en
er
g
y
s
p
ectrum
with respect to tem
p
erature
SC
Figu
re
1
2
.
(a)
Total er
ro
r
rate d
u
e to
ne
utr
o
n
ene
r
gy
s
p
ectrum
with
resp
ect
to
tem
p
eratu
r
e
SIL;
(b) To
tal trro
r
rate due t
o
neutron e
n
ergy s
p
ectru
m
with re
spect to tem
p
erature
SC
4.
CO
NCL
USI
O
N
In t
h
i
s
pa
per
,
we ha
ve i
n
ject
ed er
ro
r
on
SI
L co
n
f
iguration
of C-Elem
en
ts.
As th
e tech
no
log
y
is
scaled down, the tra
n
sistors a
r
e ve
ry
vu
ln
erab
le to
so
ft error and
th
is
h
a
s been affirm
ed from
the experi
ments
t
h
at
we
ha
d
p
e
rf
orm
e
d o
n
S
I
L.
We
de
vel
o
ped
a m
e
t
hod
t
o
cal
cul
a
t
e
t
h
e err
o
r rat
e
d
u
e
t
o
neut
ro
n e
n
er
gy
spectrum
.
This
m
e
thod ca
n
be use
d
t
o
as
ses
s
the
vulne
r
abi
lity of s
o
ft e
r
ror towards
different m
e
m
o
ry circuit
configurations
. The prese
n
te
d calculation indicates that SIL im
ple
m
enta
tions a
r
e m
o
re resistant toward soft
erro
r co
m
p
ared
wit
h
SC.
Futu
re
work
will b
e
fo
cu
sing
on
th
e i
n
corp
oratio
n
o
f
sen
s
or techn
o
l
o
g
y
with
in
el
ect
roni
c ci
rc
ui
t
s
by
c
o
nsi
d
e
r
i
n
g
di
ffe
rent
a
s
pect
s as
i
n
[1
4
-
1
6
]
.
ACKNOWLE
DGE
M
ENTS
Th
is
work
h
a
s b
e
en
supp
orted
b
y
MyRA gran
t (F02
/Sp
S
TG/137
8
/
1
6
/2
0), Un
iv
ersiti
Malaysi
a
Sara
wak.
0,
00
E+00
2,
00
E-
13
4,
00
E-
13
6,
00
E-
13
8,
00
E-
13
SIL (Erro
r
Rate /h
r) v
s
Tem
p
eratu
r
e
Out
=
0-1
Out=1-0
0,
00
E+00
2,
00
E-
13
4,
00
E-
13
6,
00
E-
13
8,
00
E-
13
1,
00
E-
12
1,
20
E-
12
1,
40
E-
12
SC(Erro
r
Rate /h
r) v
s
Tem
p
eratu
r
e
Out=0-1
Out=1-0
(
a
)
(b
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
502
-47
52
I
ndo
n
e
sian
J Elec Eng
& Com
p
Sci, V
o
l. 10
,
No
.
3
,
Jun
e
2
018
:
10
13
–
1
022
1
022
REFERE
NC
ES
[1]
S. Hauck
,
“
Asyn
chronous design
met
hodolog
ies: An
overview”,
P
r
oceed
ings
of
th
e IE
EE
,
vol 8, 1
995, pp
69 –
93.
[2]
F.
Wa
ng,
et
al
.,
“
Single event up
set: an embed
d
ed tutorial”,
21st Intern
ation
a
l C
onference on V
L
SI Design, 200
8,
pp 429 –
434, 20
08.
[3]
L. Anghel
,
et al., “Multi-l
e
vel f
a
ult eff
e
cts evalu
a
tion
”
,
Radiatio
n Effects on
Embedded
S
y
s
t
ems, Springer, 2007
,
pp. 69-88
.
[4]
H. Cha, et
al.
,
“
A
L
ogic L
evel
Model for
Alph
a Par
ticl
e
Hits
in CMOS Cir
c
uits
”
,
Interna
tion
a
l Conferen
ce o
n
Computer Desig
n
, 1993
, pp
. 538
-542.
[5]
G. F
u
chs
,
et al.,
“On the threat o
f
metastability in
an asynchronou
s fault-tolerant clock ge
n
e
ration scheme”, Fault-
Tolerant Distributed Algorithms
on VLSI Chips.
15th IEEE S
y
m
posium on Asy
n
chronous Cir
c
uits and S
y
stem
s,
2009, pp
. 1522
–
8681.
[6]
T. H
e
ijm
en
, et
a
l
.,
“A Comparative Stud
y on the Soft-Error Rate of
Flip-
F
lop fr
om 90-nm prod
uction Libraries”
44th IEEE Inter
n
ation
a
l S
y
mpos
ium on
Reliability
Ph
y
s
ics
,
2006
,
pp. 204-2011
.
[7]
R, Kum
a
r, et
al.
,
“Impact of Temperature Fluct
uation on Circu
its Characteri
stics in 180nm and 65 nm CMOS
Technologies”
, I
n
ternational S
y
mposium on Circu
its
and S
y
s
t
ems, 2006, pp. 385
8-3861.
[8]
M
.
Zhang
,
e
t
al.
,
“A Sof
t Error
Rate
Anal
ysis (
S
ERA)
Methodo
l
ogy”
, Internatio
nal Conf
erence
Computer Aided
Design, 2004
, p
p
. 111-118
.
[9]
T
.
Ka
rni
k
,
et
a
l
.
,
“Cha
ra
ct
e
r
iz
ati
on of soft errors caused b
y
single even
t upsets in CMOS pr
ocesses”,
IE
EE
Transactions on
Dependable and
Secure Computing,
vol 1, pp. 12
8-143, 2004
.
[10]
JEDEC Solid State Tec
hnolog
y
Association,
Measurement and Reporting of Alph
a
Particle and T
e
rrestrial Cosmic
Ray-Induced So
ft Errors
in Semiconductor Devices
, Octob
e
r 2006
.
[11]
P. Hazuch
a,
et
al., “Impact of
CMOS Technolog
y
Sc
aling
on
the Atmospheric Neutron Soft
Error Rate”,
IE
EE
Transactions on
Nuclear S
c
ien
c
e,
vol 47
, pp
. 2586
-2594, 2000
.
[12]
T
.
Ka
rni
k
,
et
a
l
.
,
“Cha
ra
ct
e
r
iz
ati
on of soft errors caused b
y
single even
t upsets in CMOS pr
ocesses”,
IE
EE
Transactions on
Dependable and
Secure Computing,
vol 1, pp. 12
8-143, 2004
.
[13]
X. W. Zhu,
et al., “Charge D
e
po
sition Modelling
of Th
ermal Neu
t
ron Products in
Fa
st Submicron MOS Devices”,
IEEE Transactio
ns on
Nucl
ear
Sc
ienc
e,
vol 46
, pp
.1378-1385. 200
6.
[14]
Z. Abidin, e
t
a
l
., “
L
ow Comm
on-Mode Gai
n
Instrum
e
nt
ati
on Am
plifier Archit
ectur
e Insensitive to Resis
t
or
M
i
s
m
atches
”
International Jour
nal of Electr
ical and Computer Engineering,
vo
l. 6, pp. 3247 –
3254, Decem
ber
2016.
[15]
F. Khair,
et
al. “Performance An
aly
s
is of Digital Modulation for
Coherent
Detection of OFDM Scheme on Rad
i
o
over Fiber S
y
stem
”, International Journal of El
ectrical and Computer Engineer
ing,
vo
l. 6, pp. 1
086 – 1095,
June
2016.
[16]
A. Beddiaf
, et al., “A Numerical
Model of Joul
e
Heating in P
i
ezo
res
i
s
tive P
r
es
s
u
re S
e
ns
ors
”
,
Inter
national Journa
l
of Electrical and
Co
mputer
Eng
i
neer
ing
, vol. 6
,
pp. 1223
– 1232
,
June 2016,
BIOGRAP
HI
ES
OF AUTH
ORS
Dr. Norhuzaim
i
n
J
u
lai r
e
c
e
ive
d
his
P
h
D in El
ect
rica
l &
El
ect
ron
i
c Eng
i
ne
ering f
r
om
Newcas
tle
University
, UK in 2015. He is with the Depar
t
emen
t of Electrical and El
ectronics Engineering
,
Facult
y of
Engi
neering
,
Univers
iti Mal
a
y
s
ia Sar
a
wak. His rese
ar
ch are
a
of in
tere
st is in the soft
error in
in
tegrated cir
c
uit (IC).
Dr. Ahm
e
d M. A. Haidar
is c
u
rrentl
y
an Asso
cia
t
e Professor
with the
Univ
ersiti Ma
la
ysia
Sarawak and an Adjunct Asso
ciat
e Professor with the Univer
sity
of Souther
n
Queensland,
Aus
t
ralia
. His
re
s
earch in
ter
e
s
t
s
i
n
clude s
m
art
ele
c
tri
c
it
y gr
ids
,
s
u
s
t
ainabl
e en
erg
y
and ren
e
wable
energ
y
in
tegr
ati
on, int
e
ll
igen
t p
o
wer s
y
s
t
em
co
ntrol and
optim
i
zat
ion, power
q
u
alit
y,
s
e
curit
y
,
and reliability
.
He is a Professi
onal Engin
eer w
ith the Instit
utio
n of Engineers Australia
, and
a
S
e
nior m
e
m
b
er
of the
Institu
te
o
f
El
ectr
i
c
a
l
and
Ele
c
troni
c
Engin
eers (I
EEE)
.
Dr. Abdul Rahm
an Kram
receiv
e
d his PhD in 201
4. He is with the
Departem
ent of
Ele
c
tri
cal
and
Ele
c
troni
cs
Eng
i
neer
ing, F
acu
lt
y of Eng
i
ne
erin
g, Univers
i
t
i
M
a
la
ys
i
a
S
a
rawak
.
His
res
ear
ch
areas
are in
optical, visible
light communica
tion (
V
LC), wir
e
less and antenna communication
.
Evaluation Warning : The document was created with Spire.PDF for Python.