Dual-Polarized Square-Shaped Offset-Fed. Reflectarray Antenna with High Gain and High. Bandwidth in the 60 GHz Domain. Tristan Visentin, Wilhelm Keusgen ...
Dual-Polarized Square-Shaped Offset-Fed Reflectarray Antenna with High Gain and High Bandwidth in the 60 GHz Domain Tristan Visentin, Wilhelm Keusgen and Richard Weiler Department of Wireless Communications and Networks Fraunhofer Institute for Telecommunications, Heinrich Hertz Institute D-10587 Berlin, Germany {tristan.visentin, wilhelm.keusgen, richard.weiler}@hhi.fraunhofer.de
Abstract—This paper presents the design, fabrication, simulation and measurement of a compact dual-polarized, high gain and high bandwidth offset-fed reflectarray antenna operating at 60 GHz. The square-shaped reflectarray antenna focusses two wide feed antenna beams with orthogonal linear polarizations into one narrow and dual-polarized main beam. At first, the design procedure is described. This includes the choice of a convenient double-layer unit cell comprising variable-sized crossed dipoles, the aperture efficiency optimization and the transfer of the needed progressive phase distribution to the scattering elements. Next, 3-D full-wave computer simulation results demonstrate the feasibility of the design. Promising nearfield measurement results of a fabricated reflectarray antenna validate the concept and show a high bandwidth of 8 GHz and a high gain of more than 33 dBi.
II. D ESIGN A 260 mm square-shaped reflectarray containing 104 ˆ 104 scattering elements has been designed. For the design a duallayer structure of crossed dipoles as scattering elements, each dipole arm for one orthogonal linear polarization, has been chosen. In order to form the needed phase distribution, the variable-size approach from [4] is applied to the elements. A. Reflectarray setup
Y
I. I NTRODUCTION w
A typical modern reflectarray consists of an array of printed scattering elements on a substrate which is backed up by a ground plane. These elements can be of any arbitrary type and shape and their function is to provide a preadjusted phasing to form a focussed beam when the reflectarray is illuminated by a feed antenna, in a similar way to the well-known parabolic antenna. Reflectarray antennas provide some key advantages in comparison to parabolic reflectors. One of the main benefits of the reflectarray is that it exhibits superior cross-polarization performance when compared to parabolic reflectors [1, p. 256], enabling dual-polarized transmission. Moreover, beam steering up to wide off-broadside angles in the order of ˘45˝ , as yet only known from phased array antennas, is possible with reflectarrays [2]. The principles of reflectarray antennas are described in [3]. In this paper, the challenge of implementing a compact dualpolarized offset-fed reflectarray antenna with high bandwidth (8 GHz) and high gain (35 dBi) in the 60 GHz domain has been taken. Two wide-beam feed antennas that are polarized in orthogonal directions respectively illuminate the reflectarray aperture. The intention is to collimate the two feed antenna beams into one main antenna beam with two orthogonal linear polarizations. Thus, the antenna can be used for multiplexing, demultiplexing or full-duplex applications.
#– ez
l
X
ϑi
Z #– r fe
C
y0
ϑt FRx
#– E Rx FTx
zf
#– E Tx yf αr p-polarized feed
x0
a Reflectarray printed circuit board (PCB)
b
Tx Rx 60 GHz Front end
s-polarized feed
Fig. 1. Schematic overview of a square-shaped aperture reflectarray and a 60 GHz front end with Rx and Tx feed antennas.
In the setup, the two feed antennas correspond to receive (denoted as Rx) and transmit (denoted as Tx) antennas. A medium offset position of the feed antennas between the center and the lower edge of the aperture has been chosen. In order to avoid shadowing, the main beam is tilted 20˝
An illustration of the s- and p-polarization of the Rx and Tx feed antennas according to the incidence angles ϑi is also given in Fig. 1. B. Aperture efficiency optimization The approximate cosq pϑq feed pattern model from [5, p. 128] has been chosen to calculate the aperture efficiency in the same manner as in [6], [7]. Broad-beam feed antennas with nearly 90˝ of half-power beamwidth (HPBW) are considered for the design to get a more compact antenna. These can be e.g. on-chip antennas or open-ended waveguides with very low directivity. Thus, a feed pattern parameter of q “ 1 is chosen for the aperture efficiency optimization. In order to achieve the highest possible gain, the aperture efficiency is optimized, based on the fixed parameters: x0 “ xf “ 15.635 mm and yf “ ´86.25 mm. Fig. 2 shows a contour plot of the calculated aperture efficiency over y0 and zf with the given parameters. The optimal parameters obtained are zf “ 129.34 mm
36.4
0.5
33.4 12.5
30.4
zf in m
0.4
15.4 18.3
0.3
21.2 24.1
0.2
12.4 9.4
´0.1
´0.05
0.05 0 y0 in m
produces an off-broadside pencil-beam in the direction (ϑb , ϕb ) [3, p. 34], where λ is the free space wavelength. ϑb “ 20˝ and ϕb “ 0˝ are then inserted into Eq. (2). The calculated phase distribution for Tx is plotted exemplary in Fig. 3. Both single phase distributions of Rx and Tx are superposed, where the phase distribution of Tx is transferred to the dipole arm lengths l and the one of Rx is transferred to the dipole arm widths w, respectively. ´155 0.1
´195 ´235
0.05
´275 ´315
0
´355 ´395
´0.05
´435 ´0.1 ´0.13
´475 0 x in m
0.13 deg
´515
Fig. 3. Phase distribution on a square aperture plane (a “ b “ 260 mm) for the Tx feed antenna from Fig. 1.
As unit cell, a single element from the array grid is embedded in a surrounding of an infinite number of equal elements. This infinite array approach is an approximation, but has been shown as adequate enough if the entire array comprises many hundreds or more elements [4]. For the reflectarray in this paper, a double-layer unit cell with variablesized crossed dipoles is utilized. The stacked layer approach combines the advantages of a more linear phase curve of phase-change over element-change (S-curve) with a higher bandwidth than a single layer unit cell design [10]. A view of the unit cell’s structure is given in Fig. 4. Its size is set to half of the design wavelength (2.5 mm), in order to avoid grating lobe formation of the reflectarray [2]. As substrate, the
15.4 0.1
Φpx, yq “ prf e ´ px cospϕb q ` y sinpϕb qq sinpϑb qq2π{λ (2)
24.4
32.8
35.7
Once the optimal offset feed position has been determined, the discrete phase distribution can be computed. One phase distribution is needed for each of the two feed antennas since their feed beam points are shifted to the different positions: FRx “ p´|x0 |, ´|y0 |, 0q and FTx “ px0 , ´|y0 |, 0q. The progressive phase distribution:
D. Double layer unit cell structure
18.4
29.9
C. Phase distribution
27.4 21.4 27
ϑt “ 110.7˝ or a feed rotation angle αr “ 20.7˝ which is very near to the main beam tilt angle ϑb “ 20˝ . Thus, unwanted specular reflection components outside of the main lobe are suppressed while simultaneously beam squint is minimized [8], [9]. By the stated aperture efficiency optimization, an aperture efficiency of 36 % is achievable.
y in m
off-broadside away from the feed. A schematic view of the described structure is given in Fig. 1. The feed antenna’s phase-centers lie in the same plane and are placed on a 60 GHz radio front end at the feed positions of Rx: p´|xf |, ´|yf |, zf q and Tx: pxf , ´|yf |, zf q. The feed beam points are denoted by FTx and FRx , respectively. The aperture tilt angle ϑt is 90˝ plus the feed rotation angle αr . This is the angle between the normal direction of the aperture and the feed beam direction. Red arrows show the polarization directions of the E-fields of Rx and Tx respectively. It is noted that for the Tx polarization in the y-direction, the lengths l of the crossed dipoles have to be adjusted, while for the Rx polarization in the x-direction, the widths w are responsible for the phase shift. Thus, the oblique incidence at the angle ϑi has to be dissociated from the s-polarized (i.e. polarized perpendicular to the incident plane) and p-polarized (i.e. polarized parallel to the incident plane) case. The incidence angles are computed by: ¯ ´ Mb py ´ yf q2 ` zf2 . (1) ϑi “ arccos zf
0.1
Fig. 2. Total aperture efficiency in % vs. feed height zf and y-component of the feed beam point y0 .
and y0 “ ´37.37 mm. This implies an aperture tilt angle
td2 td2
l td1
X
w td1 tb
S-parameter phase in deg
Y
ts
´150 Φi1 pS1:1,1:1 q Φi2 pS1:1,1:1 q Φi3 pS1:1,1:1 q
´250 ´350 ´450 ´550 ´650
´750 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4
ts
l in mm
Fig. 4. Double-layer crossed dipoles unit cell with two RO 3003 substrate layers and a Rogers 3001 bonding film layer in between.
Rogers high frequency material “RO 3003” with εr “ 3 and tanpδq “ 0.0013 has been chosen. Two layers of thicknesses ts “ 0.5 mm are bonded together with a single layer of the tb “ 0.038 mm thick bonding film “Rogers 3001” with εr “ 2.8 and tanpδq “ 0.003. The dimensions of the crossed dipoles are set to td1 “ 0.8 mm and td2 “ 0.2 mm, respectively. In contrast to single-polarized reflectarrays, for duallinear polarization applications, the lengths and widths of the dipole arms must be individually adjusted to allow collimation of the radiation from two spatially separated feed antennas with perpendicular linear polarizations into one main beam. For the given aperture dimensions and offset feed positions, the maximum incidence angle ϑi in the offset direction is calculated by use of Eq. (1) to be about 60˝ . Due to the equal offset positions but different polarization directions, both S-curves for s-polar and p-polar oblique incidence must be considered for the implementation of the phase distribution on the elements. A high frequency structural simulator (HFSS)-model of the unit cell is implemented and the reflection phase curves over the varied lengths l and widths w are obtained for s-polar and p-polar oblique incidence angles in 5˝ steps by a computer simulation. The results are shown in Fig. 5. All incidence angles ϕi (oblique incidence in the x-direction, not shown in Fig. 1) are set to zero and the S-curves for ϕi ‰ 0 are not regarded. The assumption made is that the incidence angles ϕi do not contribute significantly to the phase shift change of the S-curves because the angles do not lie in the offset direction and are therefore much smaller. Thus, normal incidence is assumed in the x-direction and different S-curves for oblique incidence in the y-direction are used in 5˝ steps to transfer the phase distribution to the crossed dipole dimensions. The starting phase is set to ´155˝ at the beginning of the curves which corresponds to lengths (or widths) of 0.8 mm.
S-parameter phase in deg
(a)
0 Φi1 pS1:2,1:2 q Φi2 pS1:2,1:2 q Φi3 pS1:2,1:2 q Φi4 pS1:2,1:2 q
´200 ´400 ´600 ´800
´1000 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4
l in mm (b) Fig. 5. S-curves of the unit cell from Fig. 4 for s-polar [Φi1 : ϑi “ 0˝ ; Φi2 : ϑi “ 30˝ ; Φi3 : ϑi “ 60˝ ] (a) and p-polar [Φi1 : ϑi “ 0˝ ; Φi2 : ϑi “ 25˝ ; Φi3 : ϑi “ 35˝ ; Φi4 : ϑi “ 60˝ ] (b) oblique incidence of ϑi over l.
previously described large reflectarray with 104 ˆ 104 doublelayer elements. The parameters for the implemented reflectarray after aperture efficiency optimization for q “ 1 are: x0 « 7.82 mm, y0 « ´11.16 mm, xf “ x0 , yf “ ´25 mm, zf « 39.9 mm, αr « 20.6˝ and a main beam tilt angle ϑb “ 20˝ . Similar to the large reflectarray, a maximum aperture efficiency of around 36 % is computed if the feed antennas are placed at the optimal feed positions. As feed antennas, two open-ended circular waveguides (OE-CWGs) are implemented. The corresponding 3-D HFSS-model and the obtained gain patterns for one active feed antenna at each time for ϕ “ 90˝ at 60 GHz are presented in Fig. 6. It is observed that the design approach yields very good results, even for a small reflectarray. The gain of the mainlobe peaks at ϑ “ 20˝ with a value of 26 dBi while the highest sidelobe remains approximately 13 dB below this value. Furthermore, a very high radiation efficiency of 94.4 % is calculated by HFSS. F. Construction of the reflectarray
E. HFSS-simulation of 900 elements reflectarray In a next step, a HFSS-simulation is performed for a small reflectarray with 30 ˆ 30 elements and an edge length of a “ b “ 80 mm to validate the design process. This small size reflectarray has been designed in the same way as the
Following the validation of the approach, a 104 ˆ 104 double-layer elements reflectarray has been fabricated and mounted. The mount comprises a wood frame construction that fixes the reflectarray PCB and 3-D printed slide-elements that form a holder for the feed antennas. The holder can be
26
Y
20 14 Z
8
III. P LANAR NEAR - FIELD MEASUREMENTS
2 ´5
X
´11 ´17 ´23 dBi
´29
(a)
30
Gain in dBi
20
Tx active Rx active
10 0 ´10 ´20 ´30 ´180 ´120 ´60
0 20 60 ϑ in deg
120
perform near-field measurements on the antenna. In order to get the main beam direction in the normal direction to the mount, the reflectarray is tilted about the main beam tilt angle of 20˝ and fixed to the wood frame construction.
180
(b) Fig. 6. HFSS 3-D model and simulated 3-D gain pattern, aperture illuminated by both Rx and Tx feed antennas (a). Simulated 2-D gain patterns at ϕ “ 90˝ , aperture illuminated by Rx and Tx separately (b).
adjusted in two directions to set the correct distance from the feed antennas to the reflectarray. The finished mount is shown on the photo in Fig. 7. The 3-D printed holder is made for two WR-15 open-ended rectangular waveguides (OE-RWGs), arranged in the same polarization directions as Rx and Tx on the front end in Fig. 1. With this holder it is possible to
Fig. 7. Fabricated offset-fed reflectarray antenna with mount and 3-D printed holder for two WR-15 OE-RWG feed antennas.
One problem concerning far-field measurements of antennas with electrically large apertures is that the minimum distance at which the measurement can be performed is the Fraunhofer distance. The Fraunhofer far-field begins at a distance of r ě 2l2 {λ, where l is the largest dimension of the antenna aperture (its diagonal) [11, p. 167]. For example, for the reflectarray antenna described in this paper, the far-field starts at a distance of about 54 m at 60 GHz. The inconvenience of these large distances needed for far-field measurements of high gain antennas motivates the use of near-field measurements. Planar near-field measurements as described in [12] are performed on the reflectarray. A measurement setup has been constructed and built up in the Fraunhofer Heinrich Hertz Institute’s radome. The radome is used as a measurement environment to prevent multiple reflections from the walls and ceilings. The setup comprises a vector network analyzer that is capable of measuring S-parameters up to 67 GHz. A WR-15 OE-RWG is used as measurement probe. The probe is mounted on a 2-D x-y-carriage and is able to move to positions in a discrete grid of 2 mm (0.4λ at 60 GHz). The truncated scan plane dimensions are 440 ˆ 440 mm at a scan distance of z0 “ 200 mm above the antenna under test (AUT). Since the network analyzer measures the S-parameters between its ports, only relative values of the complex electric field are measured. An antenna gain can therefore only be calculated based on a measured reference antenna with this specific setup. Thus, a standard 20 dBi pyramidal gain horn is measured as a gain reference and to verify the measurement setup. The unideal probe and its interaction with the near-field lead to errors that are corrected by the approximated analytic probe correction formulas given in [13] during post-processing of the measured data. White noise is minimized by setting the measurement bandwidth to 10 Hz. The constructed setup is shown on the photo in Fig. 8. Subsequent to the measurements of the reflectarray with only Tx or Rx active each time, the probe correction and the computation of the plane wave spectrum (PWS), the 2D normalized directivity patterns are extracted from the PWS and are illustrated in Fig. 9, overlayed by the ETSI class2 antenna radiation pattern envelope (RPE) [14]. The HPBW amounts to approximately 2˝ and the highest sidelobes occur in the E-plane with an amplitude level of ´10 dB. The calculated peak gain of the antenna at 60 GHz amounts to 33.1 dBi and remains approximately constant at 56 GHz (33.4 dBi). Only at 64 GHz, the gain decreases to about 30 dB. The main beam peaks around 0˝ , which shows that the tilt angle of the reflectarray PCB in the mount is adjusted correctly so that the antenna radiates in the normal direction of the wood frame construction shown in Fig. 7. Additionally, the cross-polarization isolation between Tx
Notebook with MATLAB control Network Analyzer “R&S ZVA67”
the conformity with the ETSI class2 RPE in almost all regions. Radome wall
2-D x-y-carriage with WR-15 OE-RWG probe arm Z Y X AUT
Scan distance z0 Millimeter wave absorbers
Normalized directivity in dB
Fig. 8. Realized measurement setup in the radome of the Fraunhofer Institute for Telecommunications, Heinrich Hertz Institute.
0
E-plane H-plane
ACKNOWLEDGMENT
´10
The research leading to these results has received funding from the European Union’s Seventh Framework Programme (FP7-ICT-2013-EU-Japan) under grant agreement number 608637.
´20 ´30
R EFERENCES
´40 ´50 ´90
´60
´30 ´10 0 10 30 ϑ in deg
60
90
60
90
Normalized directivity in dB
(a)
0
E-plane H-plane
´10 ´20 ´30 ´40 ´50 ´90
IV. C ONCLUSIONS A compact 260 mm square-shaped dual-polarized offsetfed reflectarray antenna operating around 60 GHz has been designed, fabricated and measured. Before fabrication, the design process has been validated by 3-D full-wave computer simulations for a small version of the reflectarray. Near-field measurements are then performed on the fabricated reflectarray antenna. The obtained radiation patterns show low sidelobe levels and a good conformity with the ETSI class2 RPE. A high gain of more than 30 dBi has been measured between 56 GHz and 64 GHz with a peak gain of 33.4 dBi. Furthermore, a high cross-polarization isolation of more than ´50 dB between the two linear polarizations has been measured in this frequency range. Accordingly, the presented reflectarray antenna proves its ability of dual-polarized operation while maintaining high gain and high bandwidth around 60 GHz.
´60
´30 ´10 0 10 30 ϑ in deg (b)
Fig. 9. Measured normalized gain patterns in E- and H-plane over inclination angle ϑ at 60 GHz with the ETSI class2 RPE. (a): Rx active. (b): Tx active.
and Rx has been measured. For this purpose, the two feed antennas are connected to different ports of the network analyzer and the magnitude of the S-parameter between these ports is captured. The resulting isolation between the feed antenna polarizations remains stable below ´50 dB in the frequency range of 56-64 GHz. All in all, the measurement results show a good functionality of the design approach and
[1] S. Rao, L. Shafai, and S. Sharma, Handbook of Reflector Antennas and Feed Systems: Theory and Design of Reflectors, ser. Artech House antennas and propagation library. Artech House, Incorporated, 2013. [2] F. Venneri, S. Costanzo, and G. Di Massa, “Tunable Reflectarray Cell for Wide Angle Beam-steering Radar Applications,” JECE, vol. 2013, 2013. [3] J. Huang and J. Encinar, Reflectarray Antennas, ser. IEEE Press Series on Electromagnetic Wave Theory. Wiley, 2007. [4] D. M. Pozar, S. D. Targonski, and H. D. Syrigos, “Design of Millimeter Wave Microstrip Reflectarrays,” Antennas and Propagation, IEEE Transactions on, vol. 45, no. 2, pp. 287–296, 1997. [5] Y. Lo and S. Lee, Antenna Handbook Volume I. Van Nostrand Reinhold, 1993. [6] A. Yu, F. Yang, A. Z. Elsherbeni, J. Huang, and Y. Rahmat-Samii, “Aperture Efficiency Analysis of Reflectarray Antennas,” Microwave and Optical Technology Letters, vol. 52, no. 2, pp. 364–372, Feb 2010. ˙ [7] M. Zebrowski, “Illumination and Spillover Efficiency Calculations for Rectangular Reflectarray Antennas,” High Frequency Electronics, December 2012. [8] J. Budhu and Y. Rahmat-Samii, “Understanding the Appearance of Specular Reflection in Offset Fed Reflectarray Antennas,” in Antennas and Propagation (APSURSI), 2011 IEEE International Symposium on, July 2011, pp. 97–100. [9] S. Targonski and D. Pozar, “Minimization of Beam Squint in Microstrip Reflectarrays using an Offset Feed,” in Antennas and Propagation Society International Symposium, 1996. AP-S. Digest, vol. 2, July 1996, pp. 1326–1329 vol.2. [10] J. A. Encinar, “Design of Two-Layer Printed Reflectarrays using Patches of Variable Size,” Antennas and Propagation, IEEE Transactions on, vol. 49, no. 10, pp. 1403–1410, 2001. [11] C. A. Balanis, Antenna Theory: Analysis and Design, 3rd ed. WileyInterscience, 2005. [12] A. Yaghjian, “An Overview of Near-Field Antenna Measurements,” Antennas and Propagation, IEEE Transactions on, vol. 34, no. 1, pp. 30–45, Jan 1986. [13] ——, “Approximate Formulas for the Far Field and Gain of Open-Ended Rectangular Waveguide,” Antennas and Propagation, IEEE Transactions on, vol. 32, no. 4, pp. 378–384, Apr 1984. [14] ETSI EN 302 217-4-2 V1.5.1 (2010-01), “Harmonized European Standard (Telecommunications series),” ETSI, Tech. Rep., 2010.
