DEVELOPMENT OF A BROADBAND STACKED PATCH ANTENNA ELEMENT FOR K U -BAND PHASED ARRAY ANTENNA

Transcription

1 DEVELOPMENT OF A BROADBAND STACKED PATCH ANTENNA ELEMENT FOR K U -BAND PHASED ARRAY ANTENNA ESA/ESTEC, NOORDWIJK, THE NETHERLANDS 3 OCTOBER 212 Adriaan Hulzinga (1), Jaco Verpoorte (1), Neelakantam Venkatarayalu (2), Andrew Thain (2) (1) National Aerospace Laboratory NLR, P.O. Box 153, 83 AD Marknesse, The Netherlands, Adriaan.Hulzinga@nlr.nl, Jaco.Verpoorte@nlr.nl (2) EADS Innovation Works, BP Toulouse Cedex 3, France, Neelakantam.Venkatarayalu@eads.net, Andrew.Thain@eads.net ABSTRACT A broadband stacked patch antenna element has been developed for use in a K u -band phased array antenna for airborne satellite communication. The antenna element has two output ports for two orthogonal linear polarizations. The initial design of the antenna element has been improved in order to minimize coupling between the output ports. In addition the antenna element is optimized for use of this element in a phased array antenna under scanning conditions. 1. INTRODUCTION A broadband stacked patch antenna element has been developed. This antenna element will be used in a K u - band phased array antenna for airborne satellite communication. This development has been carried out in the EU FP7 project SANDRA (Seamless Aeronautical Networking through integration of Data links, Radios, and Antennas). In aviation the need for communication increases (air traffic communication, airline logistics, passenger entertainment etc.). The SANDRA concept consists of the integration of complex and disparate communication media into a lean and coherent architecture. The communication needs for cockpit and cabin are combined into one overall communication system covering both satellite communication and terrestrial communication. For satellite communication a combined antenna for L-band and K u -band will be designed. Currently a K u -band-only antenna has been designed which will be used in a demonstrator consisting of a broadband K u -band phased array antenna with Optical Beamforming (receive-only). This antenna needs to cover the complete receive band for aeronautical earth stations and DVB-S broadcast in K u band ( GHz). The optical True Time Delays (TTDs) in the Optical Beamforming Network (OBFN) enable a squint free beamsteering over the whole band. The K u -band antenna itself also needs to cover this whole frequency band in both input impedance and radiation pattern. 2. ANTENNA ARCHITECTURE The antenna architecture has been described in detail in [1]. The architecture is based on 2x2 sub-arrays with local RF beamforming and down-conversion before optical beamforming. The complete optical beamforming architecture is shown in Figure 1. This is a two-level modular architecture where two stages of optical beamformers are used. In Figure 1A, the phasedarray antenna structure is depicted. It consists of N number of antenna tiles (here N=24 is depicted for illustrative purpose only), where each tile consists of 64 Antenna Elements (AEs) (the yellow squares in Figure 1B). Thus the total number of AEs in the system is 64xN. A monolithic microwave integrated circuit (MMIC) beamforming network will be implemented to delay and combine the signals from 4 neighbouring AEs. This is illustrated by the red squares in Figure 1B. In Figure 2 the hybrid beamforming is shown in more detail. Optical TTDs are used for the beamforming of the sub-arrays. RF TTDs or phase shifters are used for the beamforming of the elements in the sub-array. This means that each antenna tile will not have 64 RF outputs but only 16 RF outputs. These outputs will feed a 16x1 optical beamforming network (OBFN) as shown in Figure 1C. The number of 16x1 OBFNs needed is equal to the number of antenna tiles, which is N. Figure 1. The complete optical beamforming system.

2 The 16x1 OBFN consists of a laser diode, 16 low noise amplifiers (LNAs) driving 16 optical modulators, a 16x1 optical beamforming chip and a balanced photodetector (BPD) to restore the RF signal. The RF output of the BPD is then amplified by a second-stage amplifier before being fed to the modulator RF inputs of the Nx1 optical beamformer, as illustrated in Figure 1D. This beamformer delays the received signals from the first stage beamformers (Figure 1E) and combines them. Here we consider RF signals with a frequency range of 1.7 GHz to GHz, hence a 2.5 GHz bandwidth. RF FRONT-END OBFN 1 Figure 3. Top view of initial design of the K u -band antenna element ANTENNA LNA (+DC) RF TTD (OR PS) TTD Figure 2. Schematic overview of hybrid beamforming architecture. Optical TTDs are used for the beamforming of the sub-arrays. RF TTDs or phase shifters are used for the beamforming of the elements in the sub-array. 3. ANTENNA ELEMENT DESIGN 3.1. Stripline-Fed Design The initial design of the antenna element is based on a dual linear polarized patch antenna element [2], [3]. The use of stacked patches, a primary patch and a parasitic patch, increases the bandwidth of the antenna element (Figure 3 and Figure 4) [4]. The parasitic patch is larger than the primary patch and is inverted. The primary patch is fed by two apertures in the ground plane, one for each polarization [5]. Asymmetrical stripline feed lines run underneath the dog-bone shaped aperture in the groundplane. For the initial design the size of the patches, the thickness and permittivity of the substrates and the shape and size of the apertures were optimised to increase the bandwidth. Shorting pins were used to increase the isolation between the two ports of the antenna element (Figure 5). The initial design of the antenna element was simulated as an infinite array using periodic boundary conditions in the finite element method based simulation tool (Ansoft HFSS). Figure 4. 3D view of initial design of the K u -band antenna element. Initially the layer between the two patches was made of foam. Because the foam is more difficult to handle in the manufacturing process, it was replaced by a dielectric with a hole just above the patches Coaxial line to Stripline Transition In the final phased array antenna the radiating elements of the array have to be integrated with the layers of the Printed Circuit Board (PCB) that contain the RF frontend components (Low Noise Amplifiers (LNAs) and mixers). To connect the feed lines of the antenna element with the components of the RF stack, a vertical transmission line is needed. For microwave frequencies a connection by means of a normal via in the PCB layers is no longer sufficient (due to high mismatches and associated losses). Therefore a transition from vertical coaxial line to horizontal stripline transmission line has been designed. The initial design of this transmission line used an outer screen consisting of vias whereas the refined design uses half a cylinder (Figure 5).

3 ANSOFT ANSOFT ANSOFT National Aerospace Laboratory (NLR) Isolation Ku1 InfArray17a3 Curve Info db(s(2,1)) Setup1 : Sweep1 db(s(2,1)) -1-2 Figure 5. Vertical transmission line with outer coaxial conductor consisting of a screen (half a cylinder). Also shown are the ground planes with apertures and some shortening pins between the transmission lines of the two output ports (linear polarisation) 3.3. Simulations of input impedance and radiation pattern Simulations of input impedance (S 11 ) and isolation between the output ports (S 12 ) were carried out for this antenna element in an infinite array. Periodic boundary conditions were used to simulate a doubly periodic infinite array. The commercial packages Ansoft HFSS and CST Microwave Studio were used to perform these simulations. The simulated reflection loss (S 11 ) of the antenna element is shown in Figure 6. The antenna has the required broadband behaviour in the downlink part of the K u -band: the S 11 is better than -14 db in the range 1.7 GHz to GHz. In Figure 7 the isolation between the two output ports of the antenna element is shown. In the initial design the isolation between the two ports is better than 16 db. The radiation pattern of the element (at 12.7 GHz) is shown in Figure 8 for both the E-plane and the H-plane. The gain at this frequency is more than 4 db. National Aerospace Laboratory (NLR) Returnloss S11/S Ku1 InfArray17a3 Curve Info db(s(1,1)) Setup1 : Sweep1 db(s(2,2)) Setup1 : Sweep Freq [GHz] Figure 6. S 11 simulation results (reflection coefficient) of initial design Freq [GHz] Figure 7. S 21 simulation results (transmission coefficient between ports) of initial design Radiation Pattern Ku1 InfArray17a3 Curve Info db(realizedgaintotal) Setup1 : LastAdaptive Freq='12.7GHz' Phi='deg' db(realizedgaintotal) Setup1 : LastAdaptive Freq='12.7GHz' Phi='9deg' Figure 8. Radition pattern of the initial design (simulation results). 4. OPTIMIZATION UNDER SCANNING CONDITIONS 4.1. Reduction of the mutual coupling between antenna element ports In the second step of the design, the performance of the antenna element under the conditions of scanning was analyzed. In particular the active S-parameters were determined. The active S-parameters take into account the mutual coupling of the element with the neighbouring elements and also take into account the phase of the current on the neighbouring elements while receiving signals from a specific direction. The active S- parameters of the antenna element are shown in Figure 9 and Figure 1 for scanning in the planes φ= o and φ=9 o. The simulations were carried out using the frequency domain solver in CST-MWS. From these simulations it was concluded that 1. S 11 and S 22 (and hence the active reflection coefficient) degrade in performance for increasing scan angles away from the normal (broadside) direction. This degradation is not due to scan blindness [6], in which case the input impedance is very high and the active reflection coefficient is 1 at certain scan angles. 2. Performance of S 11 in the φ= plane is similar to the performance of S 22 in the φ=9 plane and viceversa. This is due to the symmetry in the structure in the two orthogonal (x and y) directions

4 3. Degradation of performance for the φ-polarization is more severe for scanning in the φ= plane than for scanning in the φ=9 plane. A similar conclusion is drawn for the θ-polarization consistent with the observation in item 2 above. 4. The observed similarity in S 12 and S 21 for broadside radiation is lost while scanning at angles away from the normal. This is not in conflict with reciprocity which dictates that the 2-port S-parameter matrix be symmetric for a 2-port network (with no nonreciprocal media). The S-parameters computed are the active S-parameters for the infinite array and are not to be interpreted as S-parameters of any other 2- port network. 5. Performance of S 12 in the φ= plane is similar to the performance of S 21 in the φ=9 plane and viceversa. This again is due to the symmetry in the structure in the two orthogonal directions. 6. The isolation between the ports for the two polarizations degrades at certain frequencies and these frequencies tend to vary with the scan angle. More specifically, the frequency at which strong coupling occurs decreases with increasing scan angle. 7. Looking at the variation of the coupling between the two ports (S 21 ) with the scan angles, it is observed that for the scanning in the φ= plane there is a stronger coupling for negative scan angles than for the corresponding positive scan angles. Similarly for the scanning in the φ=9 plane there is a stronger coupling for positive scan angles than for the corresponding negative scan angles. With the initial configuration of the shorting pins, under scanning conditions, port 1 of the antenna element can couple to port 2 of an adjacent element and vice versa. An asymmetrical variation in S 12 with positive and negative scan angles in a particular φ-plane is clearly attributed to the asymmetrical coupling between the two ports under the current configuration of the shorting pins that run along the diagonal of the antenna element (Figure 5). Thus to have a better port isolation, even under scanning conditions, a configuration for shorting pins which completely encloses each individual port was adopted (Figure 11). Again the active S-parameters were simulated (Figure 12). These simulations show the improved isolation between the ports of the antenna elements (S 12 ). A similar improvement was observed for scanning in the φ=9 o plane θ φ Figure 9. Active S-parameters for the antenna element under scanning in the φ = o plane Figure 1. Active S-parameters for the antenna element under scanning in the φ =9 o plane. Figure 11. Improved design with additional shorting pins and optimised apertures in the ground plane.

5 Scan Angle θ varied in the φ = plane Figure 12. Active S-parameters for the antenna element with the improved design under scanning in the φ= o plane. Better S 12 performance due to extra shorting pins Antenna element optimization Because the active reflection coefficient (S 11 ) degrades significantly beyond +/- 2 o scan angles (Figure 9 and Figure 1), the impedance matching of the aperture coupled patch antenna structure was optimised for these conditions. In particular the dimensions of apertures and stubs that are critical for the impedance matching were optimised for improved reflection loss (for scan angles up to 4 away from broadside). The parameters that govern the impedance matching are (as shown in Figure 11): a. L s Length of the Stub b. L b Aperture breadth c. L l Aperture length d. L w Aperture width The optimization procedure was started with varying each of the four parameters and assessing the impact on the reflection coefficient (S 11 ). Because of the symmetry in the design the variations were limited to scan angles between -4 o and o. Only S 11 was evaluated. Based on the results of these parameter variations, the following observations were made: 1. Irrespective of the parameter being varied and the plane of the scanning, the performance of the active reflection coefficient degraded with increasing scan angles. 2. Irrespective of the parameter being varied, the performance of active reflection coefficient was better in the φ=45 plane than the performance in the φ= plane. A tighter circle in the Smith Chart was observed, irrespective of the location, which demonstrates the potential to realize good impedance matching for that frequency band. This result helped to identify a proper objective function in the optimization procedure which was defined for scanning in the φ= plane. 3. Increasing the length of stub moved the impedance curves in the counter-clockwise direction. The influence indicates that while bandwidth is maintained a better overall match could be obtained. This parameter was chosen as an optimization variable. 4. Decreasing the breadth of the aperture tends to move the impedance curves to the left in the Smith Chart. This parameter with potential for impedance matching was also chosen as an optimization variable 5. Increasing the aperture length tends to move the impedance curves to the left in the Smith Chart. Again this parameter having potential for impedance matching was chosen as an optimization variable 6. Increasing the width of the aperture adversely affects the impedance bandwidth. Hence this variable was fixed at a minimum of.1mm and not chosen as an optimization variable. These observations led to an optimization that was carried out for three design variables: the breadth of the aperture, the length of the aperture and the width of the aperture. The objective was to reach an S 11 of less than - 14 db in the frequency range 1.7 to GHz. The optimization was carried out for the case with the worst performance: the plane φ= with scanning angle θ=4. New values were obtained for the dimensions of stub and aperture. The active S-parameters for the optimized element are given in. Simulations showed that the active S-parameters for the plane φ=45 also improved and are better than the results for φ= (as was observed in the case of the un-optimized design) Scan Angle θ varied in the φ = plane Figure 13. Active S-parameters for the optimized antenna element under scanning in the φ= o plane. Better S 11 performance due to optimisation of stub and aperture dimensions.

6 4.3. Design sensitivity The sensitivity of the performance metrics of the optimized design to dimension variations was investigated. This is important because the optimization parameters used for better impedance matching are susceptible to variations due to tolerances in the mechanical fabrication of the array. The parameter most susceptible to mechanical fabrication precision is the stub length, L s. The alignment of the slot forming the aperture with the stripline is critical to get the correct dimension of the stub length. The variation of the active S-parameters with respect to L s is computed with a resolution of 5 µm assuming the mechanical fabrication to be able to provide a precision of at least 5µm. The results obtained for the worst case performance i.e., for the scan in the θ=4 and φ= direction are shown in Figure 14. The other two design variables are based on the optimized design rounded off to.1mm precision i.e., L b =1.mm and L l = 1.9 mm. It is observed that indeed the S 11 and S 22 are quite sensitive to the variations in this design parameter even for variation in the order of 5 µm. The port isolation is independent of the variation in this parameter as expected. It is also noticed that although the case with L s = 1.4 mm provides a better match to 5 Ω around 11.5 GHz, the high frequency performance degrades, making this a sub-optimal design compared to the results of L s = 1.35 mm which is closer to the optimized design. The two other optimization variables, viz., the length of the aperture and breadth of the aperture were varied around the optimized design value with a resolution of 15 µm. Unlike the length of the stub parameter, these two parameters do not influence the performance of the optimized design significantly at the considered resolution L s = 1.25mm L = 1.3mm s L = 1.35mm s -1-2 Scan Direction θ=4, φ = L s = 1.2mm L s = 1.4mm Figure 14. Sensitivity analysis of the improved design for tolerances in the stub length. 5. MEASUREMENTS Manufacturing of test structures of the antenna frontend has started. First some small sub-arrays (2x2 AE) and (4x4 AE) will be made and measured. If the measurements show that these structures comply with the simulations carried out in the design phase, then a complete antenna tile will be manufactured and measured. 6. CONCLUSIONS This development has led to a broadband antenna element covering the whole frequency band from 1.7 GHz to GHz. The antenna element was optimised for use in a scanning phased array antenna. The antenna element maintains its broadband character for reflection coefficient (S 11 ) and port isolation (S 12 ) even under scanning conditions. A finite array based on the current design of the antenna element will be manufactured shortly. Subsequently the S-parameters of the antenna elements and the radiation pattern of the array will be measured. 7. ACKNOWLEDGEMENT The research leading to these results has been partially funded by the European Community's Seventh Framework Programme (FP7/27-213) under Grant Agreement n The SANDRA project is a Large Scale Integrating Project for the FP7 Topic AAT (Integrated approach to network centric aircraft communications for global aircraft operations). The project has 3 partners and started on 1st October REFERENCES 1. Jaco Verpoorte, Harmen Schippers, Pieter Jorna, Chris G.H. Roeloffzen, David A.I. Marpaung, Rens Baggen, Bahram Sanadgol, Architectures For Ku-Band Broadband Airborne Satellite Communication Antennas, ESA/ESTEC, Antenna Workshop Garg, R., P. Bhartia, I. Bahl, and A. Ittipiboon, Microstrip Antenna Design Handbook, Artech House, Norwood, MA, D. M. Pozar and D. H. Schaubert, Microstrip Antennas: The Analysis and Design of Microstrip Antennas and Arrays, IEEE Press F. Croq and D. M. Pozar, Millimeter wave design of wide-band aperture coupled stacked microstrip antennas, IEEE Trans. Antennas and Propagation, vol. 39, pp , December 1991.

7 5. A. Adrian and D. H. Schaubert, Dual aperture coupled microstrip antenna for dual of circular polarization, Electronics Letters, vol. 23, pp , November D. M. Pozar, Scan blindness in infinite phased arrays of printed dipoles, IEEE Trans. Antennas and Propagat., vol.32, no.6, 62-61, June 1984