TELKOM
NIKA
, Vol.14, No
.3, Septembe
r 2016, pp. 8
31~838
ISSN: 1693-6
930,
accredited
A
by DIKTI, De
cree No: 58/DIK
T
I/Kep/2013
DOI
:
10.12928/TELKOMNIKA.v14i3.3715
831
Re
cei
v
ed Ap
ril 4, 2016; Re
vised J
une 5,
2016; Accept
ed Ju
ne 22, 2
016
A Comprehensive Test Approa
ch on High-Power Low-
Noise Intermodulation Distortion
Lei Wang
1,2
, Jing
y
i
Zhang
1*
1
Ke
y
L
abor
ator
y of Spec
ial F
i
ber Optics and
Optica
l Access
Net
w
o
r
ks, Min
i
str
y
of Educ
ati
o
n
2
Shang
hai H
u
a
hon
g Grace Se
micon
ductor M
anufactur
i
n
g
C
o
rpor
ation
*Corres
p
onding author, e-ma
il: zhangjiny
i
@s
taff.shu.edu.cn
A
b
st
r
a
ct
W
i
th the short
age
of w
i
reles
s
communic
a
ti
on b
and
w
i
d
th r
e
sourc
e
, the r
adi
o interfer
en
ces occur
so
freq
uently. Currently, effcient
freq
uency
alloc
a
tio
n
a
l
g
o
r
ithm des
ig
nin
g
an
d Inter
m
o
duati
on
Distort
ion
(IMD) suppres
sion ar
e tw
o mea
n
s to ration
ally i
m
pr
ove the ba
nd
w
i
dth res
ourc
e
. T
herefore,
four
compre
hens
ive
appro
a
ches n
a
med sti
m
ul
us
isolatio
n, cha
nne
l crosstalk
isol
ati
on; spect
r
um sl
ight offset
and A
u
to
Lev
el C
ontrol
(AL
C
) le
ak co
ntro
l are
pro
pos
e
d
resp
ective
ly
to avo
i
d th
e
restriction
of
the
peri
phery syst
em
’
s
no
ise an
d dyna
mic ran
ge of measur
e
m
e
n
t instru
me
nts. Moreover, the hig
h
pow
er
an
d
low
noise
det
ection a
ppr
oa
ch, the auxi
lia
ry co
mp
on
ent
s ameli
o
ratio
n
and the
me
a
s
ure
m
e
n
t system
improve
m
ent a
r
e
an
aly
z
e
d
.
Fi
nally, utili
z
i
n
g
Silico
n
-On-
Ins
u
lator (SOI) R
a
dio Fre
q
u
ency
(RF) sw
itch as the
carrier to
do
the ex
per
i
m
ent b
a
sed
o
n
Adva
ntest
9
3
K tester. Ex
peri
m
e
n
t res
u
l
t
s show
that
the
compre
hens
ive
opti
m
i
z
e
d
a
p
p
r
oach
e
s ca
n k
eep
the
w
hole
system to
l
e
ss
than -
1
5
0
dB
m (ne
a
rly
170
d
B
c)
low
nois
e
ran
g
e
un
der the
lar
ge sig
n
a
l
case
s. T
he actual
i
n
termo
d
u
l
atio
n
distortio
n
sig
n
a
l
coul
d b
e
rej
e
c
t
ed
and s
a
mp
led
i
n
prec
ise
accu
racy w
h
ich
is
near
ly 2
0
% i
m
prove
d
. W
hat
’
s
more, the
ap
p
r
oach
e
s ar
e a
l
so
ben
eficia
l to the expa
nsi
on of t
he industrial
multi-site test.
Ke
y
w
ords
: int
e
rmod
uatio
n di
stortion, crosstalk, low
noise
Copy
right
©
2016 Un
ive
r
sita
s Ah
mad
Dah
l
an
. All rig
h
t
s r
ese
rved
.
1. Introduc
tion
With the fa
st developme
n
t of the social
i
n
form
a
t
ization, the
studie
s
of
wirel
e
ss
comm
uni
cati
on techn
o
log
y
are corre
s
p
ondin
g
unde
r the strong sp
otlight. Due to the limitation of
band
width
re
sou
r
ce, wi
rel
e
ss
comm
uni
cation
s
houl
d
take
full a
n
d
ratio
nal
ad
vantage
of the
band
width [1
-3]. Mean
wh
ile, IMD (Intermo
duatio
n
Disto
r
tion) verification method
s
attract
peopl
e's mo
re attention
a
nd he
nce mo
re resea
r
che
s
a
r
e involve
d
[4, 6]. At prese
n
t, high
p
o
we
r
and l
o
w noi
se
dete
c
tion
of inte
rmod
ulation
disto
r
tion is la
rgel
y re
stricte
d
t
o
the
pe
riph
ery
system’
s
noi
se an
d dyna
mic ra
nge of
measurem
e
n
t instrum
ent
s [7, 8]. Larg
e
sign
al and
low
noise IMD is
easily
dro
w
n
ed in
the n
o
ise floor of
the
whol
e
system
. It is sho
w
n i
n
mo
st repo
rted
experim
ental
result
s that aroun
d -10
0
dBm
coul
d be cont
rol
l
ed at high power stim
ul
us.
Therefore, th
e comp
reh
e
n
s
ive o
p
timize
d ap
pro
a
che
s
re
sea
r
ch a
n
d
sele
ction i
s
a very im
po
rtant
subj
ect a
s
to enha
nce IMD signal d
e
tecti
on accu
ra
cy of the
actual device.
Based
on th
e study a
nd
analysi
s
of m
easur
e
m
ent
method, auxil
iary devices
and te
st
system, thi
s
pape
r p
r
op
ose
s
the
co
mpre
hen
sive
optimize
d
approa
che
s
whi
c
h
coul
d
be
sep
a
rate
d to four parts.
The first ap
proa
ch i
s
de
fined as sti
m
ulus i
s
olati
on. The se
cond
approa
ch i
s
chann
el cro
sst
alk i
s
ol
ation.
The third ap
p
r
oa
ch i
s
calle
d sp
ectrum
sl
ight offset an
d
ALC (Auto L
e
vel Control) leak control
is in
volved in the fourth
approa
ch. Fi
nally, the paper
utilizes the SOI RF switch as carrier to
do the
experi
ment on Adv
antest 93K
tester. As a
result,
the floo
r n
o
ise is redu
ce
d
to less th
an
150
dB
m
(ne
a
rly 17
0
dBc), and th
e
act
ual a
c
curacy
is
also in
crea
se
d by over 20
%. What’s m
o
re, it al
so p
r
ovides the
ce
rtain theo
retical sup
port on
the
high po
we
r a
nd low n
o
ise intermo
dulatio
n distortio
n
m
easure
m
ent.
2. IMD Theor
etical An
aly
s
is
IMD si
gnal
is
defined
as th
at whe
n
two
or m
o
re
different freq
uen
cy
sign
als go
th
roug
h a
nonlin
ear
system, much
more h
a
rm
on
ic and
com
b
i
ned fre
que
ncy compo
nent
s are produ
ced
due to the no
nlinea
r factors. O
ne of the signal
s is m
odulate
d
with
another o
ne,
and pro
duct
the
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 14, No. 3, September 20
16 : 831 – 838
832
mixed sp
urio
us si
gnal
s. C
onsequ
ently it affects the
identificatio
n of the useful
sign
al as
sho
w
n i
n
Figure 1.
Figure 1. Intermod
ulation
Disto
r
tion Pro
ductio
n
The exa
c
t m
eanin
g
is
sh
own th
at a li
near sy
stem
contai
ns th
e
value of the
nonlin
ear
coeffici
ent. T
he inte
rmod
ul
ation p
r
od
ucti
on of
nonlin
e
a
r d
e
vice
con
s
ist
s
of
a
seri
es
of si
gnal
with
different fre
q
uen
cie
s
. The
y
includ
e fun
damental
wa
ve, second
o
r
de
r ha
rmo
n
i
c
an
d third
o
r
de
r
IMP (Intermo
dulation Di
st
ortion Pro
d
u
c
tion) a
nd
e
t
c. Unde
r the conditio
n
of small sig
nal
distortio
n
mo
del, Macla
u
ri
n expan
sion
equatio
n app
roximately equals to form
ul
a (1) u
s
u
a
lly:
∑
∞
∙
(
1
)
In the formula,
is Maclauri
n
multinomial
coeffici
ent. Then two-to
ne sign
al is supp
ose
d
to be stimulat
ed,su
ch a
s
Equation (2):
cos
cos
(
2
)
They
a
r
e co
mbined
an
d pro
c
e
s
sed by
Pr
o
s
thaph
ae
resi
s t
r
an
sform, and fu
nd
amental
wave compo
nent
ω
and
ω
is calcul
ated by simple and th
ree cu
bed terms a
s
formul
a 3:
cos
c
o
s
(3)
Combi
ned f
r
eque
ncy
com
pone
nts a
r
e
gene
rated
by quad
ratic an
d thre
e cube
d term
s
and sho
w
n a
s
formul
a (4
-6):
cos
cos
(
4
)
cos
2
cos
2
(
5
)
cos
2
cos
2
(
6
)
Acco
rdi
ng to
the a
bove
ava
ilable
cal
c
ulat
ion re
sults, di
fferent fre
que
ncie
s
are
cau
s
ed
by
the cro
s
s terms. And th
ey are
likely to f
a
ll with
in
the pass
b
and of receiver. Esp
e
cially whe
n
ω
and
ω
are very
clo
s
ed to
ω
, the two sort
s of
differen
c
e fre
quen
cy wo
uld
be ha
rmful to re
ceive
r
seri
ou
sly. And the kind of
difference freque
ncy is d
e
fined as IM
D3 (3
rd O
r
de
r Intermod
ula
t
ion
TX
BW
RX
BW
f1
f2
f2
‐
f1
f3
f4
2*
f1
‐
f2
2*
f
2
‐
f1
2
*
f1
2*f
2
3*f
1
3*
f
2
Po
w
e
r
(d
Bm
)
F
r
e
q
ue
ncy
(H
z
)
2
nd
ord
e
r
IM
P
3
rd
ord
e
r
IM
P
2
nd
Ha
r
m
o
n
i
c
3
rd
Ha
r
m
o
n
i
c
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
A Com
p
rehe
nsi
v
e Te
st Approa
ch o
n
Hi
gh-Po
we
r Lo
w-Noi
s
e Interm
odulation…
(Lei Wang
)
833
Disto
r
tion
). Relatively sp
ea
king,
be
cau
s
e the
hi
gh
-o
rder term
intermodulatio
n
si
gnal
s a
r
e
al
ways
far a
w
ay to
t
he p
a
ss ban
d of
re
ceive
r
, it bri
n
g
less
influen
ce. T
h
erefo
r
e, follo
wing
pa
per
would
mainly focu
s on the com
p
reh
e
n
s
ive o
p
timization a
ppro
a
che
s
’ a
nalysi
s
ba
se
d on IMD2
(2
nd
Orde
r Interm
odulatio
n Di
stortion)
and I
M
D3. And ve
rification
mod
e
l is al
so e
s
t
ablished o
n
four
method
s su
ch
as stimul
us
isolat
io
n, ch
annel
crossta
l
k
isolation,
spectrum
sligh
t
offset and A
L
C
(Auto Level Control
)
lea
k
control.
3. Stud
y
and Optimizatio
n
on Test
Ap
proach
Becau
s
e th
e
root ca
use
on interm
odu
lation sig
nal
is co
mpli
cate
d, the gene
ration of
intermo
dulati
on produ
cts
couldn’t be
obt
ained a
c
cu
ra
t
e
ly by theoret
ical a
nalysi
s
and
cal
c
ulati
o
n
[9-12]. Th
erefore, th
e o
p
timized
te
st
sy
st
em’s
ar
chit
e
c
t
u
re i
s
est
a
blishe
d b
a
sed
o
n
the t
r
adition
al
test sy
stem.
As dem
on
strated in Fi
gu
re 2,
four ap
proache
s a
r
e e
x
pande
d on t
he core ap
proach
architectu
re b
o
th on peri
p
h
e
ry circuit an
d algor
ith
m
o
p
timization. Detail algorith
m
and mod
e
l
on
the four app
roache
s wo
uld
be introdu
ce
d in the latter part of the pa
per.
Figure 2. Test System Architecture Opti
mization
On stimulu
s
isolatio
n, the whole o
p
timized
te
st system is esta
bl
ishe
d as
sho
w
n in
Figure 3. Th
e splitter i
s
adde
d to ca
pture t
he t
w
o-tone
sig
nal
stimulated t
o
assu
re th
e
compl
e
tion of
the DUT
(Device
Und
e
r
Test)’
s in
put
power.
DC
bl
ock is u
s
ed t
o
avoid th
e
DC
sign
al involve
d
in the
syste
m
. BPF (Ban
d Pass
Filter) and LPF
(Lo
w
Pass Filte
r
) in the
syste
m
are
used to d
o
IMD3
and I
M
D2 m
e
a
s
urement ex
clud
ed
la
rge
stim
ulus
sig
nal eff
e
ct. In traditio
nal
way,
two
sin
g
le-ton
e sign
als are combi
ned as
a
two
-
tone
si
gnal
throu
g
h
the l
o
w p
a
ss filter.
As
formula (1)
descri
b
e
s
, the test syste
m
could b
e
expand
ed to
casca
ded n
on-lin
ea
r affect.
Equation (7)
and Equatio
n
(8) sho
w
det
ails:
V
t
V
t
V
t
V
t
⋯
(
7
)
V
t
b
V
t
b
V
t
b
V
t
⋯
(
8
)
And the isola
t
or is
sub
s
tituted for the l
o
w pa
ss
filter in this way in ord
e
r to av
oid the
return wave and lea
k
of t
he adja
c
e
n
t source a
nd reflect a
s
V
ρ
V
/
√
K
,
ρ
pre
s
ent
standi
ng
-
waveratio an
d K present attenuation f
a
ctor. Me
an
while
, it’s pro
v
ed that if little sign
al leak or
reflect
betwe
en the
stim
ulus source
s exist,
serio
u
s i
n
term
od
ulation di
sto
r
tion would
be
gene
rated
du
e to the ca
scaded
effect. Thus, two is
o
l
ators
are ne
ce
ssary to b
e
input after
the
high
po
wer st
imulus sou
r
ce an
d it’s effe
ctive to avoi
d
effect o
n
ret
u
rn
wave
an
d
adja
c
e
n
t sou
r
ce
leak.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 14, No. 3, September 20
16 : 831 – 838
834
Figure 3. Test System Architecture
On chann
el
cro
s
stalk i
s
ol
ation, the wi
de sp
re
ad te
ster Advant
e
s
t 93K is
sel
e
cted a
s
experim
ent in
strum
ent. Fig
u
re 4
sho
w
s the basi
c
dia
g
ram o
n
it
s
RF in
st
rum
e
n
t
card
whi
c
h
call
PSRF (RF Pi
n Scal
e). T
h
e
r
e exi
s
ts t
w
o
measurement
loop t
r
a
c
e
s
. While l
a
rge
signal lo
opb
ack
in the mea
s
u
r
eme
n
t unit whi
c
h is th
e relatively trace pitch i
s
n’t large
eno
ugh
for the stim
ulu
s
sou
r
ce, it wo
uld ge
nerate
unde
sir
ed
crosstalk
due t
o
the no
nline
a
r fa
ctors on
the inst
rume
nt
card itself. T
he main
con
c
ern rega
rdi
n
g
the gang
ed
Mini-Coax co
nne
ctor a
nd trace is i
s
olati
on
[13]. Isolation
can
be
defin
ed a
s
el
ectri
c
al noi
se
ca
used by m
u
tual
indu
ctan
ce
a
nd
capa
citan
c
e
betwe
en a
d
ja
cent
sig
nals
due to th
eir
proximity.
Crosstalk will
mainly o
c
cur due to
the
small
con
d
u
c
tor
pitch of th
e g
a
n
ged
con
n
e
c
to
r o
r
tra
c
e. A
s
sumin
g
the
crosstal
k d
epe
nds
mainly o
n
mutual indu
ct
ance, we ca
n
use the simp
le cro
s
st
alk e
s
timation a
s
it is shown in the equatio
n (9)
belo
w
, whe
r
e
D and H
rep
r
ese
n
t the cen
t
erline sepa
ra
tion and tra
c
e
height:
Cr
osstalk
dB
2
0
l
o
g
(
9
)
Therefore, chann
el
crosstal
k isol
ation
appro
a
ch is propo
se
d to do the improvement.
93K MEAS1 trac
e is
replac
ed to 93K
MEAS2 trac
e
to ac
c
e
ss
the large
s
i
gnal thru in. And the
93K MEAS1 trace i
s
changed to
LNA(Low Noise Amplifier)mo
de to detect low
power
intermo
dulati
on disto
r
tion
sign
al.
Figure 4. Cha
nnel Cro
sstal
k Isolatio
n
‐
20
dB
DU
T
IS
O
1
IS
O
2
IS
O
3
BP
F
LP
F
HV
‐
PS
G
93
K
VS
G
93
K
MEA
S
1
93
K
MEA
S
2
Co
m
b
i
n
e
r
S
p
l
i
tte
r
DC
bl
o
c
k
PC
co
n
t
r
o
l
GP
I
B
93
K
SR
C
93K
MEA
S
1
93K
ME
A
S
2
Cr
o
s
s
Ta
l
k
Ex
c
h
a
n
g
e
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
A Com
p
rehe
nsi
v
e Te
st Approa
ch o
n
Hi
gh-Po
we
r Lo
w-Noi
s
e Interm
odulation…
(Lei Wang
)
835
On sp
ect
r
um
slight offset, SFDR
(spuri
o
us
fre
e
dyna
mic Range
)
woul
d also bring som
e
effect in the low noi
se two
-
tone inte
rmo
dulati
on
sign
als produ
ct while
d
o
ing spectrum testi
n
g
[14-16]. Fo
rm
ula (1
0)-(1
1)
descr
i
be the
system dyn
a
m
ic range
wh
ich cau
s
e
d
by
system
n
on-
linear cha
r
a
c
teristi
cs.
F
represe
n
ts floor n
o
ise, B repre
s
ent
s band
wi
dth and SNR
is sign
al noise
ratio.
1
1
0
,
(10
)
2
,
(11
)
On Advante
s
t 93K, the spu
r
iou
s
si
gnal o
c
curs in ne
arl
y
45MHz freq
uen
cy point close
d
to fundament
al freque
ncy.
As Figure 5
sho
w
s, the
expected si
gn
al is cove
red
by system no
ise
floor. In
co
nseque
nce, fre
quen
cy p
o
int
slightly
shi
ft a
ppro
a
ch i
s
i
n
volved in thi
s
pap
er.
1 K
H
z i
s
shifted
with
the two-to
ne
sign
al a
nd th
e me
asure
resol
u
tion i
s
adju
s
ted to
1
Hz to
p
r
o
c
e
ss
narro
wba
nd
sampli
ng. In
this
way, t
he a
c
tual
int
e
rmo
dulation
disto
r
tion
could
be
dete
c
ted
without excee
d
ing the op
eration pa
ss b
a
nd in the mea
n
time.
Figure 5. Spectrum Slight
Offset
Figure 6. ALC Lea
k Control
Le
a
k
an
d
Reflec
t
e
d
wa
v
e
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 14, No. 3, September 20
16 : 831 – 838
836
On ALC(A
u
to
Level Control) lea
k
cont
rol,
the appro
a
ch is
setup
on the basi
s
of the
above th
re
e a
ppro
a
che
s
. T
he front-e
nd
sign
al
sou
r
ce
co
ntain
s
a
u
t
o
level
co
ntro
l unit
as Figu
re 6
sho
w
s. Seen
from th
e out
put, it co
uld
be a
pprox
imately equival
ent to a
mixe
r [17, 1
8
]. Du
e to
the nonlin
ea
r factor, if sig
nal lea
k
from
the other
so
urce or th
e
mismat
che
d
return wave
are
impacte
d, the sp
urio
us i
n
termo
dulatio
n disto
r
tion
woul
d be e
a
s
y to be g
e
nerate
d
in
sid
e
of
sou
r
ce
ca
rd.
To avoi
d t
he inte
rferen
ce, the
swit
ch i
n
fro
n
t
of the d
e
tector shoul
d b
e
discon
ne
cted
in following e
x
perien
c
e. Th
en, the pure
signal would b
e
assured.
If the above four a
pproa
ch
es is e
m
be
d
ded in
the co
re test ap
pro
a
ch a
r
chitect
u
re, the
detect a
c
cura
cy on inte
rm
odulatio
n dist
ortion
w
oul
d
be obvio
usly
improve
d
to
over 20%. A
n
d
the actu
al d
e
v
ices’
ch
aracteristic could
be
me
asure
d
.
Follo
wing pape
r would
sho
w
the
det
ail
experim
ents and
re
sult
s.
4. Experiments and
Res
u
lts
Ba
s
e
d o
n
a
n
a
l
ys
is o
f
ab
ove
fo
u
r
c
o
mp
r
e
he
ns
ive
op
timiz
e
d a
ppr
o
a
c
h
es
, th
e te
s
t
in
g
system i
s
est
ablished a
s
shown in
Figu
re 7(a
)
(b). An
d thru mo
de i
s
sel
e
cte
d
to verify the syst
em
first. From th
e results
as l
i
sted i
n
T
abl
e 1, l
a
rg
e
sig
nal trace
with
more i
s
olato
r
s could
re
du
ce
20dB floo
r n
o
ise
effective
l
y and the
a
c
cura
cy i
s
al
so in
crea
se
d
nea
rly 20%.
Ho
weve
r, small
sign
al tra
c
e
with more iso
l
ators i
s
in va
in to the floor noise. Mo
reo
v
er, small
sig
nal tra
c
e with
out
isolatio
n wo
ul
d brin
g the ot
her i
s
sue that
highe
r
order
prod
uct of n
o
n
linea
r sy
ste
m
may shift the
bias of the
de
vice du
rin
g
I
M
D2 te
st. Th
ereby, the
re
sult i
s
illogi
ca
l inverse
ch
a
nged
as sho
w
n in
Figure 8.
(a)
(b)
(c
)
Figure 7. (a)
Overview of t
he test sy
ste
m
(b)
Detail a
u
xiliary com
p
onent
s Loo
p
(c) SOI RF Switc
h
as
DUT
HV
_P
SG
93
K_VS
G
IS
O
*
2
IS
O
*
1
Co
m
b
i
n
e
r
DC
Bl
o
c
k
AT
T
DU
T
Loa
d
AT
T
Sp
l
i
tter
BP
F
9
3
K
_
M
EAS
1
9
3
K
_
M
EAS
2
PS
R
F
Ri
n
g
DC
co
n
t
r
o
l
bi
a
s
Se
r
i
a
l
St
r
u
c
t
u
r
e
Sh
u
n
t
St
r
u
c
t
u
r
e
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
A Com
p
rehe
nsi
v
e Te
st Approa
ch o
n
Hi
gh-Po
we
r Lo
w-Noi
s
e Interm
odulation…
(Lei Wang
)
837
Table 1. Isola
t
or influen
ce
on thru mo
de
No Test
loop
T
w
o-to
ne
(dBm)
IMD3
(dBm)
Compensated IM
D3
(dBm)
Variation
(%)
1
PSG+ISO1/VS
G 22/-15
-126~-12
7
-104
18.1%
2
PSG+ISO1/VS
G+ISO3
22/-15
-124.7~-1
27
-103
18.8%
3
PSG+ISO1+IS
O2/VSG+ISO3
22/-15
-150
-127
0%
4
PSG+ISO1+IS
O2/VSG 22/-15
-150
-127
0%
Figure 8. IMD2 with/witho
ut ISO compa
r
i
s
on at differe
nt Vg
Finally, singl
e
stru
cture of
SOI RF switch as
th
e ca
rri
er (Se
e
Figu
re 7(c)) is
sel
e
cted to
do the o
p
timized
app
ro
ach ve
rificati
on (Se
e
Ta
ble 2, 3). A
c
cordi
ng to
the test results
analysi
s
,the
comp
re
hen
si
ve optimize
d
approa
che
s
wh
i
c
h
conta
i
n stimulu
s
i
s
olatio
n, ch
a
nnel
cro
s
stalk i
s
ol
ation, spe
c
trum
slight offset and AL
C (Auto Level Control
)
lea
k
co
ntrol co
uld
sampl
e
the i
n
termo
dution
distortio
n
sig
nal le
ss tha
n
-150
dBm floor n
o
ise accurately. And the
test system h
a
s hig
h
re
sol
u
tion ability even at large signal conditio
n
.
Table 2. IMD3 Res
u
lts
on SOI RF Switch
Vg
T
w
o-to
ne signal
Serial Structure
Shunt Structure
input(output
)
po
w
e
r
(
dBm)
IMD3
(dBm)
Compensated IM
D3
(dBm)
IMD3
(dBm)
Compensated IM
D3
(dBm)
1V
22/-15(
7/-27
)
-83.2
-60.2
-94.3
-71.3
2V 22/-15(
21.2/-
15.7
)
-117.8
-94.8
-127.2
-104.2
3V 22/-15(
21.5/-
15.5
)
-145.3
-122.3
-150.0
-127.0
Table 3. IMD2 Res
u
lts
on SOI RF Switch
Vg
T
w
o-to
ne signal
Serial Structure
Shunt Structure
input(output
)
po
w
e
r
(
dBm)
IMD2
(dBm)
Compensated IM
D2
(dBm)
IMD2
(dBm)
Compensated IM
D2
(dBm)
1V 22/-15(
7/-27
)
-108.3
-85.3
-120.1
-97.1
2V 22/-15(
21.2/-
15.7
)
-132.2
-109.2
-145.7
-122.7
3V 22/-15(
21.5/-
15.5
)
-138.7
-115.7
-150.0
-127.0
What’
s
mo
re,
not only lo
w
noise IMD
at
high p
o
wer st
imulus
is ca
ptured, but
al
so
actual
SOI RF
de
vice
cha
r
a
c
t
e
risti
c
s a
r
e
analy
z
ed
throu
gh th
e
co
mprehe
n
s
ive o
p
timization
approa
che
s
. It is instru
ctive and mea
n
ingful fo
r d
e
vice an
alysi
s
and mo
deli
ng develo
p
m
ent.
What’
s
more, it could lay a solid fou
ndati
on on SOI fro
n
t end modul
e integratio
n.
‐
125
‐
120
‐
115
‐
110
‐
105
‐
100
‐
95
‐
90
‐
85
‐
80
‐
75
1V
2V
3V
Po
w
e
r
(
dB
m
)
Ga
t
e
Vo
l
t
a
g
e
(V
)
wi
th
IS
O
wi
th
o
u
t
IS
O
In
v
e
r
s
i
o
n
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 14, No. 3, September 20
16 : 831 – 838
838
5. Conclusio
n
Thro
ugh th
e
above
expe
riment, the i
n
crea
se
of i
s
olatio
n bet
ween diffe
rent
stimulu
s
sou
r
ces
ca
n
effectively re
duce the inte
rfere
n
ce
on t
he retu
rn
wa
ve and a
d
ja
cent lea
k
. Ch
a
nnel
cro
s
stalk isol
ation could
d
e
crea
se th
e l
a
rge
sig
nal
effect from
so
u
r
ce
to me
asu
r
eme
n
t to a l
a
rge
extent. Spect
r
um
slig
ht offset
coul
d av
oid th
e
spuri
ous si
gnal f
r
om SFDR. A
L
C l
e
a
k
cont
rol
further prevents th
e int
e
rmo
dulaiton
pro
d
u
c
t which
gen
erated in
side
the sou
r
ce
card.
Mean
while, i
n
ord
e
r to
g
e
t accurate I
M
D
sampl
e
, narro
w ba
nd
resolution
of measureme
n
t is
necessa
ry as well. What’s
more,
device also shoul
d b
e
controlled
t
o
operate in linear regio
n
or
operating bi
as
shifted
woul
d bri
n
g
illogical
pe
rforma
nce. L
a
st but le
ast, the optimize
d
comp
re
hen
si
ve appro
a
ch coul
d not onl
y reach th
e better a
c
cura
cy and mo
re
application, but
also it co
uld b
e
extended to
multi-site test for actual in
dustri
a
l engi
n
eerin
g.
Referen
ces
[1]
Munir Ac
hmad,
Ran
u
m Biru T
u
tur. Sing
le St
age
RF
ampl
ifi
e
r
w
i
th
hig
h
g
a
i
n for 2.4GHz r
e
ceiv
er front-
ends.
T
E
LKOMNIKA T
e
leco
mmu
n
icati
on C
o
mputi
ng El
ectronics a
nd Co
n
t
rol
. 2014; 1
2
(3
): 711-71
6.
[2]
Li Li
ang,
Li
u T
a
iju
n,
Ye Yan,
et al. Electroni
ca
ll
y tun
abl
e i
m
ped
ance-m
a
tchin
g
net
w
o
rks
for intelli
gen
t
RF po
w
e
r am
plifiers.
T
E
LK
OMNIKA
T
e
le
communic
a
tio
n
Co
mp
uting El
ectronics an
d Contro
l
. 20
13;
11(1
1
): 667
3-6
678.
[3]
Yu Ha
i, Z
hou
Ando
ng. T
h
ird-
order Interm
od
ulati
on Pr
obl
e
m
s in the
Sh
ort-
w
a
v
e
C
o
mmu
nicati
on.
Ship
Electron
ic Engi
neer
ing
. 2
014;
46(4): 74-
76.
[4]
YIN Haic
he
ng.
Res
earch
an
d
Ana
l
ysis
of 3r
d or
der
interm
odu
latio
n
for
R
F
Cabl
e Ass
e
mblies.
Op
t
i
c
a
l
F
i
ber & Electric
Cabl
e an
d T
heir App
licati
ons
. 2013; (5): 20-
33.
[5]
Bin W
ang, Li
jia
Chen, Hu
a Z
ong, et al.
T
he Study on Passi
ve Intermodu
la
tion Res
earch
Mode
ls
. IEEE
Internatio
na
l C
onfere
n
ce
on
Mechatro
nic S
c
ienc
es
. Electri
c
Eng
i
ne
eri
ng
and
Com
puter
(MEC). 20
13
:
309
9-31
03
[6]
Z
hen
xi
ang Y
i
, Xi
ao
pin
g
Li
a
o
. Measur
eme
n
ts on Interm
odu
latio
n
Dist
o
rtion
of Cap
a
citive P
o
w
e
r
Sensor Bas
ed
on MEMS Cant
ilever B
eam.
IEEE Sensors Journal
. 201
4; 14
(3): 621-6
22.
[7]
Hao
pen
g Z
h
e
ng, F
e
i Da
i, Don
g
li
n Su,
et al.
A mea
s
ure
m
e
n
t met
hod for a
n
te
nna ter
m
i
n
a
l
inter
m
o
dul
atio
n sensitiv
ity
. Micro
w
av
e, An
tenna, Pro
p
a
g
a
tion, a
nd EM
C T
e
chnolo
g
ie
s for W
i
reless
Commun
i
cati
o
n
s (MAPE). IE
EE 4th Internat
ion
a
l S
y
mpos
iu
m on. 2011: 5
5
1
-55
4
.
[8]
Che
n
L
iji
a, L
i
n Sh
u, Ya
ng
Ca
itian,
et
al.
An
alysis
a
nd M
eas
ure
m
ent of
Ante
nn
a 3r
d Ord
e
r
Intermodulation
.
T
he anten
na
conferenc
e. 2
009: 17
84-
178
7.
[9]
Rob
e
rt A W
i
tte. Spectrum and
Net
w
ork Me
as
ureme
n
ts. U.SPT
R prentic
e hall. 19
93: 10
5-
110.
[10]
Ye Min
g
,
Xi
ao
Yi, et a
l
. E
x
p
e
riment
al r
e
se
arch
o
n
passiv
e
int
e
rmod
u
lati
on
effect of mi
crostrip l
i
nes
.
Chin
esse Jo
ur
nal of rad
i
o sci
ence
. 20
14; 29
(3): 472-4
75.
[11]
LI
Xua
n
, Z
H
A
O
Shan
g-h
ong,
Z
HUZ
i-
xin
g
, et
al. In
fl
ue
nce
o
f
photo
d
i
ode
n
onli
n
e
a
rit
y
on t
he th
ird-or
de
r
inter-modulation distortion.
Optical Co
mmu
n
i
c
ation T
e
ch
no
l
ogy
. 201
3; (1): 44-4
6
.
[12]
W
a
rren W
J
, Palo A
l
to, Calif, He
w
l
ett WR.
An Ana
l
ysis of
the Inter
m
od
u
l
atio
n Metho
d
of Distortio
n
Measur
e
m
ent
. Vacuum El
ectronics C
onfere
n
c
e. IVEC 2004
Fifth IEEE Internatio
nal. 2
004:
176-1
77.
[13]
Z
hang Z
h
i
g
a
n
g
, Jiang W
a
nshu
n.
Influe
nce of Isolati
on betw
e
e
n
Generators o
n
3rd ord
e
r
Intermodu
latio
n
Me
asure
m
e
n
t
. Confer
enc
e
of the
n
a
tio
n
a
l
micro
w
a
v
e m
illim
eter
w
a
ve.
20
11:
146
4-
146
7.
[14]
Li Ch
en, Mar
c
Solal, Je
an
Briot.
A Nonlin
ear Mas
o
n
Model for 3
r
d Order Har
m
o
n
ic a
n
d
Intermodu
latio
n
Si
mu
latio
n
s
of SAW
Dupl
exers
. IEEE International Ultrasonics
S
y
m
posium
Procee
din
g
s. 2
012: 56-
60.
[15]
Sun T
i
quan,
Xu F
e
ng
hui, L
u
Juan, et a
l
.
Novel a
nd si
mp
lifie
d pass
i
v
e
inter
m
o
d
u
l
ati
on distorti
o
n
me
asur
e
m
ent
usin
g vector n
e
tw
ork analy
z
er
. Proceed
ing
s
of the 2014
9th IEEE Conferenc
e on
Industria
l Elect
r
onics a
nd Ap
p
lic
atio
ns, ICIEA. 2014: 16
88-1
690.
[16]
Li Min
g
tai, W
a
ng Z
unfe
ng, Li
T
ao.
T
e
st
techniq
ue for p
a
ssi
ve inter-
mo
du
l
a
tion
distortio
n
. Procee
din
g
s
of 2013 IEEE
11th Internati
ona
l
Confer
en
ce on Electro
n
ic Meas
urem
ent and Instru
ments, ICEMI.
201
3; 201
3(1): 370-
373.
[17]
He Sha
n
, Saav
edra, Car
l
os E.
Desig
n
of a lo
w
-
v
o
lta
ge an
d l
o
w
-
dist
ortion m
i
xer throu
gh vo
lterra-seri
e
s
analy
sis.
IEEE Transactio
n
s o
n
Microw
ave Theory a
nd Tec
hni
ques
. 2
013;
(61): 177-1
84.
[18]
Z
han Z
h
i
q
i
a
n
g
,
Lai
Lei.
Cal
i
b
ratio
n
for
out
put
p
o
w
e
r
an
d
PIM measur
e
m
ent acc
u
rac
y
of pass
i
ve
intermo
dul
atio
n tester.
W
I
T
T
r
ansactio
n
s o
n
Infor
m
ati
on
and
Co
mmuni
cation
T
e
chn
o
l
ogi
es
. 2
014;
(51): 116
9-1
1
8
1
.
Evaluation Warning : The document was created with Spire.PDF for Python.