Professional Documents
Culture Documents
3 - Two Methods For The Measurement of Substrate Dielectric Constant-Das-D Pozar
3 - Two Methods For The Measurement of Substrate Dielectric Constant-Das-D Pozar
3 - Two Methods For The Measurement of Substrate Dielectric Constant-Das-D Pozar
636
VOL.
MTT-35>
K. DAS, SUSANNE
the accurate
M. VODA,
and convenient
measurement
Both
methods use the precision measurementcapability of the HP-851O Net-
of the dielectric
work
Analyzer
transmission
constant
system
of a microwave
and a rigorous
substrate
theoretical
analysis
of multilayer
lines [6], and hence can also be used for the measurement
the frequency
dependence
of the relative
the
1.0 percent
can be obtained
order
are propused.
of
Measurements
were
done
for
dielectric
various
constant.
Accuracy
by use of these
substrates
and
of
on
techniques.
gave
results
as
predicted.
AND
DAVID
ment
is done.
used
for
HE
MEASUREMENT
microwave
made using
[1]-[6].
The
of the dielectric
integrated
various
method
circuit
(MIC)
constant
substrates
of
can be
resonant methods,
as described in
described in [1] and its modified
In microstrip
ing
fields
[2]-[5],
the sub-
In
constant
is determined
frequencies
and stripline
of the dipole
account empirically,
resonance
method,
by measuring
and substrate
are usually
dielectric
constant.
constant
Both
of these techniques
tions
have enough
with
them
ment.
measure
reactance
and/or
to cause significant
In addition,
suffer from
transitions,
constant
associated
in a phase measure-
the dielectric
errors intro-
error
In both
can be flexibly
constant
of any
dielec-
line is the
(or inferred
constant
by working
the effective
transmission
measured
The dielectric
then determined
program
methods,
ceff of a particular
that is actually
from
mea-
of the substrate,
backwards,
c,, is
using a computer
transmis-
to the program
is
of the substrate.
to connector
suring
methods
In both methods,
reactance/mismatch
the differences
into
duced by coax-to-rnicrostrip
together
of the dielectric
will
is canceled
in phase between
out by mea-
the fringtaken
IEEE
dimensions.
resonance techniques,
resonator
MEMBER,
deviation
of the dielectric constant from sample to sample
in a large number of substrates of approximately
the same
surement).
of
quantity
INTRODUCTION
POZAR,
measurement
substrate
industrial
tric
I.
M.
7, JULY 1987
NO
can conveniently
as it varies with
frequency
and position.
II.
THE
TWO-MICROSTRIP-LINE
METHOD
This method
is capable of accurately
measuring
the
dielectric
constant of a substrate with an accuracy of the
order of 0.5 1.0 percent, and also can be used to measure
the dielectric
constant
of a substrate as it varies with
frequency.
Besides taking into account other possible errors affecting
sures that
method
the measurement
error
is suggested
effective
dielectric
accuracy,
constant
measurement
of the standard
this method
is canceled.
technique,
enThis
of the
substrate
to
discussed in
111.
A.
of multilayer
transmission
lines [7]. Because of the nature
of the analysis, the technique can be used for the measurement of the variation
of dielectric
constant
through
a
range of frequencies
large
by changing
without
the frequency
much effort
range
over which
the experi-
Manuscript
received November 7, 1986; revised February 9, 1987. This
work was supported
in part by the National
Science Foundation
under
Grant ECS 8352325.
The authors
are with the Department
of Electrical
and Computer
Engineering,
University
of Massachusetts,
Amherst, MA 01003.
IEEE Log Number
8715005,
0018 -9480/87/0700-0636
as possible
launchers
probe
probe
accurate
results.
results, in contrast
End
to
an end-launch
connection.
The difference
between the
electrical lengths is measured using the HP-851O Network
Analyzer
system. Assuming
the four connectors
to be
identical,
the electrical length difference Al, between the
$01.00
01987 IEEE
DAS
et al.:
METHODS
FOR THE
MEASUREh4ENT
OF SUBSTRATE
DIELECTRJC
CONSTANT
637
TABLE
MEASUREMENT
lp~
OF A+
METHOD
END
FOR RT\DUROID
5880
THE
(c,
IWO-MICROSTRIP-LINE
=
2.20)
SUBSTRATE
eff (calculated)
A@(measured)
f(GHz)
[ -
USING
LAUNCH
w
k-
1.0
58.5
1.896
1.5
88.0
1.907
2.0
117.0
1.896
2.5
146.0
1.890
3.0
175.5
1.896
3.5
205.0
1.901
4.0
234.0
1.896
[ -+-----
Ipq
GRMJND
(DEGREES)
PLANE
I
Fig.
1.
Experimental
setup forthetwo-microstrip-linemethod,
the lines.
dielectric
From
constant
this,
we can determine
of the microstrip
dielectric
constant
frequency
gation
constant
of
the
line
substrate
around
analysis
solution
the
to be canceled
microstrip
identical,
transitions
and cancel
- ceff
test
propavalues of
all four
where
8 implies
Improved
measurement
be obtained
by measuring
the transfer
~,, can
phase difference
A+
the measurement
statistical
to obtain
B. Error
average
an
cr.
two
Al,
difference
in electrical
In a typical
Hence,
order
lengths
as the dielectric
the percentage
of the substrate
microstrip
lines
lengths
of the
constant
error introduced
c. of the substrate.
in determining
of
lPI and
dielectric
IP2 (the
constant
physical
c~ff of the
microstrip
line is related to the measured difference
in
electric lengths Al, and physical lengths AIP = lP1 IPZ as
(1)
c-,
:=%=2(%+%)
(3)
It can now
the dielectric
-0.1
the
error
Analysis
physical
of
be noted
constant
in the measurement
of
to the percentage
the
lengths
difference
two lines,
of these data
and
AIP
ent frequencies.
Table I shows some measured values of
calculated
C.ff. A
A+ versus f, and the corresponding
accurate
2)
error in measurement,
respectively
magnitude
termined
L
=2(%+3)
coax-to-
Al,.
through
interest.
f3&
&eff
could be used
out because
the
The normalized
the effective
and hence
= 3.54 cm in the
constant c, calcu-
cm, 8(A1J
2.2), we would
have
of measuring
AIP is limited
by the uncertainty
in the position of the
connector,
as well as the accuracy of measuring the line
lengths. Thus, end-launch connectors are found to be more
reliable than probe connectors. Also, the assumption
that
IEEE TRANSACTIONS ON MICROWAVE THEORY AND TECHNIQUES, VOL. MTT-35, NO. 7, JULY 1987
638
the four
connectors
end launchers
are identical
method,
the substrate
method
for
I
two microstrip
constant
The
fastened
of the dielectric
constant
method
overcomes
\
1310UND
PLANE
disad-
countered
in it. Instead
different
lengths,
substrates
referred
to as the standard
is accurately
referred
/gCLA:/
to as the test
-+.+
++
STRIPLINE
MICROSTRIPLINE
(a)
twoen-
S21
REF
lines of
two different
1.001
Units
200.3
mUn
it=/
hp
whose dielectric
beforehand;
substrate,
One of them is
substrate,
known
1P
MmmipLINE
over
the
the difficulties
on both
constant
This
method
is an improvement
microstrip-line
method and overcomes
TEST
of
vantages.
111.
%rmmmd
on a large
these
d-=o
w
GROUND
PLANES
of measuring
next
END LAUNCH
lines have
the dielectric
substrate.
valid
substrates
may be more
the
whose dielectric
ELECTR
is
con-
other
stant is to be determined. The standard substrate is permanently connected to the test setup, but the test substrate
can be very easily removed
be
tested,
and
measurement
hence
and replaced
can
be very
of a large number
As in the previous
method,
by another
flexible
to
of substrates.
this technique
involves
measurement
stripline
or without
with
tor mismatch
A.
one to
use for
a cover substrate),
the
,<
,1
/--/
two lines (a
and so connec-
substrate,
whose dielectric
constant
some other accurate measurement
gested
method
scribed
earlier
interested
sample
is
in
the
two-microstrip-line
Section
in the deviation
to sample
line on a standard
rather
II.
In
method,
of the dielectric
than
START
is determined
using
technique.
One sug-
constant
in the absolute
STOP
S21
one is
REF
from
dielectric
z
1.001
Units
200.3
mUn
its/
hp
substrate
must be measured
accurately.
This
substrate is connected to the HP-851O Network
system by two connectors at the ends of the 50-0
line (see Fig. 2(a)). The substrate under test (the test
substrate) is etched to remove the copper on one side, and
is cut into a rectangular
shape. The dimensions
piece need not be of any specific size, and
of the test
hence the
START
STGP
test
tom to form
2.500000000
3.500000000
GHz
i3+z
(c)
substrate
microstrip
method can be flexibly used for measurement of the dielectric constant of any size substrate. It should, however, be
greater than about two wavelengths wide, but can be of
any suitable length. The length, of course, will be constrained by the dimensions of the clamping structure (discussed below). It can be noted here that the final result is
not very sensitive to the measurement of the dimensions of
The
GHZ
GHz
(b)
de-
2. !500000000
3.500000000
is used to cover
line and is clamped
a stripline
structure.
a portion
from
of the
top to bot-
Fig.
2.
(a) Experimental
stripline
cal Smith chart plot of SZl (b) with and (c) without
obtained
from the HP-851O Network Analyzer.
method.
Typi-
DAS
et al.:
MSTHODS
used on both
sides to clamp
OF SUBSTRATE
it should
not extend
and
again microstrip
line.
The impedance
of the stripline
beyond
the standard
The entire
639
CONSTANT
pressure should
substrate
DIELECTRIC
transmission
line, stripline,
(Z,)
is different
and
from
Analyzer,
the previous
dielectric
constants
and
of the layered
of the bare
stripline
with
line
the standard
sub-
cally
in both
solely
in the connectors
dielectric
point
line
13
dJTE8T)mib
.
m
>
%
=
o
40
identi-
length
is due
of the stripline
the electrical
on the right-hand
@ 7
v
m
/
11
constant
14
any error
appears
to the effective
15
(a)
of the
microstrip
leaving
is different
effective
that
length
like
chart
10
a small
(1.0~).
is given as
8
%2 (TEST)
(4)
(b)
where
Fig. 3. Calibration
graph for (a) c.l(standard)
= 2.2, dl (standard)=
0.1575 cm= 62 roils, W= 0.5 cm, c,2(test) = 2.0-3.0, dz(test) = 10-60
1,
change in electrical
1P
physical
length
~,ff,
effective
dielectric
c~ff~
standard
effective
(standard
Then,
from
length,
roils,
and frequency = 3.0 GHz
and for (b) C,l (standard) = 10.2,
dl (standard)=
0.127 cm= 50 roils, W= 0.12 cm, C,2(test)= 5.0-15.0,
d2 (test) = 10-60 roils, and frequency=
3.0 GHz.
substrate
of the stripline
of
(with
microstrip
Now
line
from
only).
=8
8(W
(4),
)(
(5)
G=G+:.
From
the
value
of
Al,
and
1P (measured)
and
and
(b).
These
curves
of the stripline
were generated
structure
with
using
=
Of the two factors
()
iYAl,
81P
.
AI, + ~
(7)
while
a full-wave
two dielectric
slabs
NP.AI,
+ 12
(theory),
C.ff, can be calculated from (5), and hence the
unknown
c ,2(test) can be determined
from ~~ff,, c,l(standard), W, d2(test), and dl(standard)
by calculation
[7] or
by using a set of calibration
graphs, as shown in Fig. 3(a)
analysis
81,
Al,
1P
or
&
=eff.
Cefff
z 8G
6Ale
(9)
c effs
= 2 Ip. ceff,
Analysis
= o.
(6)
640
IEEE TRANSACTIONS
However,
the contribution
rough
effs
estimate
-1/2&,2
TECHNIQUES,
TABLE
VOL.
MTT-35,
NO. 7, JULY
1987
II
can be checked
Substrate
tested
(lo)
(test).
from
Fig.
3.(a) and
(b). Finally,
6,2 (test)
example,
if
iheff~
=2=
~effs
1P-5.0
cm,
Ceff,
ceff~ -2.0,
method
used
measured
r
62mils
2.2
2.189
ST/DUROID
2.19
31mils
2.2
50mils
10.2
10.80
60m11s
2.55
11
2.54
(11)
588o
RT/DUROID
6DI0.2
OAK
OAK
30m11s
2.55
11
2.53
8A1, - ().1
RT/DUROID
588o
KT/OUROIO
62mi1s
2.2
11
2.2D
20m11s
2.2
11
2.19
50mils
10.2
II
10.15
and
E?
5880
8AI,
4
lP.
manufacture!,
thickness
RT/DUROID
iic,z(test)
For
AND
&
This
THEORY
ON MICROWAVE
mm, we have
588o
RT /DUROI O
6010.2
&,2(test)
(12)
= 0.4 percent.
c,z(test)
C. Discussion
ent.
A few important
points
can be mentioned
regarding
the
The
calibration
standard
substrate.
substrates
graphs
Similar
were obtained
for
differ-
a specific
can
above measurement
technique.
i) One side of the test substrate can quickly be etched
free of copper and made ready for test. The size of the
be obtained.
ix) The variation
of the dielectric
constant
over a
frequency
range can be easily measured. This is done by
ordinary
ii)
small
is good enough).
method
measures
can
the dielectric
dielectric
just
ruler
This
dimensions
and
(accuracy of an
by
constant
shifting
from
it
constant
None
of
ently
the variation
sidewise
of a
or turning
it in
of substrates
types
of substrates
of approximately
of different
the
dielectric
to
of
con-
thicknesses.
introduced
in the measurement
of dimensions
of the substrate
because of
is negligible
the standard
is avoiding
the
IV.
of the standard
to minimize
the temperature
uncertainties
larly, better
which
RESULTS
methods
accurately.
can so conveni-
AND
CONCLUSIONS
substrates
from
have
sample
single
a large
variation
piece of substrate.
samples
of
dielectric
constant
of these substrates
constants
in different
from
as
method,
measurement
of the dielectric
constant
quick,
reliable
of microwave
sub-
way.
APPENDIX
&l
OF A CASCADE
introduces
(not error) in the measurement process. Simiresults will be obtained if the thicknesses of
OF THREE
TRANSMISSION
LINE
SECTIONS
for very
dependence
reported
and this
frequency
of this
a different
same dielectric
constant,
the deviation
from sample
sample can quickly be determined using this method.
iv) A fixed test setup can be used for a number
different
care
of the previously
on a sample
direction.
iii) In a large number
take
A signal
of unit
amplitude
incident
on port
1 (see Fig.
4(a)) on the forward path reaches plane pz with an amplitude Te-JfO+olJ. Part of it gets transmitted
to the third
section and to port 2 ( = TT. eJ(e +I +2) ); the other part
gets reflected
and undergoes
reflection
partly
at plane
delay equivalent
to 28 and a
gets transmitted
to the third
section
and to port
2,
DAS
et al.:
METHODS
OF SUBSTRATE
CONSTANT
J-+
-r+xi
[2]
5
d
d.~
DIELECTRIC
mm 2
mm q
Zm
z~
50.0
7:
SO.fl
MICROSTRIP
=RlpLiNE
[3]
.r
Zm
MICROSTRIP
641
L. S. Napoli and
determination
of
integrated
circuit
pp. 664-665,
July
[4]
Zm
Theoiy
\
\
\/
/\
\
\
/
-.\
/,
\/
\
,\
/
\:\
f
\
/\\
.
\\
\
(
)/
(
[5]
(a)
1.0
[6]
X%-%
,/
substrates:
IEEE Trans.
383-386,
Apr. 1981.
[7]
[8]
,,
..
..
s-
._~5.
;&...____.._~-.:- -:Z-.
----
______
I
\
//
,,
\,-/
\/
/~.
-1.0
TT
(b)
Fig. 4. (a) Cascade of three transmission
lines, i.e., microstrip
line,
stripline,
and microstrip
line. (b) SZl of the cascade for smafl mismatch.
[9].
Thus,
gets reflected
the generalized
Nirod
K. Das and D. M. Pozar, A generalized
spectral-domain
Greens function
for muftilayer
dielectric
substrates with applications to mnkilayer
transmission
lines.
IEEE
Trans. Microwave
Theory Tech., vol. MTT-35, pp. 326-335, Mar. 1987.
S. Hubbel
and D. J. Angelakos,
A technique
for measuring
the
effective
dielectric
constant
of a rnicroswip
line,
IEEE
Tram.
Microwave
Theoty Tech., vol. MTT-31, pp. 687-688, Aug. 1983.
R. E. Collin,
Foundations
of Microwave
Engineering.
New York:
McGraw-Hill,
ch, 5.8.
and partly
pp.
,1,
~:
.-.
Microwave
;2
1
//
A\
7/
\/
,\
[9]
<
1973.
Y!
Mar.
P. H. Ladbrooke,
M. H. N. Potok, and E. H. England, Comments
on A quick accurate method to measure the dielectric constant of
microwave
integrated
circuit substrates,
IEEE
Trans. Microwave
Theoiy Tech., pp. 570-5712, Aug. 1973.
P. H. Ladbrooke,
M. H. N. Potok, and E, H. England,
Coupling
error in cavity resonance measurement
of MIC dielectrics,
IEEE
Trans. Microwave
Theory Tech., pp. 560-561, Aug. 1973.
R. M. Prrnnel and B. W. Jervk, Two
simple methods
for the
measurement
of the dielectric
permittivi~y
of low-loss microstrip
expression
and so on
as
Szl =
7777~-X@+%+@2)(l+
r2e-2j0
1 _
rq~-qjo
(Al)
between
the microstrip
Q, or Irl <1/3,
and stripline
lr12 <0.1,
small
Thus,
r,
we have T=l+
S21 looks
(Al)
as
. . . )
+6,)
TT1e-j(6+e,
rz~-zjo
r,
and T=1
toward
r;
TT =1.
the right-hand
[1] ASTM
She joined
the Raytheon
Company,
Missile
Systems Division, in 19S6, where she is currently
working on numericaf antenna analysis.
(A2)
delay equivalent
David M. Pozar (S74 M80) was born in Pittsburgh, PA, on January 25, 1952. He received the
B.S. degree in 1975 and the M.S. degree in 1976,
both in electrical engineering,
from the Universit y of Akron. He received the Ph.D. degree in
electrical engineering from Ohio State University
in 1980.
During
the course clf his undergraduate
degree, he spent one year as an Engineering
Assistant at the National
Security
Agency,
Fort
Meade, MD. He was a Graduate
Research As-
642
IEEE TRANSACTIONS
Engineers
(IEEE).
He is also an Associate
Member
of the
ON MICROWAVE
THEORY
AND
TECHNIQUES,
VOL.
International
Union
of Radio Science (URSI),
Commission
B. He has
served as an Associate Editor of the IEEE TRANSACTIONS ON ANTENNAS
AND PROPAGATION and as an Associate Editor of the IEEE ANTENNAS
AND PROPAGATION SOCIETY NEWSLETTER. In 1984, he received an NSF
Presidential
Young
Investigator
Award,
as well as the Keys
to the
Future
Award,
from the IEEE Antennas
and Propagation
Society. In
1986, he received the R.W.P. King Best Paper Award from the IEEE
Antennas
and Propagation
Society.