PIERS Proceedings, Cambridge, USA, July 5–8, 2010
212
An Effective Strategy for Designing Probe-fed Linearly-polarized
Thick Microstrip Arrays with Symmetrical Return Loss
Bandwidth
Laila F. Marzall1 , Daniel C. Nascimento2 , R. Schildberg2 , and J. C. da S. Lacava2
2
1
TSM Antennas, Brazil
Laboratório de Antenas e Propagação, Instituto Tecnológico de Aeronáutica, Brazil
Abstract— A new strategy for designing probe-fed linearly-polarized thick microstrip antennas
is proposed in this work. Symmetrical return loss bandwidth is obtained by designing the patch
to operate at the zero input reactance condition (Xin = 0). Applications for a single microstrip
element and a linear array of suspended ones are presented. Experimental and simulated results
validate the proposed strategy.
1. INTRODUCTION
When the standard design procedure is applied to probe-fed linearly-polarized (LP) microstrip
radiator with a thick substrate — which is a practical way to increase bandwidth — the antenna
exhibits an inductive input reactance that is difficult to match to the feeder network, and affects
significantly its bandwidth symmetry [1, 2]. Many techniques are available for compensating the
probe’s inductance, but their use depends largely on the expertise of the antenna designer. Series
capacitors, which can be implemented in several ways, are a possibility [3]. Other techniques, like
changing the probe’s geometry, can be utilized [4]. However, if such modified elements are used
in a linear array, the complexities of its feeder network will likely increase, and, consequently, the
difficulty in designing it. To overcome this limitation, a new strategy for designing probe-fed LP
thick microstrip antennas is presented. The main outcome of this new proposal is the development
of arrays with symmetrical bandwidth, resulting therefore in simpler feeder network designs. First,
the new strategy is applied to the design of a single element. Subsequently, it is used for designing a
WLAN array of suspended microstrip antennas to comply with the IEEE 802.11b (2.4–2.4835 GHz)
protocol and the ETSI EN 301 525 V1.1.1 (2000-06) standard [5]. Experimental results validate
the proposed strategy.
2. SINGLE ELEMENT: COMPARISON BETWEEN THE STANDARD AND THE NEW
DESIGN
The typical geometry of a probe-fed LP rectangular patch microstrip antenna is presented in Fig. 1,
where `p denotes the patch length, wp the radiating width, yp the probe position along the y-axis,
and h the substrate thickness. In this case, the antenna is directly fed by a 50-Ω SMA connector.
According to the standard procedure for designing a LP patch in the fundamental mode TM10 , the
operating frequency is set up at the maximum input resistance point.
Following this procedure and using the commercial software HFSS [6] for optimizing the radiator
dimensions, a rectangular antenna consisting of a h = 6 mm thick, air (εr = 1) substrate, fed
by a 1.3-mm diameter coaxial probe, was designed to operate at 2.45 GHz. Using a rectangular
ground plane (125 mm×100 mm), the following optimized dimensions were obtained: `p = 50.4 mm,
wp = 65.5 mm and yp = 15.75 mm. Results for the input impedance and the reflection coefficient
magnitude (|Γ|) are presented in Fig. 2(a). As shown, the maximum input resistance occurs at
the operating frequency (2.45 GHz). As a result, the antenna input impedance is highly inductive
(Zin = 50 + j66Ω, at 2.45 GHz) and can not be perfectly matched to a 50-Ω SMA connector
(|Γ| = −5.5 dB) without an external network. Hence, the radiator bandwidth is asymmetrical in
relation to the operating frequency. Thus, a new approach for designing probe-fed thick microstrip
antennas is proposed next. Differently from the first approach, it consists of designing the patch
to operate at the zero input reactance condition (Xin = 0). Using this new criterion, the antenna
is redesigned to operate at 2.45 GHz for the same air-layer thickness, ground plane dimensions and
probe diameter used in the standard design. The new optimized dimensions are: `p = 55.0 mm,
wp = 71.5 mm, and yp = 2.0 mm. Results for the input impedance and the reflection coefficient
magnitude of the new antenna are shown in Fig. 2(b). Now, the antenna matches perfectly the
50-Ω SMA connector and presents a symmetrical bandwidth in relation to the operating frequency
Progress In Electromagnetics Research Symposium Proceedings, Cambridge, USA, July 5–8, 2010
Rectangular patch
Feed point
wp /2
y
yp
x
213
wp
lp
h
Ground plane
Figure 1: Linearly polarized probe-fed microstrip antenna.
0
-1
190
-6
75
-2
150
60
-3
45
-4
30
15
0
2.1
2.2
2.3
2.4
2.5
2.6
110
-18
70
-24
-5
30
-30
-6
-10
-36
-50
2.1
-7
2.8
2.7
-12
Re[Zin]
Im[Zin]
|Γ|
2.2
Frequency [GHz]
2.3
2.4
2.5
2.6
2.7
|Γ| [dB]
Impedance [Ω]
90
Impedance [Ω]
230
Re[Zin]
Im[Zin]
|Γ|
|Γ| [dB]
0
105
-42
2.8
Frequency [GHz]
(a)
(b)
Figure 2: Input impedance and reflection coefficient magnitude: (a) standard design, (b) new design.
y
z
wr
wp
Patch substrate
hs1
hgap
2.5 mm
lr
lp
x
yp
Reflector
hr
x
4.5 mm
hs2
BF substrate
(a)
(b)
Figure 3: Suspended microstrip antenna geometry: (a) top view, (b) side view.
(2.45 GHz). In addition, as observed in Fig. 2(b), the maximum resistance is greater than 50 Ω and
occurs at a frequency below 2.45 GHz.
3. SUSPENDED MICROSTRIP ANTENNA DESIGN
The project of a suspended microstrip radiator, based on IEEE 802.11b (2.4–2.4835 GHz) protocol
and ETSI EN 301 525 V1.1.1 (2000-06) standard, is discussed next as another application of the
proposed strategy.
The typical multilayer structure is shown in Fig. 3: the rectangular patch is printed on a
dielectric substrate of thickness hs1 and an air gap of thickness hgap separates it from the ground
plane.
PIERS Proceedings, Cambridge, USA, July 5–8, 2010
214
In this particular configuration, the patch is directly fed by a 50-Ω coaxial probe at x = 0
and y = −(`p /2 − yp ). The multilayer structure under consideration utilizes three types of PCB
laminates: Arlon CuCLad 250GX (hs1 = 1.524 mm) for the patch (wp = 58.7, `p = 46.1 and
yp = 15.8 mm, based on the zero input reactance criterion), Arlon AR450 (hs2 = 1.1938 mm) for
the array beam forming (BF) network, and FR-4 for the reflector (wr = 100 mm, `r = 100 mm and
hr = 1.6 mm). Furthermore, the element must operate at 2.45 GHz and cover the aforementioned
bandwidth. For optimizing the element BW, the air gap spacing was set at 2.5 mm, resulting in a
bandwidth of 4.8% at 10-dB return loss condition (Fig. 4).
4. LOW-CROSS-POLARIZATION SUSPENDED PATCH LINEAR ARRAY
Using the suspended microstrip element previously designed, a low-cross-polarization WLAN array
that meets the ETSI standard minimum gain of 14 dBi is proposed [7]. To comply with the gain
and cross-polarization requirements, an eight-element array, divided in two subgroups with 180◦
phase shift excitation, was designed. To control the side lobe level (SLL), the Dolph-Chebyshev
method was used to determine the excitation coefficients for a −20 dB SLL. The array topology is
presented in Fig. 5 (wr = 200 mm, `r = 800 mm, ws = 100 mm, `s = 720 mm, and d = 0.75λ0 ) and
its impedance matrix is given in Table 1. As seen, there is low mutual coupling between the array
elements, thus simplifying the feeder network design.
Hence, the individual reflection coefficient magnitudes of the eight elements of the array present
a similar behavior, as shown in Fig. 6.
Based on this fact, the feeder network topology could be implemented in a simple way, as shown
in Fig. 7. Just four T-junction power dividers were required: two on the first level (1N1 and 1N2),
one on the second level (2N) and the last on the third level (3N), where a microstrip line was
used to implement the 180◦ phase shift between the two array subgroups. The elements’ input
0
BW = 4.8%
-5
|Γ| [dB]
-10
-15
-20
-25
-30
-35
2.350
2.375
2.400
2.425
2.450
2.475
2.500
2.525
2.550
Frequency [GHz]
Figure 4: Reflection coefficient magnitude: Note the symmetry of the bandwidth in relation to the operating
frequency.
Table 1: Impedance matrix (Ω) of the array elements simulated by HFSS.
N
4
Z4N
Z3N
53.3 + 7.4i −4.8 − 5.8i
Z2N
Z1N
Z5N
Z6N
Z7N
Z8N
2.2 + 0.1i
−0.4 + 0.4i
8.9 + 5.4i
−2.9 + 0.8i
0.4 − 0.7i
0.1 + 0.2i
3 −4.8 − 5.8i
52.7 + 6.2i
−4.2 − 5.7i
2.1 + 0.1i
−3.0 + 0.9i
0.4 − 0.7i
0
0
2
−4.2 − 5.7i
51.8 + 5.8i
−4.3 − 5.7i
0.4 − 0.7i
0
0
0
2.2 + 0.1i
1 −0.4 + 0.4i
5
8.9 + 5.4i
6 −2.9 + 0.8i
2.1 + 0.1i
−4.3 − 5.7i
52.7 + 5.9i
0
0
0
0
−3.0 + 0.9i
0.4 − 0.7i
0
54.1 + 7.6i
−4.8 − 5.8i
2.2
−0.4 + 0.4i
0.4 − 0.7i
0
0
−4.8 − 5.8i
52.4 + 6.1i
−4.2 − 5.7i
2.1 + 0.1i
7
0.4 − 0.7i
0
0
0
2.2
−4.2 − 5.7i
52.2 + 6.0i
−4.3 − 5.7i
8
0.1 + 0.2i
0
0
0
−0.4 + 0.4i
2.1 + 0.1i
−4.3 − 5.7i
52.7 + 5.9i
Progress In Electromagnetics Research Symposium Proceedings, Cambridge, USA, July 5–8, 2010
215
impedance and respective Dolph-Chebyshev excitation coefficients for the required SLL level are
shown in Table 2. The feed network was placed behind the reflector to avoid spurious radiation.
The final array geometry is depicted in Fig. 8.
Experimental and simulated results for the reflection coefficient magnitude are shown in Fig. 9.
As expected, symmetrical return loss BW was established.
lr
ls
wr
y
ws
d
x
Figure 5: Array topology: two subgroups with 180◦ phase shift excitation.
0
-5
-10
|Γ| [dB]
-15
-20
Element 1
Element 2
Element 3
Element 4
Element 5
Element 6
Element 7
Element 8
-25
-30
-35
-40
-45
2.30
2.35
2.40
2.45
2.50
2.55
2.60
Frequency [GHz]
Figure 6: Individual reflection coefficient magnitude behavior.
Divider 2N
I8
Divider 1N2
Element 8
I7
Divider 2N
I6
Element 7
Divider 1N1
Element 6
Divider
3N
I5
Element 5
I4
Divider 1N1
Element 4
I3
Element 3
I2
Divider 1N2
Element 2
I1
Element 1
Figure 7: Array BF network topology: composed by simple T-junction power dividers.
Table 2: Input impedance (Ω) and Dolph-Chebyshev excitation coefficient of the array elements.
N
1
2
3
4
5
6
7
8
ZN
50.3 + 0.6i
45.1 − 7.0i
48.8 − 4.6i
43.5 − 3.8i
44.3 − 3.2i
48.3 − 4.5i
45.4 − 6.2i
50.0 + 0.3i
IN
0.58
0.66
0.88
1.00
1.00
0.88
0.66
0.58
PIERS Proceedings, Cambridge, USA, July 5–8, 2010
216
Figure 8: Final array geometry: backside view.
0
Simulated (BW = 6.1%)
Measured (BW = 7.3%)
-5
|Γ| [dB]
-10
-15
-20
-25
-30
-35
2.30
2.35
2.40
2.45
2.50
2.55
2.60
Frequency [GHz]
Figure 9: Array reflection coefficient magnitude.
5. CONCLUSION
This paper presents a new effective strategy for designing probe-fed LP thick microstrip antennas.
It consists in designing the patch to operate at the zero input reactance condition (Xin = 0),
which results in a symmetrical return loss bandwidth. As external matching networks are no
longer required, the antenna construction process is simplified. Besides, the new approach can be
applied to the design of microstrip radiators of arbitrary patch shapes regardless of the substrate
thickness. Experimental results for a linear array of suspended microstrip elements validate the
design strategy.
ACKNOWLEDGMENT
This work was partially supported by CAPES, FUNTTEL-RHODES under Grant 01.05.0154.00
and FINEP-SINAV under Grant 01.07.0540.00.
REFERENCES
1. Garg, P., P. Bhartia, I. Bahl, and A. Ittipiboon, Microstrip Antenna Design Handbook, Artech
House, Boston, 2001.
2. Volakis, J. L., Antenna Engineering Handbook, 4th Edition, McGraw-Hill, New York, 2007.
3. Hall, P. S., “Probe compensation in thick microstrip patches,” Electron. Lett., Vol. 23, No. 11,
606–607, 1987.
4. Chang, F. A. and K. L. Wong, “A broadband probe-fed patch antenna with a thickened probe
pin,” Proc. Asia-Pacific Microwave Conf., 1247–1250, Dec. 2001.
5. Marzall, L. F., “Antenna arrays with suspended microstrip radiators,” M.Sc. Thesis, Technological Institute of Aeronautics, São José dos Campos, Brazil, 2009 (in Portuguese).
6. HFSS, Ansoft Corporation, www.ansoft.com/products/hf/ansoft hfss/.
7. Marzall, L. F., R. Schildberg, and J. C. S. Lacava, “High-performance, low-cross-polarization
suspended patch array for WLAN applications,” Proceedings of IEEE International Conference
on Antennas and Propagation, Charleston, USA, June 2009.