This paper reports on the design, fabrication and test of a five-element linear array developed at K-band and using LTCC-compatible (low-temperature co-fired ceramic) technology. The array was printed in a multi-layer laminated slab and radiated from one of the edges (endfire). In the literature, several antennas using LTCC technology have been presented but none were endfire-type antennas. The main advantage of the linear end-fire array is that there is more board space available for placing the electronics than for tile arrays. Also, several linear arrays can be stacked together to form a planar array and still have plenty of room to layout the electronics. This is especially important at the higher frequencies where room is at a premium.

The array radiating element was an open waveguide realised in laminated technology. The laminated waveguide (LWG) vertical walls were formed by a series of vias connecting the upper and lower ground planes. Compared to striplines embedded in a dielectric medium, waveguides have the capability to confine the electromagnetic fields, thus avoiding cross-coupling with adjacent elements or nearby electronic circuits. The element developed was fed through a special coaxial-to-waveguide transition. It exhibited a 1GHz bandwidth centered at 20.7GHz and a 2.7dBi gain. Electromagnetic simulations agreed well with the measurements.

A five-element array was developed (Fig. 1) using the successful element. Electromagnetic simulations were also performed for comparison with measurements. The analysis focused particularly on the scan impedance and the radiation pattern characteristics. Results in terms of return loss were comparable to those for a single radiating LWG, and the mutual coupling between the elements was lower than -20dB across the bandwidth. Simulation of radiation patterns were carried out and analysed for several scan angles and frequencies. The simulated boresight radiation pattern for a five-element array was stable over the bandwidth and the gain reached 9.7dBi at 20.7GHz. A feed system was setup for use with the array when measuring the radiation patterns. It included line stretchers used to adjust the phase excitation of each element, thus giving the possibility of providing scan radiation patterns. A maximum gain of 9.2dBi gain was measured in the boresight direction at 20.7GHz. This result was in good agreement with the prediction allowing for a reasonable error margin in the gain calibration. The measurements of the radiation patterns for various scan angles and across the operating bandwidth agreed with the simulation results, and thus, demonstrated the scanning capability of this array. An example for a 20o scan angle is presented in Fig. 2.

The laminated waveguide array is an interesting candidate for communication and radar systems, allowing for direct integration of the antenna with the RF electronics.

Fig. 1: Array of five radiating laminated waveguides

Fig. 2: Measured radiation patterns for a 20o scan angle

Planar printed antennas, are often required to be fed by parallel feednetworks, which, when printed on the same substrate as the radiating elements, create mutual coupling, spurious radiation and excite surface waves. This considerably affects the array efficiency.
Numerous architectures were proposed in the past; some of them use the multilayer structure (which, by itself, exhibits low efficiency), or perpendicularly fed structures.
The latter consist of a substrate for the radiating element and another substrate which for the feed-network.
The paper reviews some of the basic configurations proposed until now, and shows the specific improvements introduced by the proposed architecture. The goal of this development was to design a wide-band circularly polarized printed element, which can be easily integrated in an array, and separates the radiating space from the beamformer.
Linearly polarized elements suitable for this purpose are well known: LTSA (with its variations), the quasi-Yagi dipole, and other types of printed dipoles.
The main drawback of these elements is that the only polarization achievable with these elements is linear and the polarization vector has to be parallel to the substrate. These limitations are significant and some effort was invested in the development of alternatives.
In [1, 2], an aperture-fed patch fed by a perpendicular substrate is presented, in which the polarization of the element is perpendicular to the feeding substrate. In [3], the problem seems to be solved; however the solution is achieved using a pair of crossed dipoles with one feeding point. This architecture does not allow multiple polarizations. The architecture proposed (Figure 1), consists of two back-to-back microstrip substrates perpendicularly feeding a pair of crossed dipoles. Each half-dipole is individually fed by a microstrip line. The four microstrip lines are fed separately and by suitably arranging the phase of each feed, all polarizations are achievable: i.e. linear (horizontal, vertical and ±45°), and circular (RHCP or LHCP).
In all polarization modes, the polarization vector can be independent of the orientation of the feeding network substrate.
REFERENCES
[1] David M. Pozar, Robert W. Jackson Jr.; An aperture coupled microstrip antenna with a proximity feed on a perpendicular substrate, IEEE Trans. Antennas Propagat., vol. 35, pp. 728 - 731, June 1987.
[2] Naftali I. Herscovici, D. M. Pozar; Full-wave solution for an aperture-coupled patch fed by perpendicular coplanar strips, IEEE Trans. Antennas Propagat., vol. 42, pp. 544 - 547, April 1994.
[3] Aleksander Nesic, Sinisa Jovanovic, Ivana Radnovic; Wideband printed antenna with circular polarization, 1997 IEEE Int. Antennas Propagat. Symp. Dig. vol. 35, pp. 1882 - 1885, June 1997.

Frequency selective surfaces (FSS) are often used as bandpass or bandstop radomes for antennas. For many radome applications, the FSS is curved and of finite extent, rendering the usual assumption that FSS is a doubly-infinite, doubly-periodic, planar structure invalid. Therefore, the conventional method of moments (MoM) based on Floquet¡¯s theorem is inadequate. In addition, when the FSS is located in close vicinity of the antenna, mutual coupling between the antenna and FSS cannot be neglected due to the strong effects on the current distribution on the antenna. Therefore, a method for accurate analysis of finite conformal antenna-FSS system must consider the effects of the finite size and curved shape of the FSS, as well as the mutual coupling between the antenna and FSS.

In this paper, a full-wave method based on the volume-surface integral equation is applied to analyze microstrip antenna arrays with finite FSS radome. The antenna is fabricated on finite-size substrate and ground plane. Both the antenna and FSS can be conformal. In the method, the entire structures encompassing the antenna array and the FSS are included in the solution domain. Dielectric materials (substrate) are replaced by equivalent volume currents, while conductive bodies (patch array, ground plane, FSS array and feeding probes) are replaced by equivalent surface and wire currents respectively. The volume integral equation is then established over the dielectric region using the fact that the total electric field is the sum of the excitation field and the scattered field. The surface integral equation is applied on the conductive surface to enforce vanishing total tangential electric field. The resultant equations are solved using method of moments (MoM). Although this method is very rigorous and takes into account the non-planarity, finite size and mutual coupling of both antenna arrays and FSS, it is a computationally very expensive technique. To alleviate this problem, the precorrected-FFT method is used to reduce the memory requirement and accelerate the matrix-vector products in the iterative solution of the equation. Further, to improve on the convergence behaviour of the system, the incomplete LU (ILU) pre-conditioner is employed for better efficiency.

Based on the code developed, we analyzed antenna arrays with different bandpass or bandstop FSS. Both radiation and scattering properties of the antenna array with and without the FSS are calculated. The effects due to the presence of different FSS are studied. For scattering analysis, the FSS radome and antenna array are illuminated by a plane wave from exterior region, while for radiation case, the array elements are excited by coaxial probes. Results obtained showed that it is possible to minimize the pattern distortion (for both radiation and scattering cases) due to the radome by proper choice of the shape and size of the FSS, as well as the distance between the antenna and FSS.

Spherical antenna arrays present a natural choice if complete hemispherical coverage with nearly constant beam width is needed. This makes them an optimal solution for satellite tracking applications, allowing simultaneous communication with multiple satellites from one base station. On the other hand, spherical periodic surfaces find their use as frequency selective radomes or sub-reflectors in reflector antenna systems.

In the analysis, spherical arrays are often approximated with a corresponding planar structure, and the spherical geometry is later accounted for by an appropriate coordinate and phase transformation. However, in many practical applications one cannot neglect the structure curvature. This is particularly important when accounting for the mutual coupling effects, since the magnitudes of mutual coupling coefficients depend strongly on the curvature. Mutual coupling can drastically change the embedded element pattern of each antenna element, and has to be rigorously included in the analysis procedure.

In this article, the rigorous analysis of arrays (of patches or printed dipoles), embedded in a multilayer spherical structure, is performed using the spectral-domain approach and the Moment Method. Since the problem is defined in the spherical coordinate system, the vector-Legendre transformation is applied to the patch current, and the elements of the MoM matrices are calculated in the spectral domain. The original three-dimensional problem is thus transformed into a spectrum of one-dimensional problems.

The vector-Legendre transformation is calculated easily if the patch is situated at a sphere pole. But, to calculate the mutual coupling between elements in the array one needs to calculate the vector-Legendre transformation of basis/test functions with domains at different parts of the sphere. One approach is to describe the currents on each element in a local coordinate system, and then numerically calculate the needed terms using formulas that relate the global coordinate system with local ones. However, this proves to be very time-consuming.

In order to make an efficient program we have established an analytic relationship between vector-Legendre transformations of basis/test functions defined on the central patch (at the pole, α=β=0°) and on the patch located at an arbitrary position (α_n, β_n). In the first step, two auxiliary functions, J^A and J^V, are defined in a manner similar to the representation of the electric field via vector and scalar potentials. Their definitions do not depend on the coordinate system, i.e. they are valid in both global and local coordinate systems. Moreover, their representations in different coordinate systems can be linked with help of the Vilenkin’s addition theorem for spherical harmonics. This simplifies significantly the computation of vector-Legendre transformation of basis/test functions for patches not at the pole, and improves the speed of mutual coupling computation by two orders of magnitude. The same approach is implemented in the analysis of spherical periodic structures, to establish a relationship between spectral domain representations of plane waves impinging on the sphere from different angles, allowing for much faster calculation of monostatic radar cross-section (RCS) for different angles of incidence of the illuminating electromagnetic wave.

An antenna for a Signal Identification System has been designed. The objective of this design is to detect the angle of the arrival signal by electronic sweeping and monopulse system in the horizontal plane. The antenna is working at 1.06 GHz with a bandwidth of 80 MHz (7.54 %). A sweeping of the aiming angle signal with a 3 dB loss level in the edges is desired, as well as a 16.5 dB maximum gain of the total antenna. For all these reasons, a planar array antenna is designed with a number of 32 radiating elements implemented in the structure and divided in 8 columns and 4 rows. Each column is connected to a phase shifter network (Figure 1) and every column will be connected to a stripline network which will feed them all.

Two different prototypes have been designed (Figure 2). In the first one, the structure is based on a multi-layer configuration with rectangular patches (vertical polarization is required) coupled through slots by a feeding microstrip line. Two slot configurations have been probed due to the reduce width of the horizontal ground-plane (column width is approximately 0.5 ë0): U slot, and S Slot form. The U slot showed a better behavior in terms of cross-polar component. The microstrip line is printed over a dielectric with low losses which must be considered in the feeding network. The second design consists on a double stacked patch configuration fed by a coaxial probe.

Same amplitude and phase feeding is desired for the 4 elements composing each column. In the first design, a mixed corporate feed and serial feed has been combined to obtain the same feeding phase. Due to the reduced dimensions of the antenna a complex structure of the feeding network is required. In the second prototype, a corporate fed divider has been designed to obtain same phase and same amplitude in 4 elements and impedance matching in the feeding point of the column.

Two prototype columns of each planar array antenna design were built and measured. A -30 dB level is obtained in the entire work frequency band (1.02 - 1.1 GHz) for the design based in feeding by slots, while a - 15 dB level is obtained in the probe fed design (Figure 3). Although slots design shows a better behavior in terms of bandwidth, stacked patch design achieves coupling levels lower than -25 dB between two consecutive columns. In both two cases, single element radiation pattern indicates that an 8.5 dB gain has been obtained, so it is supposed that the desirable gain of 16.5 dB will be achieve by all elements of the array. Besides, the -3 dB beam width is approximately of 100 ° so a cover has been achieved.

These designs make it a very interesting proposal due to the small size of the antenna, the low cost materials and the results obtained.

We propose a novel waveguide slot array antenna in which radiating transverse long slots are arrayed on the top face of an oversized rectangular waveguide. A hard surface is placed on the bottom face of the guide to force a quasi-TEM mode to propagate whithin the waveguide. The proposed linearly polarized antenna has high efficiency and directivity and it is cost-effective for mass applications. Simulation results confirm a good performance.

Waveguide slot array antennas are becoming a promising solution for communications at the millimeter-wave band. Efficiencies higher than 80% have been achieved for circular polarization by using RLSAs. However, to achieve linearly polarized RLSAs with high efficiencies is quite a difficult task. Therefore, it is more usual to resort to rectangular waveguides for linear polarization. The first planar arrays made use of monomode waveguides. Later, slot array antennas in oversized rectangular parallel plate waveguides were proposed to simplify the antenna. Nevertheless, fields in such oversized waveguides are difficult to control in order to achieve a uniform distribution.

In this paper a hard surface is employed on the bottom face of the guide to force a quasi-TEM mode. There are several ways of implementing a hard surface [1]. The most effective one employs a corrugated surface, where longitudinal corrugations allow longitudinal propagation and inhibit transverse propagation, so reflections from the sidewalls are efficiently eliminated. However, there are other low-cost ways of manufacturing hard surfaces. These are strip-loaded grounded dielectric slabs, where strips may be either narrow, or wider if via holes are used all along the strips, and frequency selective "gangbuster" surfaces printed on a grounded dielectric slab [2]. Therefore, a hard surface makes the uniform field used as excitation preserve its original distribution while is propagating along the guide. Obviously the fields vanish on the metallic sidewalls, but this problem might be overcome employing perfect magnetic conductors instead. Moreover, results show a behavior fairly independent of frequency.

Once a uniform field has been achieved throughout the guide, we can put transverse slots all along the width of the guide as radiating elements.

References:

[1]Per-Simon Kildal, "Artificially soft and hard surfaces in electromagnetics", IEEE Trans. Antennas Propag., vol. 38, no. 10, pp. 1537-1544, Oct. 1990.

[2] S. Maci and P.-S. Kildal, "Hard and soft gangbuster surfaces," in Proc. URSI EMT Symp., Pisa, Italy, May. 2004.

Suárez-Fajardo, C.¹; Ferrando-Bataller, M.²; Antonino-Daviu, E.²; Vico-Bondia, F.^2
1Universidad Distrital Francisco José de Caldas, COLOMBIA;
²Universidad Politécnica de Valencia, SPAIN

The switched beam antenna, a smart antenna technology, consists of an array of antennas, a beam- forming network and a switching matrix. It may be used to generate n beams to improve the carrier to interference ratio (CIR) and the reuse of frequency in cellular systems, increasing system capacity.

Several beam switching techniques in multiple beam system applications have been reported by various authors and a few of these techniques have been considered in this paper. Skahill and White [1] used a Butler Matrix and a switching matrix to excite only the sector of the circular array that contributed to formation of the desired radiation pattern. Krairiksh et al [2], developed a flat four-beam compact array antenna, designed using a circular array of four antennas and four 1-bit phase shifters. Kuga and Arai [3] introduced a flat four-beam switched array antenna and the beam switching was carried out by switching input terminals of the hybrid couplers through terminal one to terminal four. This paper describes a simplified system composed of a circular array and a passive beam-forming network with several ports. It includes two NxN Butler matrices, a switching matrix and a weights box, including the magnitude and phase of the excitation coefficients, which synthesize the required pattern. With the first Butler Matrix any beam may be selected and given the phase shift to deflect the pattern. Several narrow and directional beams can be generated simultaneously around 360°.

To achieve SLL and Crossover level improvement to the adjacent beams, the beams generated by the system could be increased by adding Butler matrices with several ports and increasing the number of antennas used. Nevertheless the system cost would be raised. On the other hand, if Butler matrices with few ports and antennas were used, the synthesized pattern would have poor levels of SSL and crossover. If we excite two adjacent beams simultaneously in a circular array with directional elements arranged in a multiple beam system, we can generate cosine amplitude taper at the outputs of the Butler Matrix. This leads to a remarkable decrease in the side lobe level.

To generate additional directional beams, the beams generated by the system described before could be modified by adding switched phase shifters at the input of the second Butler matrix. With an additional linear progression of phase shifts across the input ports of the second matrix, it would produce linear angular shifts of the resultant directional pattern.

References

[1] G. Skahill and W. White, "A new technique for feeding a cylindrical array," IEEE Trans. Ant. Propag., Mar. 1975, 253-256.

[2] M. Krairiksh, P. Ngamjanyaporn and C. Kessuwan, "A flat four beam compact phased array antenna," IEEE Microwave and wireless components letters, vol. 12, No. 5, May 2002, 184-186.

[3] N. Kuga and H. Arai, "A flat four beam switched array antenna," IEEE Trans. Ant. Propag., vol. 44, Sept. 1996, 1272-1230.

[4] C. Suárez, M. Ferrando and A. Valero, "Pattern Synthesis of Uniform Circular Arrays with Directive Elements", 2004 IEEE AP-S International Symposium and USNC /URSI National Radio Science Meeting, Monterrey California, June 20-25, 2004.

High-gain planar antennas have been a topic of continuous research during last decades because of their mechanical advantages for microwave and millimeter-wave applications. Among them, radial-line slot array-antennas (RLSAs) emerge as one of the best choices for massive production in DBS-TV applications. Some of the features that make these antennas so appealing are their simplicity, low profile, low cost and high efficiency for high frequencies. The structure of a RLSA consists of a set of slots placed on the upper face of a simple parallel plate waveguide. Such guide is fed at its center, typically by a coaxial probe, so that a cylindrically outward TEM mode is propagating inside the radial guide.

As a result, the suitable arrangement of the slots provides the desired radiation pattern and polarization. However, while the circularly-polarized RLSA (CP-RLSA) has been successfully developed and it is commercially available, its equivalent with linear polarization (LP-RLSA) presents some inherent drawbacks. The standard disposition distributes the slots in concentric rings spaced half wavelength, in order to produce linear polarization and a directive beam in broadside direction. As a consequence, such initial design presents a very poor reflection coefficient due to coherent reflections. Many solutions have been proposed to enhance the return loss performance involving the addition of canceling slots or the introduction of a tilt in the main beam. Nevertheless, the performance achieved by a LP-RLSA remains rather far away from the designs of circular polarization.

With the aim of improving the basic LP-RLSA performance, an alternative radiator was proposed by the authors, consisting of a parasitic dipole above the slot. This radiating element improves its matching to the radial guide and consequently minimizes reflections inside the guide. Nevertheless, the matching problem is only solved partially since these reflections still add up coherently. As a consequence, although the reflection coefficient offers good levels at the design frequency, the impedance bandwidth presents very low values.

This paper proposes new topologies of strip-loaded RLSA that presents a wider band performance without compromising other features. Firstly, the basic LP-RLSA configuration with slot-dipole elements is modified by inserting a set of transverse cancelling slots distributed in rings. The suitable design of the involved parameters minimizes the radiation distortion of the new elements. A further improvement converts each cancelling slot into a pair that radiates a pure linear polarization. Finally, the whole antenna can be composed just of slot-strip radiators with an inbuilt removal of reflections.

The key point is to consider a set of four slot-strip elements as the basic radiating cell, so that their reflections are cancelled out. These topologies demand an accurate design of the whole antenna supported on an efficient and accurate analysis. The results obtained will be presented at the symposium, where the significant improvement of the matching bandwidth performance achieved by these new topologies will be shown.

Ka band printed arrays play an important role in many applications like LMDS (Local Multipoint Distribution Systems) and interactive high data rate satellite communications for which low profile antennas are required. In particular, Ka band mobile satellite Tx/Rx terminals find potential applications in civil transportation with large vehicles (bus, trains, aircraft and ships). As terminals have to be integrated within the structure of the vehicles, antennas have to be flat, small, light and robust. The high frequency bands involved allow for a significant reduction of the overall antenna size. However, a further reduction in size could be achieved if the transmit and receive arrays could be integrated to form a single radiating aperture.

In this paper, a new dual-band Tx/Rx antenna element is presented. It is based on a short-circuited patch resonating at 20 GHz which includes a radiating circular waveguide working at 30 GHz. Such an element is compact, light, easy to build and it allows the control of the surface wave emissions which can strongly reduce the antenna efficiency. Because of the large separation between the Tx and Rx frequencies, a different inter-element spacing for the transmit and receive array have to be chosen to avoid grating lobe formation at the higher frequency band.

To solve this problem an interleaved arrangement of dual-band elements with single band radiators resonating at the higher frequency is proposed. In the following, the design of the planar dual band array will be outlined describing the geometry and the main array features. Full wave simulation results and measurements will be presented and discussed.

A high-gain antenna that consists of a rectangular resonant cavity with a metallic grid instead of one wall was proposed in [1]. The grid eyelets are uniform, and the antenna is treated using the metamaterial approach.

We propose a modification of this antenna using on a different physical insight into the antenna operation. Instead of the dense grid, we consider nonuniform radiating slots (Fig. 1), which constitute a planar array. This approach allows us to design an antenna with a pencil-beam radiation pattern, low sidelobes, and two orthogonal polarizations.

We consider the resonant cavity to operate on the TEz101 mode. This is a high-order mode as h < < a,b. (The lowest resonant mode is TMz110.) The magnetic field over the cavity ceiling is approximately given by Hx(x,y,h) = Ho sin(pi*x/a), where Ho is a complex constant. The magnetic field is in-phase all over the ceiling and is tapered going towards the walls at x=0 and x=a.

The array of slots radiates upwards and the electric field is parallel to the y-axis. If the radiating slots are identical, their excitation is naturally tapered along the x-axis. To obtain the pencil-beam radiation pattern and low sidelobes, the excitation should also be tapered in the y-direction. This is achieved by a taper of the slot dimension parallel to the x-axis (going away from the plane y=a/2).

The slots do not have to be close to each other. We follow the classical approach for two-dimensional arrays, where the slot separation is on the order of half wavelength. The size of a slot determines the amount of energy leaking from the cavity through the slot. If the slot is too large, the resonant field structure is overly disturbed, so that the slots are not properly excited.

To obtain the orthogonal linear polarization, we consider the antenna to operate on the TEz011 mode and repeat the above reasoning. As the result, a set of nonuniform slots is obtained as shown in Figure 1.

The TEz101 and TEz011 modes are excited by feeders on the floor. By combining these two modes, the antenna radiates two orthogonal circular polarizations.

The antenna was simulated and optimized using the program Wipl-D [2], and the results were verified experimentally. As an example, an antenna was designed for 5.8 GHz. The dimensions of the cavity (Fig. 1) are a=b=115 mm and h=20 mm. The main-lobe beamwidth is 32 degrees, and the gain is 13 dBi. The bandwidth is 10 MHz.

REFERENCES

[1] G. Benelli, S. Enoch, G. Tayeb, P. Vincent, J.-M. Geffrin, P. Sabouroux, H. Legay, Resonant cavity antennas, Proc. of the 27th ESA Antenna Technology Workshop on Innovative Periodic Antennas: Electromagnetic Bandgap, Left-handed Materials, Fractal and Frequency Selective Surfaces, 9-11 March 2004, Santiago de Compostela, Spain. ESA-WPP-222, pp.107-111, 2004.

[2] B. Kolundzija, J. Ognjanovic, T. Sarkar, M. Tasic, D. Olcan, B. Janic, D. Sumic, WIPL-D Pro. v5.1, WIPL-D, 2004.

Fig.1.

Leaky coaxial cables (LCX) are distributed antennas which can be efficiently employed in places, like mines and tunnels, where the usual antenna systems fail. They consist of an array of small slots on the external conductor of the cable. In [1], an integral equation technique has been developed for the analysis of this structure, based on radial transmission line theory. It is well known that if the number of elements of the array is large, the storage and inversion of the moment matrix [A] becomes problematic. To overcome this problem, various techniques have been proposed in the past. Here, an "eigencurrent approach" is presented, see [2].

The method is based on the determination of the approximate eigencurrents of the entire structure, to be used as entire domain basis functions to diagonalize the moment matrix and avoid its inversion. The array is first decomposed into N_sub equal subarrays (which may even consist of a single element), whose eigencurrents u_n^sub are determined numerically. Then, we introduce the sets E_n formed by entire domain basis functions for the entire array. In particular each basis function is zero everywhere except on one subarray, where it is equal to u_n^sub. Let E be the union of these E_n. Suppose we compute the moment matrix of the entire array using the elements of E as basis and we group the eigenvalues and eigenvectors, according to which set E_n the dominant component belongs to. It is shown that the coupling between the sub-arrays must be taken into account only for a very small number of sets E_n, whose union we call E^cpl, and can be neglected for the others, E^unc. Therefore, the set of approximate eigenvectors of [A] is the union of the set E^unc and of the set of eigencurrents of the reduced moment matrix obtained using E^cpl as basis functions. In this way, we need to construct and diagonalize only small moment matrices, with a considerable CPU-time saving.

As a simple example, consider a LCX with 40 circumferential slots, whose geometric and electric characteristics are: inner/outer conductor radius=3.4/8.8mm, slot angular width α=180 deg, slot width s=3mm, slot separation L=15cm, å_r1=1.26 and frequency f=3GHz. A 60% CPU-time saving is obtained with respect to the full-wave analysis with an error, for the magnetic equivalent current distribution, less than 6%, in L2-norm.

References

[1] G. Addamo, R. Orta, D. Trinchero, R. Tascone and P. Gianola, "Slotted Coaxial Cables for wireless communications", Loughborough Antennas & Propagation Conference 2005, Lougborough, Great Britain

[2] D. Bekers, "Finite Antenna Arrays: An Eigencurrent Approach", Ph.D. Thesis, Eindhoven University of Technology, The Netherlands, 2004

In this paper we present a novel method based on the concept of best approximation in Hilbert spaces, to find the related current distribution of a circular aperture for an array design. Taylor presented a low sidelobe and narrow beamwidth design for circular apertures assuming that the current distribution is phi-independent (in here phi is the angle due to X axis in X-Y plane) For a general solution which is presented here the distribution obtained is not necessarily symmetric and the final current distribution achieved to generate a desired pattern is in terms of polynomial expansions of "r" and imaginary exponentials of phi. (exp(j*n*phi)). The method is a general method and it is capable of becoming as precise as desired at the expense of obtaining more expressing terms, though just for about 10-20 terms the results are very satisfying! After appropriate sampling of aperture through the method presented by Hodges↓ , very good and satisfying results are obtained. In the figure attached we have shown two examples of array design through this method. The first design (left figure) aims to achieve a low sidelobe and narrow beam pattern. After applying our method and appropriate sampling of aperture current we obtained a sidelobe level of -24.6 dB through 204 elements located on an aperture of radius 4*lambda with square mesh grid of 0.5*lambda spacing. In the next design we wished to generate a 3D flat top pattern covering the solid sector |u|<0.4,|v|<0.4. (u=sin(theta)*cos(phi) and v=sin(theta)*sin(phi)) As we can see in the right figure, excellent results are obtained through this method. ↑_{Hodges, R.E. and Rahmat-Samii, Y.:"On sampling continuous aperture distribution for discrete planar arrays", IEEE Trans. Antenna Propag., 1996, 44, (11), pp. 1499-1508}

In this work, a compact corporate fed linear antenna array system is presented. The antenna elements are probe fed circular polarized microstrip patch antennas. The 8x1 linear array will be the building block of a larger 2D array. Output of each row of the array will pass through a phase and amplitude control block and then will be combined to construct a phased array system where beam steering in the other dimension will be mechanical.

Microstrip antenna arrays formed by corporate transmission line feeds have considerably bulky feed networks. In the presented 8x1 microstrip array with corporate probe fed patch antennas, the aim is to reduce the overall size of the feed network. A special feed topology is applied as shown in Fig.1. Using this topology it is possible to combine all the antenna elements on the backside of the array. The connector present in the actual system will be a top-mounted connector where the phase/amplitude control PCB will be mounted in vertical direction.

The presented structure consists of three layers: (i) feed network (ii) patch antennas (iii) radome. The feed network and the patch antennas are connected through plated via holes which served as the probe feeds for the antennas. The top surface of the trimmed square patch antennas is covered with a thin dielectric layer.

All three layers are RO4003 substrate (ε_r= 3.38, tan δ=0.0027). The thickness of 0.81 mm is used in the feed network and the patch antenna layers and 0.51 mm thick substrate is used for the third layer as the radome.

The antenna array characteristics are measured; the return loss of the overall design, which is shown Figure 2(a), has satisfactory results in 8.3-8.4 GHz. Besides, the radiation pattern of the array system shown in Figure 2(b) has a good agreement with the design which is aimed to have about 9 degree 3dB beamwidth. The axial ratio of the antenna array is measured as 0.8 dB which is close to circular polarization. The return loss measurement is done with HP8720D vector network analyzer; whereas, the radiation pattern and polarization measurements are done with HP8757C scalar network analyzer in an anechoic chamber.

More detail and information about the design of the antenna array with corporate probe feed and measurement and simulation results will be given in full paper.

With the growing demand for wideband mobile communications, more and more attention is being paid to the design of new handsets. The present mobile terminals are expected to show increased bandwidth as well as low-loss impedance matching while maintaining low-profile geometry. During the last years a new antenna system for handsets, the planar inverted F antenna (PIFA) has appeared to be the best substitute for the monopole or helix antennas utilized in the past years in mobile system manufacturing. The geometry of this internal antenna The structure consists of a conductive top plate lying over a finite sized ground plane. The plate and the ground plane are interconnected through a feed wire and a shorting strip. The performance of this antenna can be adjusted by alteration of the dimensions of each element, by changing the size and the contour of the top conductive plate or the size of the ground plane. An addition of a parasitic element to the initial PIFA structure provides an enhanced operational bandwidth.

In this study, a PIFA extended to contain a parasitic element (PIFA - Parasitic) is studied. The parasitic element consists of a top plate connected with the ground plane through a parasitic wire and a shorting strip. Optimization of this structure concerning impedance matching at the operating frequency and attainment of a sufficient resonant bandwidth is performed with the use of the Genetic Algorithms (GA) approach. The dimensions of the ground plane, of the top plates, of the wires and of the shorting strips are varied, as well as the distance between the active and the parasitic element, to obtain the optimum antenna. The Super Numerical Electromagnetics Code (SNEC) simulation package, based on the Method of Moments (MoM), is used. The aforementioned optimization parameters are depended only on the wavelength at the frequency of examination, thus, keeping the electrical dimensions of the structure constant.

This report deals with analysis of geometric parameters unequidistant antenna array (AA) with irregular amplitude distribution on its working frequencies. As criterion of optimization, the overlapping coefficient of effective range of frequencies was selected. Under certain conditions apart in equidistant AA from a main lobe existence of parasitic interference lobes higher, than ordinal is possible. Their position, relative to area of real corners depends on a distance between elements in wavelengths and value of phase shift between currents in adjacent elements. The most effective way to struggle with parasitic interference lobes is the usage of unequidistant AA. This report presents studies of influence of elements coordinates in unequidistant antenna array with unequal amplitude distribution on AA scanning sector in different range of frequencies.

By using various current distributions, it could be possible to get a various unequidistant AA. In this work unequidistant antenna arrays were analyzed, which have been obtained by two kinds of current distribution in a original equidistant AA: namely a distribution "cosine in m degree on a pedestal" and "parabola in s degree on pedestal". After this the pattern of antenna array was calculated and corner at which the parasitic interference lobe of any order than zero exceeding level -10 dB in relation to a main lobe was found.

Conclusion.

1. Reducing a relative current amplitude on edges of equidistant antenna array increase a scanning sector of unequidistant antenna array, herewith every type of amplitude distribution in equidistant antenna array have value of pedestal under which scanning sector would be maximum. Changing a value of relative current amplitude on unequidistant antenna array edges, influences on scanning sector at overlapping coefficient become greater than one, i.e. maximum scanning sector depends on the type of amplitude distribution, either under greater or under smaller pedestals.

2. In general the decreasing of relative current amplitude on edges of unequidistant antenna array leads to the worsening of results i.e. parasitic interference lobes might appear because influence of elements on edges decreases.

Hidden automotive antennas are often printed on the glass areas of vehicles. However the introduction of heated windscreens and metal-coated solar reflection glass affects the performance of hidden antennas on glass and alternative sites must be found to place antennas. In this paper we report a novel alternative type of antenna installed under an aperture in the roof of a sports utility vehicle below a plastic panel akin to a sun roof slot as shown in Fig.1. Measurements and simulations of the antenna system have been carried out and are reported in this paper. The antennas are printed on a flexible polyester substrate which is laid on a plastic carrier for rigidity and attached 10mm below the roof aperture which is covered by a plastic panel. In order to gauge the performance of the roof aperture antennas the gain was measured across the FM band and compared with that of two other types of antenna commonly used on cars, an 80cm long mast antenna and a printed antenna situated on the side window. Measured radiation patterns for each antenna are plotted in Fig. 2 for horizontal polarisation at 100 MHz. Simulation of the radiation patterns shows overall that there was reasonable agreement although for horizontal polarisation the null angles are not always accurately predicted. The entire vehicle body was simulated with the antennas in the aperture. This may partly be due to the fact that internal furnishings such as the seats were not included in the model and are the subject of a further investigation. The overall gain of the roof aperture antennas was close to that of a reference roof mount 80cm mast antenna and slightly exceeded the performance of a side glass window antenna. Overall this novel antenna has strong performance but needs further refinement to improve radiation pattern diversity to overcome multi-path reception conditions. The computer model was acceptable and further improvement is being investigated. The full paper will provide details of the full antenna diversity performance and comparisons between the model and measurements. Antenna performance at 100 MHz, 200MHz and 450-800 MHz will be reported.

A high altitude airship (HAA) offers a great potential as a host platform for low frequency antenna array. The main advantage is the large surface available allowing good performances for applications like radar tracking and telecommunications at lower cost than satellites. The antenna array has to be conformed to the ellipsoidal shape of the airship hull and must fill wideband, low weight, low power, low profile conditions. The array lies on the side of the airship and must achieve a bandwidth centered on 500 MHz as well as digital beam forming capability. The unusual configuration of our array must be pointed out; we assume that no ground plane will be used because of low weight and low profile conditions and because equipment must lie outside the hull.

The usual modeling tool (FEKO) based on method of moments (MoM) is no longer appropriate for such a large array. Indeed, the number of antennas is too high and modeling becomes time consuming. Nevertheless, it is also possible to obtain the radiation pattern of the array if the coupling between elements is low. Thus, only the radiation pattern of the isolated element has to be computed by the MoM.
Therefore, after studying the radiation characteristic of a single antenna element, coupling between two elements will be examined according to the frequency and the array curvature. Then, considering a small array, radiation patterns obtained with exact method and computed with neglected coupling will be compared in order to study the frequency and curvature influence on array radiation. This comparison has shown that main lobe reconstruction is fulfilled (Fig. 1) but differences still remain at low frequencies.

Furthermore, since the radius of curvature is not the same according to the angle, the position of the array on ellipse will be investigated using the approximated method. Such conformation leads to non uniformly curved array shapes. Study of the array position is expected to characterize the radiation pattern of such array in order to exploit thereafter this non linearity for digital beam forming (DBF). Radiation patterns obtained with uniform element excitation and for several array positions on ellipsoid will be plotted and compared (Fig. 2).

Fig. 1: 600 MHz radiation pattern of 8 antennas, conformal array (5° curvature) obtained with MoM and approximated method.

Fig. 2: 600 MHz radiation pattern of 10 antennas array, at two different positions on the ellipse

The emergence of novel mobile multimedia applications with high quality services and global coverage will increase the need for mobile satellite systems. Ground terminals are key components of satellite mobile communication networks and the availability in the market of performing and low-cost equipments will be a key issue for a successful commercial deployment.

Besides technical and electrical specifications, vehicular (cars, trains, airplanes ...) mobile antennas will have to satisfy specific requirements like aerodynamics (low-profile), mechanical strength, conformability and easiness of integration on curved surfaces. Moreover, all these requirements have also to take into account the stringent aesthetic requirements of vehicle manufacturers. In this paper we will present the details of the development of a planar array for the reception of Direct Video Broadcasting at Ku band. The antenna is suitable for high speed vehicles like cars, trains and airplanes.

The Ku-band front-end has been designed to receive direct satellite TV and broadband multi-casting information through the Ku-band (10.7-12.75 GHz) satellite transponders. The antenna is based on an hybrid pointing system where the azimuth scan is performed mechanically while the elevation scan is performed electronically. The hybrid phased array antenna is composed by 12 planar wideband dual-linearly polarised printed subarrays fed by a corporate feeding network (see Fig. 1). The radiating sub arrays are mounted inclined in order to maximise the directivity for low elevation angles. Active blocks, including Low Noise Amplifiers and phase shifters, are embedded in the feeding network to optimise the efficiency of the entire array and to allow electronic beam scanning in elevation. A main feeding network connects then all the active elements to the LNB. As already mentioned, the antenna is based on a hybrid beam pointing system where the azimuth tracking is performed mechanically while the elevation tracking is performed electronically. With respect to full mechanical solutions, this approach allows to reduce the thickness of the antenna down to 5-6 cm and to fully integrate it in the vehicle structure.

The dimensions of the presented breadboard are 60 centimetres wide by 75 cm long and 4 centimetres tall (see Fig. 1).

The scanning capabilities of this antenna structure have been demonstrated and will be presented in this paper as well as technical details concerning the realization of the parts of the antenna. The receiving performances of the antenna, while not being fully compliant with the DVB-S service, show the viability of the concept. Fig. 1: Picture of the antenna array

1. Introduction

Sector antennas in modern radio transmission systems have to meet stringent requirements on the one hand, but the manufacturing process should be cost-effective on the other hand.
In the horizontal plane a constant illumination in the given sector with strong sidelobe level attenuation has to be realised. In the vertical plane, a directive radiation pattern, avoiding deep nulls is required.
For lower frequencies, microstrip technology is state-of-the-art to realise such kind of sector antennas /2/, whereas for higher frequencies, waveguide feeding concepts, exhibiting very low losses, are preferred /1/.
The following concepts for a Point-to-Multipoint system in the 10 GHz frequency range will be presented in the contribution at time of conference, comparing the electrical performance, the complexity of manufacturing as well as cost aspects of a 90° sectorising base station antenna in horizontal polarisation.

2. Antenna in plain waveguide technology

Fig.1 shows a photograph of the 4x8 element slot antenna and the radiation pattern over the frequency band. ETSI EN 302085 radiation pattern envelopes class is met without any restrictions. Crosspolar discrimination is about 30 dB, the gain we measured is better 15 dBi in the centre of the sector and better 11.5 dBi over the whole 90° sector.

Fig.1: Slot antenna;photograph and pattern

3. Antenna with linear array of circular waveguides and additional corrugations

The same waveguide network (bottom layer) can also be used for the realisation of the excitation coefficients in elevation plane feeding a linear array of circular waveguides. Of course, this radiating element characteristic has to be modified to get a sectorising pattern. This can be done by corrugations with optimized width and depth.

Fig.2 shows a photograph of the first prototype on the left and first measurement results of this antenna.

Fig.2: Circular waveguides with corrugations

4. Concepts of planar microstrip antennas

Well-known realisations for such types of antennas are microstrip antennas like broadband, slot-coupled patches /2/, but these solutions suffer from the fact that they need two layers. To reduce costs, a one-layer low cost substrate solution for the antenna in terms of a feeding network on one side and a slot layer on the other side is investigated.

Different shapes of slots to get a well-suited trade-off of backradiation of the slot, crosspolar discrimination, bandwidth etc. are actually examined.

Fig. 3 shows the slot and network layer of this planar microstrip solution. Actually, measurements results of this realization with an H-shaped slot are not available.

Other shapes like the bow-tie slot will also be investigated.

Fig.3: Slot- and network layer

5. Conclusions

Different concepts for sector antennas in the 10 GHz frequency range have been presented. First measurement results are compared. The aim of the presentation at the conference will be to give a comparison in terms of electrical performance, costs, design risk, mechanical parameters etc.

6. References

/1/ M. Munk, J. Christ, "Sector Antenna in Plain Waveguide Technology", Twelfth Int. Conference on Antennas & Prop., Exeter, 2003
/2/ Dr. H. Ansorge,K.-H. Mierzwiak, U. Oehler, "Antenna solutions for point to multipoint radio systems," ECRR, Bologna, 1996

The use of a triangularly arranged set of slits is evaluated in this paper as a means of reducing the size of a two-port, dual polarized patch antenna. Furthermore, the proposed design achieves dual frequency operation.

Forty slits, varying in height, are considered on the perimeter of a square patch, ten on each side, to reduce the operating frequencies. The aforementioned discontinuities effectuate 38.8 % reduction in the area of the antenna, compared to the area of a non-meandered, orthogonal patch operating at both 1.8 GHz and 2.1 GHz.

To achieve dual frequency operation the height of the central slits in the x and y directions, shown as h_x,h_y below, are unequal and the same applies to the difference in height between any two neighbouring slits, shown as difx,dify. The remaining geometrical characteristics of the slits, i.e. width w_s and distance d , are the same in both dimensions. The antenna feed points are located in the middle of two adjacent sides in order to improve the isolation between the two ports. Appropriate matching circuits, not displayed below, transform the input impedance at the aforesaid feed points to the standard, 50 Ohm, value. Some interim results approximating the frequencies of 1.8 GHz and 2.1 GHz are tabulated below.

Operating Frequency(GHz)	Reduction in length (%)	Reduction in frequency (%)	Total area reduction (%)
1.742(simulated) , 1.728(measured)	15.2	26.8	38.8
1.997(simulated), 1.995 (measured)	27.9	14.6

At this point it should be noted that the results displayed at the second column of the aforementioned table refer to non meandered square patches resonating at the indicated frequencies. Also the results of the "frequency reduction" column refer to a square patch without slits equal in size to the proposed antenna. Finally the area reduction has been determined with reference to an antenna resonating at two frequencies, namely 1.8GHz and 2.1GHz.

Simulated and measured results for the scattering parameters of an antenna resonating close to the desired frequencies are diplayed in the graph. A first deduction is that the small variation in slit height along the x and y direction does not disturb significantly the symmetry of the S₁₂ and S₂₁ parameters. Furthermore, the slight discrepancy between the simulated results (displayed with dashed lines) and the measured results (displayed with continuous lines) was caused by the limited number of meshcells which was used for the simulation, because of computer memory constraints.

Finally, the antenna can easily be made to resonate at exactly the frequencies of 1.8 GHz and 2.1 GHz (or any other desired pair of frequencies satisfying the inherent geometrical restrictions imposed by this design) by changing the height of the slits along the x and y dimension. Final simulated and measured results will not deviate more than 10 MHz from the goal frequencies.

ReferencesBR>1. "Dual polarized microstrip patch antenna, reduced in size by use of peripheral slits"Notis, D.T.; Liakou, P.C.; Chrissoulidis, D.P., 7th European Conference on Wireless Technology 2004, Page(s): 273 - 276

GPS applications require uninterrupted reception regardless of the position of the receiver. Therefore the employment of a circularly polarised antenna is required. Some advantages can be achieved with the integration of the antenna to e.g. clothing. Above all, several different antenna configurations can be considered and an antenna exceeding the performance of the antenna integrated to the receiver can be designed. The main challenges related to this solution include the small permittivity and low height of the antenna substrate (garment), the selection of materials and method for the manufacturing of the antenna element into a flexible and elastic substrate, and the effect of the human body to the antenna performance. As these problems are solved, a lightweight antenna transparent to the user can be achieved.

Flexible antennas for GPS reception have been designed and their performance on two different substrates has been evaluated. Two antenna types, a corner-truncated square patch and a planar inverted F-antenna were studied. Antenna patterns were manufactured using conductive pastes and silk screen printing technique. Different mixing ratios of silver and carbon pastes were evaluated. It was found that the performance of 25/75% carbon/silver paste mixture does not significantly differ from that of 100% silver paste, whereas the losses of 50/50% mixture are considerably higher.

Antenna substrate was constructed by pressing a textile plastic on both sides of the fabric. One side of the fabric was covered with one layer and the other side with two layers of the plastic in order to increase the thickness of the substrate. Two different fabrics were tested – Gore-Tex and cotton. The permittivity of the antenna substrate was specified using a parallel-plate resonator measurement. Applying the three-layer textile plastic on Gore-Tex, relative permittivity of 2,3 and thickness of 1,3 mm were achieved. The evaluation of the designed antenna elements showed that the accuracy of the permittivity measurement was adequate (measured return loss for patch antenna on two substrates are represented in Figure 1). Axial ratio < 1 dB was achieved with the studied structures (measured radiation patterns for patch antenna on Gore-Tex are represented in Figure 2).

Figure 1 Measured return loss for a patch antenna fabricated on Gore-Tex (Gore) and cotton (Cott) using silver paste (100% Ag) and silver/carbon paste mixture (75% Ag / 25% C).

Figure 2 Measured radiation patterns for a patch antenna fabricated on Gore-Tex using silver paste (100% Ag) and silver/carbon paste mixture (75% Ag / 25% C).

I Introduction

The understanding and the evaluation of the mutual electromagnetic coupling between antennas arranged in array configurations are of paramount importance in the study of phased array (PA) antennas. The scanning capabilities of PAs are strongly influenced by the mutual couplings between different elements of the array. Effects such as the scan blindness [1], the gain drop observed as the antenna beam is steered from the broadside direction and the (strong) variations of the individual antenna input impedances with the scaning angle variations could be predicted and prevented after a deep understanding of the underlying physical phenomena.

In the case of arrays consisting of microstrip patch-like antennas (MPAs) the amount of energy coupled from one antenna to another depends on many parameters. In particular, it depends on the properties of the dielectric substrate on top of which the antennas are printed.

This contribution describes an experimental approach to identifying the contribution to the mutual couplings due to the direct interaction through the dielectric substrate hosting the array. The cavity enclosure of the individual antennas is experimentally demonstrated to be a viable solution for reducing the mutual couplings. By resorting to the printed circuit board (PCB) manufacturing process the cavity enclosure of the antennas is achieved without introducing any significant manufacturing complication.

II Experimental approach

The experimets have been carried out on two non-uniformly distributed sets of antennas (see Fig. 1). The spatial distribution of the patch antennas is the same in the two cases. In the first case, standard U-slot patch antennas are employed (see Fig. 1(a)) while in the second case the U-slot patches are enclosed in metallic cavities (cavity-backed U-slot patch antennas - CUP) whose side walls are mimicked by rows of equally-spaced metal-plated through-holes (see Fig. 1(b)). The measurements of the transmission coefficient between the feeding points of couples of antennas gives a measure of the electromagnetic coupling between the antennas under consideration. An example of the frequency dependance of the aforementioned transmission coefficient is presented in Fig. 2. It can be easily observed a significant reduction of the mutual couplings due to the presence of the cavities.

References

1. D. M. Pozar, and D. H. Schaubert, "Scan Blindness in Infinite Phased Arrays of Printed Dipoles," IEEE Trans. Antennas Propagat., Vol. AP-32, No. 6, pp. 602-610, June 1984.

Fig. 1: Non-uniform array configurations: (a) U-slot patch antennas; (b) cavity-backed U-slot patch (CUP) antennas.

Fig. 2: Mutual coupling between the 10-11 pairs of U-slot patch and CUP antennas (spacing 15 mm).

Portable FM receivers often come with a short built-in antenna. Sometimes the earphone cable acts as an antenna. These structures are often inefficient and poorly aligned, degrading reception quality. Well-opened half-wave antennas would improve reception, but they are inconveniently large to carry (1.5 m at the FM broadcast band).

The antenna can be made flexible and integrated into worn or carried equipment. The integration of the antenna into a garment eases the use but introduces a new problem: the vicinity of a human body alters antenna properties such as resonant frequency, bandwidth, and radiation pattern. This is due to the high dielectric constant of the human body, ranging from 30 to 50. The effect is difficult to simulate, as the density of the calculation mesh increases rapidly with increasing dielectric constant, and thus this study is forced to rely on measurement results.

The power received by the antenna may be transferred into the receiver either by simply plugging the antenna into the receiver, or by employing a transmission line coupler.

In this study we present wearable antennas for FM reception. The structures are planar, enabling the antenna to be printed on a garment. The antennas are specifically designed to work near the human body. Various structures, such as dipoles, meandered dipoles, and loop antennas, are considered and measured.

At the FM broadcast band (around 100 MHz) the height of the human body is nearly a half-wave in free space. Thus, the body can be seen to work in the resonance region. Thus the height and alignment of the user and the placement of the antenna on the body affect the behaviour of the antenna. The antenna structures are measured worn by different people posing in various manners. The wearers are chosen from both sexes in such a manner that the effect of varying height and weight can be studied. The variation in body alignment is taken into consideration by measuring the values with the testee standing, sitting, and with outstretced arms. All measurements are performed with varying antenna position on the body, and both in horisontal and vertical polarisation.

Measurement results for resonant frequencies, impedance bandwidths, and radiation patterns for various antennas are given and discussed. The effect of the human body on the antenna parameters is studied by comparing free-space results with those measured when the antenna was being worn.

Microstrip patch antennas are attractive elements at microwave frequencies due to their advantageous features in terms of low profile, low cost and lightweight. Researches on antennas for minisatellites SSETI - Express and ESEO and for radioamateur communication with International Space Station (ISS), that was carried out by our team, have shown that microstrip antennas can generate high quality circular polarization within a broad angle of the main beam, and achieve broad impedance bandwidth (about 12% with RL better than -10 dB). One of the main disadvantages of the patch, however, is low gain (typically not higher than 8 dBi). Traditionally, reflectors would be considered to achieve high gain solutions, nevertheless their overall size and unwieldy nature make these antennas unattractive. Other structures, commonly used for high gain applications, are arrays, but they usually suffer from high feed network loses. Therefore I have decided to concentrate my studies on gain enhancement methods for single antenna element.

My research confirmed that gain of the antenna can be increased significantly by setting a parasitic patch above fed element. However, in some cases it can be difficult to achieve good impedance matching. Also a dielectric superstrate layer above a microstrip patch antenna has remarkable effect on its gain. Major disadvantage of those methods is remarkable increase in antenna height. Use of dielectric material with high permittivity can also narrow impedance bandwidth. I have performed full-wave analysis of such structures using commercial tool - Ensemble of Ansoft. The most interesting antennas were produced and measured. My research have shown great advantages of combining those methods. If we put a dielectric layer in the space between the patches the distance between them can be reduced.

During my studies I put special attention to PBG (Photonic Band Gap) materials. They are periodical structures composed of metallic and dielectric elements. The PBG structure has a frequency range (gap) within which electromagnetic wave propagation is forbidden in one, two or three directions (1D, 2D or 3D PBG). Using that we can greatly enhance antenna gain by reducing distribution of surface waves - a main cause of loss in gain. The other very important characteristic is the ability to achieve a resonant frequency inside the forbidden bandwidth by introducing a defect into the PBG structure. I performed electromagnetic simulations for different types of PBG antennas including antennas on PBG substrate, antennas with PBG superstrate and defect resonator antennas.

I considered also other gain enhancement methods such as dielectric lenses, shorted metallic patches placed in a proximity of the patch and antennas with slotted ground planes. In my paper I will present results of simulations and measurements of investigated structures and compare considered means of gain enhancement at the patch level.

1 - Introduction

We propose here an effective and simple design methodology for the impedance matching of aperture-coupled microstrip antennas. Fig. 1 compares the classical approach and the proposed one. In the first case (Fig. 1a), the antenna design usually requires a large number of full wave electromagnetic (EM) analyses because of the cross-dependence between the dimensions of the radiating elements, the coupling slot and the feed network. Therefore such an approach is time consuming even though fast antenna design softwares are now available.

The new methodology proposed here is schematised in Fig. 1b. Although applicable to any one-port slot-fed antennas, its implementation is illustrated in Fig. 2 by considering a simple patch antenna.
The basic concept consists in substituting the open-ended feeding microstrip line by a two-port one, and in computing the resulting scattering matrix [S]. The procedure comprises two steps:
(1) Design and optimisation (EM simulations) of the radiating element and coupling slot by maximizing the radiation loss ( i.e. 1 - |S₁₁|² - |S₂₁|² ),
(2) Application of circuit theory to design the feed network, substituting the antenna by the [S] matrix obtained in step (1).

2 - Applications

Our methodology allows a significant reduction of the total number of EM simulations using simple microwave circuit theory for the feed network design. Moreover, it can be used to assess the antenna performance before the final feed network design. This approach is particularly relevant when designing (i) wide band planar antennas using a frequency domain formulation or (ii) directive EBG (Electromagnetic Band-Gap) resonator antennas for which impedance matching is a well-known and not really solved challenge. In the final paper, our impedance matching method is described in detail by designing two multi-layers planar antennas:
(1) A 30%-bandwidth stacked patch antenna at 7GHz (MoM analysis) whose final reflection coefficient is presented in Fig. 3,
(2) A set of EBG resonator antennas excited by an embedded slot-fed patch antenna. The resonant cavity is made from the ground plane of the feed and an upper partially reflecting surface ( PRS ). Several EBG antennas with various directivity ( i.e. reflectivity of the PRS ) are designed in Ku_band. Our procedure for impedance matching is successfully applied for low, moderate and high Q resonator. As an example, Fig. 4 compares the S₁₁ computed with the circuit analysis and a global FDTD simulation. These results demonstrate the relevance of our design methodology.

3 - Conclusion

An effective and simple design methodology for the impedance matching of aperture-coupled microstrip antennas has been presented. Two applications of this technique have been proposed : a 30% bandwidth stacked patch antenna and EBG resonator antennas excited by an embedded slot-fed patch antenna.

In this work a new method to suppress the harmonic radiation from a square electromagnetic coupled microstrip patch antenna is proposed. The goal of the design is the elimination of the resonances at the 2nd and 3nd harmonic frequencies to reduce spurious radiation. The study shows the possibility of controlling the second harmonic resonance matching varying the inset of the microstrip line. The size of the feeding microtrip line can be tuned to maximize the return loss for the second resonant frequency. In order to suppress the third harmonic a resonator is placed underneath the antenna’s feeding line. The resonator consists of a printed patch with a via connected to the ground plane (mushroom type) in a multilayer configuration.

Experimental results are in good agreement with the simulations. The input return loss measured indicates that the structure is effective for the reduction of the spurious radiation and it is demonstrated how the application of tuning length technique and mushroom-cell eliminates the second and third harmonics. Besides the loaded feeding line improves the return loss level of fundamental antenna mode and retains a good radiation performance. Since the resonator is very small and it is placed under the patch metallization a compact design is achieved, given that no extra surface is needed to introduce the filtering cell. So the simplicity of this compact low harmonics design, has potential application in active microstrip antennas for communications systems.

Figure 1 shows the sketch of the printed antenna with the mushroom-cell and simulated results for the reference patch and for the complete structure are presented in figure 2.

[1] G. Splitt and M. Davidovitz, "Guidelines for Design of Electromagnetically Coupled Microstrip Patch Antennas on Two_Layer Substrates", IEEE Tran. on Antennas and Prop. vol 38, n°7, july 1990, pages 1136-1140

[2] Y. Horii and M. Tsutsumi, "Harmonic Control by Photonic Bandgap on Microstrip Patch Antenna", IEEE Microwaves and Guided Wave Letters, vol 9, n°1, January 1999

[3] Y.J. Sung and Y.S. Kim, "An Improved Design of Microstrip Patch Antennas Using Photonic Bandgap Structure",IEEE Transactions on Antennas and Propagation, vol 53, n°5, may 2005