Professional Documents
Culture Documents
2008 Yh Ieee Ap
2008 Yh Ieee Ap
Ying Hu
David R. Jackson
Jeffery T. Williams
Stuart A. Long
Varada Rajan Komanduri
ABSTRACT
The input impedance and radiation pattern of the inset-fed rectangular microstrip patch
antenna are investigated. It is concluded that a shifted cosine-squared function describes
well the variation of the resonant input resistance with the feed location (point of contact
between the microstrip feed line and the notch) for a given patch and substrate geometry.
The parameters of the shifted cosine-squared function depend on the notch width, for a
given patch and substrate geometry. The cross-polarization level is seen to increase as
either the notch width or depth increases. However, for a fixed input resistance level, the
notch depth decreases as the width increases, so that the cross-polarization level is not
very sensitive to the notch width for a fixed input resistance.
1
I. Introduction
The location of the contact between a feeding transmission line and a microstrip patch
antenna is commonly used for impedance control. The inset-feed technique integrates a
patch antenna together with a microstrip feed on a single planar substrate. The input
impedance of the inset-fed patch varies, similarly to the coaxial probe-fed patch, as the
For a coaxial probe-fed rectangular patch antenna shown in Fig. 1(a), customary analysis
methods use a transmission line model or a cavity model to calculate the input impedance
[1]-[9]. More rigorous numerical approaches, such as the method of moments are also
commonly used [10]. Calculations show that the resonant input resistance of the probe-
fed rectangular patch decreases as the feed approaches the center of the patch, and as the
width of the patch increases [10]. Furthermore, the dependence of the resonant input-
squared function,
"!x #
Rin = A cos 2 $ n % (1)
& 2 '
where xn = 2 x / L (see Fig. 1(a)). The cosine-squared formula has been confirmed by
The impedance behavior of the inset-fed rectangular patch shown in Fig. 1(b) has not
been systematically investigated to the same extent as the probe-fed patch, although the
same cosine-squared function is often used to design such patch geometries. Some earlier
2
measurements pointed out that the dependence of the input resistance on feed position
demonstrated that the input impedance of an inset-fed patch dropped more rapidly than a
cosine-squared function as the feed position moves away from the edge of the patch [11-
13].
The purpose of this paper is to more fully explore and characterize the behavior of the
resonant input resistance, which takes the form of a “shifted” cosine-squared function.
The effects of the inset-feed on the radiation pattern are also considered, and a study of
notch widths S, line widths Wf, and inset depths xf. The input impedance was obtained by
de-embedding the calculated input impedance at the end of the microstrip line (M-M in
Fig. 1(b)) to the contact point between the feed line and the radiating patch (C-C in Fig.
1(b)). The resonance frequency was chosen as the frequency that maximized the input
resistance at the contact point (therefore there will be a small input reactance at the
used inside the notch region. That is, the interaction between the feed line and the
surrounding patch in the notch region was assumed to be negligible, so that the feed line
within the notch region acts as an isolated microstrip transmission line. This assumption
3
was verified for the gap widths assumed here (results are omitted), where the ratio of slot
width S to the line width Wf was greater than or equal to 2. (Even for those situations
where the gap width is small enough so that line-patch interactions are not negligible, one
may still assume an isolated microstrip line in the de-embedding procedure as a way to
artificially define the input impedance at the contact point, though the impedance so
defined will no longer approximately agree with that measured by directly feeding the
It was observed that the inset feed causes the voltage minimum to shift away from the
center of the patch (the location predicted by Eq. (1) for the probe-fed patch), and move
towards the edge of the patch. Another observed effect of the inset feed is that the slope
of the curve describing the voltage versus feed position is no longer zero at the edge of
the patch, as predicted by Eq. (1). It was found that both effects are captured to a fair
"! #
Rin = A cos 2 % (xn $ B )& . (2)
'2 (
In general, the coefficients A and B depend on the notch width S, the aspect ratio W/L of
the patch, and the substrate permittivity εr and thickness h. The coefficients are
approximately independent of the line width Wf. In the next section results will be given
to verify the accuracy of Eq. (2), and values for the coefficients A and B will be given for
varying substrate εr and h, for a commonly used patch aspect ratio W/L = 1.5.
4
III. Results for Input Resistance
Three patch geometries with different dielectric substrates are studied to characterize the
inset-fed patch. Table 1 records the width and length of the patch for each substrate.
These dimensions are scaled to produce a common resonant frequency of about 2.3 GHz
(the exact resonance frequency depends on the notch dimensions). The substrate
Figure 2 first validates the simulation for the inset-fed patch by comparing results for the
de-embedded resonant input resistance at the contact point from two simulation packages
(Ansoft HFSSTM, which uses a finite element analysis, and Ansoft Designer®, which is
based on the method of moments) with experimental measurements [11] on the patch
with εr = 2.42. For this result S = 1.14 cm and Wf = 0.38 cm (corresponding to Z0 = 50 Ω).
(In all simulations the conductivity of the copper patch and ground plane was taken as 5.8
× 107 S/m while the loss tangent of the substrate was neglected.) The resonant input
resistance is determined for each notch depth by finding the frequency at which the de-
embedded resistance at the contact point is a maximum. (This frequency varied slightly
as a function of the notch depth.) The de-embedded input impedance at the contact point
is slightly inductive due to the fact that the line/patch junction has some inductance
Results (omitted) were then obtained for varying line widths. For the air-filled patch
results were obtained using lines that have characteristic impedances of 50 Ω and 100 Ω.
For the patch with εr = 2.42, results were obtained using feed lines that have
5
characteristic impedances of 50 Ω and 25 Ω. It was observed that the de-embedded input
Given that the line width does not significantly affect the resonant input resistance, a
standard 50 Ω line is used to feed an inset-fed patch with εr = 2.42 and different notch
widths. Figure 3 shows the Rin vs. xn curve for each notch width, along with the results of
CAD formula (2). The coefficients A and B were determined by a least-squares fitting
procedure (details omitted). It is seen that the CAD formula predicts the resonant input
resistance fairly well. The agreement is also observed to hold for different substrate
Table 2 summarizes the values of the coefficients A and B for the three different
permittivities indicated in Table 1, where all of the patches have the same substrate
thickness of 0.127 cm and the same (typical) aspect ratio of 1.5 (see Table 1), and all are
fed by a 50 Ω feed line. The parameters A and B are observed to depend on the substrate
permittivity.
Table 3 shows values for the coefficients A and B for the three permittivities used
previously (1.0, 2.42, and 10.2), for two different substrate thicknesses (the one used in
Table 2 as well as one that is twice as thick, h = 0.254 cm), for a fixed ratio of notch
width to line width of 3 (the value used most commonly in the literature). The line widths
were chosen to be the same as that for the S/Wf = 3 case in Table 2, corresponding to a 50
Ω feed line for the thinner substrate. Note that this corresponds to using a different notch
6
width for each different permittivity. For the thicker substrate the impedance of the feed
lines is higher than 50 Ω, but the de-embedded input resistance is essentially independent
of the line width, as noted previously. The patch dimensions are the same for both
substrate thicknesses, and are given by Table 1. It is observed from Table 3 that the
coefficients do depend on the substrate thickness to some extent. For the high permittivity
case (εr = 10.2) it was observed that when the substrate thickness was doubled the CAD
formula was significantly less accurate than in all of the other cases, presumably due to
the stronger interaction between the line and the notch, attributable to the narrower notch
in the high permittivity case, and the greater spread of the fringing fields, due to the
thicker substrate. Hence, results for this case are omitted in Table 3. The results show that
the A coefficient increases with increasing substrate permittivity, while the trend for the B
Table 4 shows values for the coefficients A and B for four permittivities: the three used
previously in Table 2, along with εr = 6.15. The substrate thickness is the same as for the
thicker substrate used in Table 3 (h = 0.254 cm). For this set of results a fixed notch
width of 1.14 cm was used (which is the same notch width used for the εr = 2.42 case in
Table 2). The line widths were again chosen to be the same as that for the S/Wf = 3 case
in Table 2, corresponding to a 50 Ω feed line for the thinner substrate. For the εr = 6.15
case the line width was 0.187 cm. The patch dimensions are again the same for both
substrate thicknesses, and are given by Table 1 for the low, medium, and high
permittivities. For the εr = 6.15 case the patch dimensions are W = 3.90 cm and L = 2.55
cm. The ratio of notch width to line width now varies from about 2 for the air substrate
7
case to about 9 for the εr = 10.2 case. In all cases the notch width is wide enough so that
there is little interaction between the line and the notch, even for this thicker substrate
case. These results show that the A coefficient again increases as the substrate
permittivity increases. The trend with the B coefficient is again not completely clear, but
for this case of a fixed notch width it remains approximately constant as the substrate
permittivity increases.
Figure 4 shows a comparison between the exact resonant input resistance and the results
from CAD formula (2) for the three permittivities εr = 1.0, 2.42, and 10.2, for the thicker
substrate case (h = 0.254 cm) in Table 4. For each permittivity, the CAD formula uses the
coefficients A and B from Table 4. The agreement is seen to be very good, although there
is a noticeable error when the contact point is near the edge of the patch for the high
permittivity case.
Figure 3 shows that for a particular input resistance, a wider notch requires a shallower
inset depth. Increasing either the notch width or depth results in a greater disturbance of
the patch currents relative to that of the rectangular patch, increasing the cross-
polarization in the H plane (there will be no cross polarization in the E plane regardless of
the notch dimensions, due to symmetry). In a previous work, piecewise sinusoidal basis
pulse functions were used in the vicinity of the contact point between a semi-infinite
microstrip line and the patch surface for an inset-fed patch [14]. A large current
8
fluctuation was found at the contact point [15], which would be expected to result in
cross-polarization.
In this paper, the effects of the notch width and depth on the radiation pattern are
investigated. First, simulations are performed for a fixed notch depth and a changing
notch width, and then for a fixed notch width and changing notch depth. For this study,
the patch has a substrate dielectric constant of 2.42. Figures 5 (a) and (b) show the notch
width and depth effects, respectively. As expected, the cross-polarization level increases
as the notch gets wider or deeper. However, for a fixed input resistance level, the
competing factors of notch width and depth compensate to maintain a somewhat constant
cross-polarization level. At xn = 0.875, the cross-pol level in Figure 5(a) exceeds the co-
pol level. This corresponds to the feed point where Rin approaches zero in Figure 3. At
this point the patch essentially loses its main resonance and performs very poorly.
A patch with a substrate dielectric constant of 2.42 with three different notch widths, S =
0.76 cm, 1.14 cm, and 1.52 cm, was then constructed in HFSSTM. A 50 Ω line fed all
patches. The notch depth was tailored to match the patch to the line impedance for each
notch width. The patch with the narrowest notch width was found to have the smallest
cross-polarization, although the cross polarization was not very sensitive to the notch
width, due to the competing effects of notch width and depth mentioned previously.
9
V. Conclusions
The input impedance behavior of the inset-fed rectangular microstrip patch antenna has
been investigated. The de-embedded resonant input resistance of the patch at the contact
point between the line and the patch versus the notch depth (location of the contact point)
independent of the width of the microstrip feed line. However, they depend on the
substrate permittivity and thickness, as well as the notch width and the aspect ratio of the
patch. Tables were presented to give the values of these coefficients for a commonly used
aspect ratio W/L = 1.5 and the most commonly used ratio of notch width to line width of
S/Wf = 3, for three different permittivities and two different substrate thicknesses.
The radiation pattern of the inset-fed patch was also studied with the main focus on its
cross-polarization level in the H-plane. Increasing either the notch width or depth
increases the cross-polarization level. However, for a fixed input resistance, an increase
in the notch width requires a decrease in the notch depth. Therefore, for a fixed input
resistance, the cross-polarization level is not very sensitive to the notch width.
References
[1] J. R. James, P. S. Hall, and C. Wood, Microstrip Antenna Theory and Design, Peter
[2] J. R. James and P. S. Hall, The Handbook of Microstrip Antennas (two volume set),
10
[3] D. M. Pozar and D. H. Schaubert, Editors, Microstrip Antennas: The Analysis and
[4] P. Bhartia, Millimeter-Wave Microstrip and Printed Circuit Antennas, Artech House,
1991.
[5] K. F. Lee, Editor, Advances in Microstrip and Printed Antennas, John Wiley, 1997.
[6] R. Garg, P. Bhartia, I. J. Bahl, and A. Ittipiboon, Editors, Microstrip Antenna Design
[8] R. Bancroft, Microstrip and Printed Antenna Design, Noble Publishers, 2004.
[9] C. A. Balanis, Antenna Theory: Analysis and Design, 3rd Ed., Wiley, 2005 (pp.727-
752).
antennas”, IEEE Trans. Antennas Propag., Vol. 30, no. 6, pp. 1191 – 1196, Nov.
1982.
input impedance on feed position of probe and microstrip line-fed patch antennas”,
IEEE Trans. Antennas Propag., vol. 49, pp. 45-47, Jan., 2001.
[12] R. Zhong, X. Tang, L. Wang, and X. Zhang, “Study of microstrip-line inset-fed and
11
[13] T. Samaras, A. Kouloglou, J. N. Sahalos, “A note on the impedance variation with
[14] G. Ghione and C. Naldi, “Parameters of coplanar waveguides with lower ground
12
List of Tables
Table 2. A and B coefficients for common inset-fed patch geometries with various
dielectric substrates (εr = 1.00, 2.42 and 10.2) and a 50Ω feed line. The units of A are
Table 3. A and B coefficients for common inset-fed patch geometries with three different
dielectric substrates (εr = 1.00, 2.42, and 10.2) and S/Wf = 3. The patch geometry is given
in Table 1 and the feed line has the width shown in Table 2. Results are shown for h =
0.127 cm (left value) and h = 0.254 cm (right value). The units of A are Ohms and the
Table 4. A and B coefficients for inset-fed patch geometries with various dielectric
substrates (εr = 1.00, 2.42, 6.15, and 10.2) and S = 1.14 cm. The patch geometry is given
in Table 1 for εr = 1.00, 2.42, and 10.2. For εr = 6.15 the dimensions are W = 3.90 cm and
L = 2.55 cm. The feed line has the width shown in Table 2 for εr = 1.00, 2.42, and 10.2.
For εr = 6.15 the width of the feed line is 0.187 cm. The substrate thickness is h = 0.254
cm. The units of A are Ohms and the units of B are radians.
13
List of Figures
Figure 1. (a). Coaxial probe-fed rectangular patch. (b) Inset-fed rectangular patch.
0.127 cm.
Figure 3. The de-embedded resonant input resistance versus the normalized feed position,
for an inset-fed patch with three different notch widths. εr = 2.42, Wf = 0.38 cm. Results
are obtained from a moment-method simulation, using Ansoft Designer®. The discrete
data points are the results from the moment-method simulation, and the solid curves are
the results from the proposed CAD formula, using the values of A and B in Table 2.
Figure 4. The de-embedded resonant input resistance versus the normalized feed position,
for an inset-fed patch with S/Wf = 3. The values of the line width Wf are shown in Table
2. The substrate permittivity is εr = 2.42 and substrate thickness is h = 0.254 cm. The
patch dimensions are those shown in Table 1. Results are obtained from a moment-
method simulation, using Ansoft Designer®. The discrete data points are the results from
the moment-method simulation, and the curves are the results from the proposed CAD
Figure 5. Calculated H-plane radiation patterns of an inset-fed patch with various notch
depths and widths, with εr = 2.42 and a substrate thickness of h = 0.127 cm. The patch
14
dimensions are shown in Table 1. Results are obtained from a finite element simulation,
(a) The notch width is fixed at S = 1.14 cm and the notch depth is varied. The curves
correspond to different notch depths and polarizations as follows: i: xn = 0.126, cross-
polarization, ii: xn = 0.625, cross-polarization, iii: xn = 0.875, cross-polarization, iv: xn =
1.0, cross-polarization, v: xn = 0.875, co-polarization, vi: xn = 0.126, 0.625, and 1.0, co-
polarization.
(b) The notch depth is fixed at xf = 1.01 cm and the notch width is varied. The curves
correspond to different notch widths and polarizations as follows: i: S = 0.57 cm, cross-
polarization, ii: S = 1.90 cm, cross-polarization, iii: S = 2.28 cm, cross-polarization, iv: S
= 3.04 cm, cross-polarization, v: co-polarization for all.
15
Table 1
εr W (cm) L (cm)
1 8.91 6.06
2.42 5.94 4.04
10.2 2.97 2.02
Table 2
Table 3
εr A (Ω) B (rad)
1.00 194 | 177 -0.27 | -0.24
2.42 197 | 188 -0.21 | -0.12
10.2 318 | NA -0.03 | NA
Table 4
εr A (Ω) B (rad)
1.00 165 -0.17
2.42 188 -0.12
6.15 290 -0.16
10.2 385 -0.15
16
W W
L L
S
C C
xf xf
g
Wf
M M
17
Figure 2
Figure 3
18
Figure 4
19
Figure 5(a)
Figure 5(b)
20