High efficiency and circular polarization in SATCOM phased arrays using tri-ridge apertures

Abstract

A new antenna array architecture is proposed here to overcome a current bottleneck in phased arrays concerning the enhancement of radiation efficiency and avoidance of scan blindness. In contrast to the conventional approach of using patches with circular or square geometries that exploit 90$^\circ$ or 180$^\circ$ rotational symmetry, this work proposes to use waveguide radiating elements based on 120$^\circ$ symmetry. To implement such symmetry, the tri-ridge aperture is proposed, and its capability to scan widely within a broad frequency bandwidth is demonstrated. A modal analysis is performed to explain the physical phenomena underlying such superior performance. A successful experimental validation is provided by means of a monolithic prototype built in metal additive manufacturing. The measured results show, for the first time, that it is possible to achieve high radiation efficiency and circular polarization when scanning widely in a broad bandwidth. Such an accomplishment constitutes a major achievement in the field of active electronically steered phased arrays, and impacts significantly the capabilities and potential of modern radar and communication systems.

Introduction

Phased arrays are key elements in modern communication systems due to their capability for electronic beam steering and their implementation as Active Electronically Steered Antennas (AESAs) is playing a crucial role in the advancement of satellite communications (SATCOM) technology, since they allow real-time optimization of the system’s performance^1,2. In the context of the upcoming networks (i.e., OneWeb, Starlink, Amazon Kuiper, Telesat Lightspeed³), AESAs are the brick that will guarantee ubiquity, mobility and high-date rate, for both ground users and satellites. AESAs radiating panels are typically built as two-dimensional arrays of elementary radiators, and they enable agile beamforming by precisely controlling their individual feeding in phase and amplitude. Such panels must comply with several key features. Firstly, circular polarization (CP) is commonly needed to enable simple and mobile user terminals that do not face alignment issues. Secondly, the arrays should avoid grating lobes (GLs) and blind spots while scanning within a wide field of view (FoV) to assure ubiquity. And finally, high gain is needed in order to enable high values of EIRP (equivalent isotropic radiated power) and G/T (gain over temperature noise ratio), which are the system parameters that correspond to high throughput⁴.

Figure 1

Computed value of G/T as a function of the loss per radiating element and the number of elements in the array $N_E$⁴. Parameters for calculation (considering Ka-band scenario): directivity of the array assumes 5 dBi unit cell and 10log10($N_{E}$), mismatch loss = 0.2 dB, physical temperature of the antenna = 290 K, antenna noise temperature = 20 K and receiver noise figure = 2 dB (excluding antenna loss).

The current trend in AESA design is to choose an antenna technology that provides CP and wide FoV scanning, and to adjust the number of elements in the array in order to reach certain gain (and associated G/T and/or EIRP). As an illustrative exercise, Fig. 1 shows a theoretical G/T computed for arrays with different number of elements. If one sets 8 dB/K as reference value for boresight performance, an array whose RF loss is 1.5 dB (all loss between antenna aperture and input of the beamforming circuit), would require around 800 elements to reach the target, while an array whose RF loss is 3 dB would require around 1500 elements. Such loss values are typical average RF loss values for state-of-the-art examples using respectively dielectric patches or liquid crystal^5,6,7. Applications such as commercial aviation in Ka-band require G/T values that are as high as 15 dB/K⁸ and would require a much larger number of elements. Unfortunately, the consequences of increasing the number of elements are not negligible: size and/or power consumption and/or cost will be AESA characteristics strongly influenced by such an AESA construction approach. Figure 1 also suggests that for a fixed number of elements, it is also possible to increase the G/T by reducing the antenna RF loss. To do so, we propose to move towards waveguide technology. The target of 8 dB/K could be also reached with around 500 elements using a waveguide aperture with an RF loss of 0.5 dB. Up to now, waveguides are not used in wide-angle scanning AESAs because efficient miniaturization approaches have not been found. This paper aims at bridging this gap by proposing a new concept based on 120° symmetry, unlocking for the first time high-radiation efficiency and CP in SATCOM AESAs.

Regarding CP AESAs in the literature, a classical design mindset is found in the majority of examples: the use of radiating elements with 90° or 180° rotational symmetry. Such is the case of microstrip patch arrays for SATCOM ground terminals^{9,10,11,12,13,14} or for Low-Earth Orbit (LEO) satellites^15,16, horn aperture arrays for Geostationary or Medium-Earth orbit satellites^17,18,19,20 and ultrawideband connected arrays^21,22,23,24. Most CP AESA elements have square or circular shapes, thus supporting two degenerate modes that induce radiated horizontal and vertical electric field components. Elements presenting other orders of symmetry have been barely explored in the literature, mainly because of their increased manufacturing complexity and the fact that the generation of CP may seem less intuitive. In particular, elements presenting 120° rotational symmetry can also enable radiation of CP or dual-linear polarization, since they can also support two fundamental modes with orthogonal polarizations^25,26. The examples in the literature implementing such symmetry in AESA elements are very scarce and they consider patch-based radiators²⁷.

Waveguide aperture radiators (namely horns when they are electrically large) serve as the antenna workhorse for applications prioritizing RF power loss minimization²⁸. This technology is particularly convenient at millimeter-wave frequencies, such as Ku- or Ka-bands, where the generation of RF power is less efficient compared to lower frequencies^5,29. Waveguide AESAs with high radiation efficiency have been proposed in the literature, but they consist of arrays with large apertures that have very limited scanning capability (lower than $\pm 25^\circ$) due to the appearance of grating lobes^17,18,19,20. Expanding the FoV relies on reducing the aperture size, prompting the need for solutions to address a well-known issue: the occurrence of scan blindness²⁸. Such an anomalous phenomenon prevents AESAs from radiating at specific angles within the FoV due to the excitation of spurious modes. The integration of impedance matching external sheets (or WAIMs, for wide angle impedance matching) and other bandgap structures offer a promising solution to mitigate scan blindness^30,31. However, most of the solutions exploiting such external surfaces only operate under single linear polarization^30,31,32. Moreover, they bring significant drawbacks, as increased complexity in design, fabrication, assembly and deterioration of the AESA polarization and radiation efficiency due to their non-negligible insertion losses.

The present contribution demonstrates the strong potential of array antennas based on a 120$^\circ$ symmetry to solve the problem of scan blindness without the need for external matching layers. Throughout this paper, the methodology employed to justify the physical operation of radiating apertures is based on a modal analysis approach. For the sake of clarity, the case of large apertures is firstly analyzed in order to explain how CP can be generated by exciting two degenerated modes. Canonical apertures like a square and a circle are considered, but also triangular (the latter, with 120$^\circ$ symmetry). Then, aperture miniaturization is considered by means of the well-known strategy of employing ridges³³. A comparative modal analysis is carried out for apertures loaded with four and three ridges, presenting respectively 90° and 120$^\circ$ rotational symmetry. This study shows that the bandwidth of the quad-ridge waveguide is drastically reduced as the size of the aperture is miniaturized. On the contrary, the tri-ridge case presents a bandwidth that can reach almost an octave. It should be noted that antennas based on waveguides loaded with three ridges have been already proposed to conceive a different type of electrically large leaky-wave antenna³⁴, thus not conceptually overlapping with the configuration presented here.

The analysis of miniaturized apertures allows to explain the phenomena of scan blindness. By comparing the quad-ridge geometry to a canonical empty square²⁸, it is shown that the origin of scan blindness is different in each case and that it is not possible to avoid scan blindness while preserving the 90° symmetry. Subsequently, the case of arrays built with tri-ridge apertures is presented, showing their capability to solve the previous problems and to scan widely and to generate CP within a broad frequency bandwidth. Such performance is specially well suited to SATCOM applications, and the concept is finally validated experimentally considering a radiating panel for Ku-band LEO spaceborne AESAs.

The article is structured as follows: “Introduction” explains the generation of CP in 120$^\circ$ symmetric apertures in comparison to canonical square and circular ones; “Waveguide apertures: generating CP and miniaturization” presents the modal analysis of ridged miniaturized apertures; “Scanning limitations in aperture arrays with 90$^\circ$ rotational symmetry” discusses the issue of scan blindness in sub-wavelength 90-degree symmetric apertures; “Wide-angle scanning with tri-ridge apertures and 120$^\circ$ rotational symmetry” presents the blindness-free performance of tri-ridge phased arrays; “Experimental validation using 3D-printing and discussion” provides experimental validation of a design example for LEO spaceborne AESA radiating panel; Finally, “Conclusion” concludes the article by summarizing the findings and exploring potential applications for this innovative radiating element.

Waveguide apertures: generating CP and miniaturization

The generation of CP radiation from apertures has traditionally relied on waveguide horns with square or circular geometries. Those waveguide shapes can support two orthogonal modes that generate horizontal (H-pol) and vertical (V-pol) radiated field components, respectively³⁵. In this subsection, the canonical cases (square and circular) are analyzed together with an equilateral triangular waveguide horn, which is characterized by 120° symmetry (which as been barely explored in the literature^25,26,36,37).

Figure 2

Analysis of three waveguide horns using with square, circular and triangular waveguide cross-section. (a) Horn geometry and modal distribution of the two fundamental modes at the input port. (b) Normalized directivity CP radiation pattern for each horn in the planes $\varphi =0^\circ$ and $\varphi =90^\circ$ (co-pol is RHCP, and cross-pol is LHCP).

Figure 3

Quad-ridge and tri-ridge circular waveguides: (a,c) The field distribution of the three first modes, (b,d) show the cutoff frequency of the three first modes as a function of the height of the ridges (h/R).

Figure  2 presents the three radiating apertures under consideration. The three horns have equal radiating surface, $4\lambda _0^2$ ($c=\lambda _0/f_0$, c being the speed of light in free space and $f_0$ the operating frequency). Figure 2a shows the geometry of the three horns and the E-field distribution of the two fundamental modes at the input ports. The modes in the square and circular cross-sections are the well-known TE$_{01}$ and TE$_{10}$ (square waveguide) and the TE$_{11}$-H and TE$_{11}$-V (circular waveguide)³⁸. In both cases, the orthogonal modes share their dispersion characteristics and present degenerated field distribution with 90° rotational symmetry. On the other hand, the modes in the triangular waveguide (M1 and M2) are less intuitive: they are mathematically orthogonal to each other, they have the same cutoff frequency and share the same dispersion³⁶, but they do not present the same field distribution. Figure  2b shows the CP radiation patterns. These patterns have been computed through full-wave simulation and considering that Mode H and V at the ports are excited in quadrature (this type of excitation is possible by means of an OMT (ortho-mode transducer), even for the triangular case^39,40. In Fig. 2b, the co-polarized patterns correspond to right-hand circular polarization (RHCP) and the cross-polarized ones correspond to left-hand circular polarization (LHCP). It is also worth noting that the square and circular horns present a pattern that is symmetric with respect to $\theta =0$° in both $\varphi$ planes, while in the case of the triangular horn, symmetry only appers for $\varphi =0$ °. The pattern symmetry is consistent with the symmetries of the waveguide modes (Fig. 2a). It is worth mentioning that the horns in Fig. 2a have not been particularly optimized to present any specific pattern shape, since the goal of this exercise is to demonstrate that all horns can produce CP with good (and comparable) polarization purity. These horns are electrically too large and cannot be used in wide-FoV AESAs. In order to enable AESAs having the requirements discussed earlier, it is necessary to miniaturize the apertures so that they can operate well above cutoff while being grouped in arrays with sub-wavelength periodicity (typically close to $\lambda _0/2$ when aiming at scanning close to endfire²⁸).

Two main strategies can be distinguished in the literature to miniaturize waveguide apertures⁴¹: loading them with dielectric materials, or with metallic ridges. The first one increases the aperture effective length as a result of the dielectric permittivity, while the second one results in an elongation of the aperture perimeter as a function of the length of the ridges. Both strategies allow to reduce the cutoff frequency of the fundamental modes in the aperture while keeping their periodicity constant. While both approaches are valid, this article seeks for alternatives to avoid dielectric materials in order to maximize radiation efficiency, therefore only the second design technique will be discussed on the remaining of the article. Phased array apertures using apertures with one or two ridges can be found in the literature, however they only operate with linear polarization^42,43,44.

Figure  3 presents two examples of circular waveguides that are miniaturized through ridges: a waveguide with four ridges following 90$^\circ$ rotational symmetry, and a waveguide with three ridges following 120$^\circ$ rotational symmetry they will be named as quad-ridge and tri-ridge waveguides in the remaining of the article. Figure  3 shows the E-field distribution and the evolution of the cut-off frequency with respect to the ridge penetration for the two fundamental modes (M1-Q/T, M2-Q/T) and the first higher-order mode mode (M3-Q/T).

In Fig. 3a, one can see that M1-Q and M2-Q resemble the modal solution of the circular aperture shown in Fig. 2a, except for the fact that M1-Q and M2-Q are mainly concentrated in the space between the four ridges. Fig. 3b shows that the cutoff of M1-Q and M2-Q decreases as the ratio h/R increases thus showing how the penetration of the ridges enables miniaturization of the waveguide. The behaviour of M3-Q is different: the cutoff increases until $h/R < 0.2$, where it reaches a maximum, and then it decreases and comes close to M1-Q and M2-Q when h/R is further increased. In practice, this implies that single-mode bandwidth of this waveguide is very narrow for electrically-small cases. This problem was already studied in other works^33,42,45.

Figure 3c shows M1-T, M2-T and M3-T. As it happened between the quad-ridge and the empty circular waveguides, the field distribution of M1-T and M2-T is very similar to the modal solution supported by the triangular waveguide (see Fig. 2a), again except for the fact that M1-T and M2-T are concentrated in the space between the three ridges. Figure 3d shows the evolution of the cut-off of the three first modes. The two fundamental modes evolve in a similar manner to those in the quad-ridge waveguide, which implies that miniaturization is also possible with this waveguide. On the other hand, M3-T presents a very different behaviour, since it does not come close to the fundamental ones for any ridge penetration. In practice this implies that the single-mode bandwidth is wide. For the sake of comparison, quad-ridge and tri-ridge waveguides characterized with $f_c^h/f_c^0 = 0.5$ would exhibit a single-mode bandwidth of $BW = 5.77\%$ and $BW=96.4\%$, respectively. This fractional single-mode bandwidth is calculated as $BW=2(f_c^{\text {M3}}-f_c^{\text {M1}})/(f_c^{\text {M3}}+f_c^{\text {M1}})$. It should also be noted that some of the authors of this paper have already demonstrated that it is possible to overcome the issues of M3-Q through evanescent waveguide sections, yet the bandwidth (limited to 10%) is not wide enough to address the majority of SATCOM AESAs⁴⁶.

Scanning limitations in aperture arrays with 90$^\circ$ rotational symmetry

Two-dimensional scanning arrays are generally limited by their periodicity due to the excitation of Floquet harmonics, since they drain energy from the desired main lobe and create undesired GLs. Another additional undesired effect related to GLs may appear: the so-called scan blindness. One may think that by reducing further the periodicity of the array (miniaturizing the radiating elements), such problems will be alleviated since the appearance of GLs is shifted towards higher frequencies. This section analyzes the physical phenomena behind such limitations and explains why it is not possible to overcome them if the 90$^\circ$ symmetry is preserved in the array.

For the sake of benchmarking, two study cases are considered: a quad-ridge aperture and the reference case presented in a seminal handbook about phased arrays²⁸. The latter consists in a sub-wavelength square waveguide aperture arranged in an infinite periodic square lattice. Square aperture and lattice are inherited from the well-established reference²⁸, but similar conclusions would be obtained with other shapes respecting the 90° symmetry. Figure  4 shows both the performance of the reference square (first row), and the quad-ridge (second row) apertures. As it is explained in the previous sections, both support two fundamental modes and can generate CP. Each column in Fig. 4 deals with a polarization case: H-pol (x-axis aligned), V-pol (y-axis aligned), and a combination of both with an OMT in order to produce RHCP. Each of the plots show the active realized gain (embedded pattern²⁸) of the apertures for each polarization when the array is fed with a phase law that produces a beam scanning in $\theta$ from 0° to 60$^\circ$ in $\varphi = 0^\circ$ plane. The ordinate axis shows how the realized gain changes with frequency, considering a value that is normalized with respect to $f_c^{\text {M1}}$, which is the cutoff of the fundamental M1 mode in the waveguide input port. For a fixed frequency of operation $f_0$ (the horizontal line $f/f_c^{\text {M1}}=1.1$ in the plots), the square and quad-ridge arrays present a periodicity of 0.7$\lambda _0$ and 0.54$\lambda _0$, respectively. The latter is smaller thanks to the miniaturization effect brought by the ridges, as explained in the previous section. Moreover, these heatmaps also display white dashed lines representing the cutoff frequencies of the first GL and of the first higher-order mode in the aperture (M3). Note that the cutoff of the GL depends on the periodicity of the array and decreases with $\theta$²⁸, while the cutoff of the M3 mode only depends on the geometry and dimensions of the waveguide, thus remaining as an horizontal line in all graphs.

Figure 4

Active realized gain for H-pol, V-pol, and RHCP of two infinitely periodic aperture arrays when scanning to an angle $\theta$ along the $\varphi = 0^\circ$ plane. A hollow (i.e. non-ridged) square aperture is considered in (a–c), and a quad-ridge one in (d–f), both arranged in square lattices. The geometry and size of the apertures are depicted as an inset in the first figure of the row, being $\lambda _0$ the wavelength at $f_0=1.1f_c^{\text {M1}}$. The white dashed lines named as GL and M3 show the cutoff frequencies of the first higher-order Floquet harmonic and the first higher-order mode in the apertures input, respectively. All results have been computed with a commercial full-wave solver.

All graphs in Fig. 4 show that, when $\theta =0^{\circ }$ (array pointing at boresight), the gain increases with the frequency, as it could be expected since the apertures become electrically larger³⁵. Note that the antennas start radiating when operating above the cutoff of the first mode in the input of the horn (i.e. when $f/f_{c}^{\text {M1}}=1$). By inspecting the first row in Fig. 4, it is possible to analyze the scanning capabilities of a hollow sub-wavelength aperture. Let us first consider the scanning for a fixed frequency (i.e. moving along an horizontal line in the heatmaps), for example at $f_0/f_{c}^{\text {M1}}=1.1$. For V-pol (Fig. 4a), it can be observed that the gain remains almost constant until reaching the dashed line corresponding to the cutoff frequency of the GL. At such point, the realized gain experiences an inflection point, and it starts to decrease when scanning further. As it is explained above, when the antenna operates above the GL line, secondary lobes appear creating diffraction in unwanted directions and reducing the gain of the main lobe. A significant difference is seen when analyzing H-pol (Fig. 4b): a strong drop of the gain is observed just before the cutoff of the GL (the dark blue area at the left side of the GL curve). A strong mismatch happens at the antenna input port when trying to scan the beam towards such direction. This is consistent with the benchmark studies²⁸ and it is called scan blindness. It is apparent that such blindness appears when scanning with H-pol but not with V-pol. Physical insight on this phenomenon can be obtained by considering the array aperture as a periodic structure and applying a Floquet expansion approach^47,48. When scanning with H-pol (i.e. the E-plane), the field couples to a Floquet harmonic of TM nature³¹, which brings a singularity to the impedance of the discontinuity thus inducing a peak of full reflection at a certain frequency close and below its cutoff. Total reflection does not occur in the V-pol since the field couples to a TE harmonic in this case (i.e. H-plane scanning) and does not generate such a resonance. This type of unexpected resonances in arrays of sub-wavelength elements were named as “extraordinary” transmission/reflection in the 90s⁴⁹. When the two polarizations are combined to create CP (Fig. 4c), blindness prevails, which makes the antenna hardly usable for scanning purposes. Note that the case of CP is not considered in the seminal reference²⁸ and it is here analyzed for the first time.

The second row in Fig. 4 considers the quad-ridge waveguide. The ridges miniaturize the aperture and enable operation with a lower periodicity of P$=0.54\lambda _0$ at the frequency $f_0$. By comparing to the upper row in Fig. 4, it can be observed that the GL is pushed towards higher frequencies and the white dashed line appears shifted towards the right on the lower row. On the other hand, as it is discussed in previous sections, the line associated to the M3 mode has now moved towards lower frequencies. As a consequence, though the array operates well below the appearance of grating lobes when scanning at a given frequency, the excitation of the M3 mode brings again a scan blindness effect for the H-pol scanning. In the case of V-pol, Fig. 4d, the problem does not occur because the scanning in the $\varphi = 0^\circ$ plane enforces anti-symmetry and does not enable the excitation of M3 mode. When both polarizations are combined to form RHCP, Fig. 4f, the antenna is again unusable due to the scan blindness effect coming from the H-pol. In conclusion, the miniaturization of the element has improved the scanning range but within an extremely narrow bandwidth. Note that one can analyze how this array operates with a periodicity P$=\alpha \lambda$ by looking at the performance at the frequency (horizontal line) corresponding to $f=f_0\alpha /0.54$.

Wide-angle scanning with tri-ridge apertures and 120º rotational symmetry

Figure 5

Analysis of a tri-ridge infinitely periodic aperture array when scanning to an angle $\theta$ along the $\varphi = 0^\circ$ plane. (a) Geometry of the aperture, with dimensions $h/R=0.77$, $P=0.73\lambda _0$ where $\lambda _0= c/f_0$ and ridge thickness 1.1 mm. (b) Cutoff of the GL in arrays with same periodicity along x axis, comparing triangular and square lattice. The periodicity is chosen so that at $f_0$ the GL appears when scanning at $\theta =50^{\circ }$ in the triangular lattice. (c–e) Active realized gain in horizontal, vertical, and circular polarization. The white dashed lines named as GL and M3, show respectively the cutoff frequencies of the first higher-order Floquet harmonic, and the first higher-order mode in the apertures input. All results have been computed with a commercial full-wave solver.

This section illustrates how the 120º symmetry approach allows to overcome the limitations highlighted in the previous section. First of all, by adopting a triangular lattice, it is possible to push further the appearance of GLs^31,50. Secondly, the use of tri-ridge apertures provides an effective means to avoid the early excitation of higher order modes in the aperture. Figure  5 shows the tri-ridge waveguide array and its performance in infinite array. An illustrative design is evaluated here with the criteria of launching the first GL at the frequency $f_0$ when scanning the main beam at $\varphi =0^\circ$ and $\theta =50^{\circ }$. Figure  5b compares the cutoff of the GL in arrays with triangular and square lattices having the same periodicity along the x-axis and scanning in the $\varphi = 0^\circ$-plane. The red dashed lines in that figure allow to validate the design criteria, since at $f/f_0$ the cutoff of the GL is found at $50^{\circ }$. It is obvious from Fig. 5b that the arrangement with the same periodicity in a square lattice would launch the GL at a much lower angle (around $20^\circ$). However, the change of lattice does not have any impact on the M3 mode, which is only determined by the aperture geometry (as justified in Fig. 3).

In analogy to the analysis performed in the previous section, Fig. 5c–e show the realized gain heatmaps for H-pol, V-pol and RHCP when scanning in the $\varphi =0^\circ$ plane. In the present representations, the frequency is normalized over the frequency $f_0$. As it happened in the previous antennas, radiation is induced through the aperture above the cutoff frequency of the fundamental mode, which now corresponds to the frequency $f/f_0=0.8$. When comparing the plots in Fig. 5 with those in Fig. 4, one important difference can be identified. While the GL line is pushed towards the right of each heatmap thanks to the triangular lattice (defined in Fig. 5b), the horizontal M3 line is kept at high frequencies (at around $f/f_0=1.38$) thanks to the aperture geometry. This way, the tri-ridge shape allows to effectively solve the problem observed in Fig. 4, and miniaturize the aperture size while avoiding the early excitation of the M3 mode. As a result, the single-mode bandwidth has increased significantly (which corresponds to the area where the realized gain is non-null and below the white dashed lines of GL and M3). Within this bandwidth, stable realized gain can be achieved in the array both at broadside and when scanning. As a consequence, Fig. 5e shows that by combining the H-pol and V-pol with an OMT, RHCP operation is enabled with enhanced performance in terms of bandwidth and scanning. It should be noted that only $\varphi =0$º is shown in Fig. 5 for the sake of brevity. However, the performance when changing $\varphi$ is very similar, and no new phenomena appears, as it will be shown in the next section.

For the sake of comparison, the 1 dB realized gain bandwidth for $10^\circ$ scanning is computed in all the considered aperture arrays. Such a bandwidth is marked with a red arrow in Fig. 4c, Fig. 4f and Fig. 5c  respectively for the square, quad-ridge and tri-ridge apertures. The resulting values are 3.9%, 5.7%, and 21.77%, respectively. This comparison demonstrates the significant  bandwidth increase brought by the 120º symmetry, which almost quadruples the bandwidth of the 90º symmetry designs. Therefore, it can be concluded that elements with a symmetry of 120$^\circ$ are particularly suitable for phased arrays with wide FoV since they allow to increase the array bandwidth by avoiding the problems that high-order modes can generate.

Experimental validation using 3D-printing and discussion

The tri-ridge waveguide design presented in Fig. 5 is here validated experimentally by fabricating and measuring a radiating panel operating between $[f_{low}, f_{high}] = [10.7, 12.7]$ GHz. For this demonstrator, we consider a meaningful scenario of a Ku-band SATCOM AESA for LEO satellites having an Earth FoV of $\theta _E=\pm 60^\circ$ and a user coverage of $\theta _U=\pm 50^\circ$, as illustrated in Fig. 6a. This scenario implies that, when scanning within the user overage, the first GL must always remain outside of the Earth FoV. By considering an array of triangular lattice, this need can be satisfied with an array with periodicity equal or smaller than $0.71 \lambda _{h}$²⁸ (where $\lambda _{h}$ corresponds to the wavelength at the $f_{high}$). Since the design of Fig. 5 is normalized with respect to $f_0$, it is therefore necessary to consider operation in the range $[f_{low}, f_{high}]=0.71/0.73 \times [10.7/12.7, 1] f_0=[0.82, 0.97]f_0$.

The fabricated antenna array is shown in Fig. 6b, and it consists of a finite arrangement of 167 elements. 3D printing has enabled monolithic prototyping of such complex piece (machine is EOS M290, build material is an Aluminium alloy, and the surface finish is Copper-plating in order to improve the conductivity^51,52). Each radiating element consists of the tri-ridge aperture described in Fig. 5 (aligned to the xy-plane), fed by a circular tri-ridge waveguide section. This section provides active impedance matching capabilities and it has five steps. The profile of the ridges has been modulated along z³⁴, resulting in a spline-like shape that multiplies the depth of each step by the factors [1, 0.8, 0.73, 0.69 0.99] (the first value corresponds to the step interfacing with free-space). This shape does not follow the typical profile of ridge horn antennas^43,46, and the reason is that the geometry optimization has considered the full scanning FoV. This strategy also results in a reduction of the operation bandwidth of the antenna at broadside. Increasing the complexity of the impedance matching section, as well as other optimization strategies, could be considered in order to enlarge the antenna bandwidth. Yet, the followed approach is found sufficient to satisfy the needs of a LEO SATCOM AESA. Finally, each radiating element integrates a septum OMT^39,40 in circular waveguide, where the two input ports have a single ridge that integrates a coaxial interface (waveguide-to-coaxial end-launch transition with mini-SMP ports). All ports in the array have been connectorized for characterization purposes. In an eventual AESA implementation (a sketch is shown in Fig. 6c), such connectors would not be necessary, and the signal would be launched in each waveguide element from electronic cards having the beamforming circuits⁵³.

Figure 6

(a) Definition of Earth FoV ($\theta _E$) and user coverage ($\theta _U$) for a LEO satellite (R is the Earth radius, h is the orbital height, $\varepsilon$ is the maximal user elevation). From “Earth-Global Elevation Model with Satellite Imagery (Version 2)” by Kevin M. Gill, licensed under CC BY 2.0. (b) 3D-printed tri-ridge waveguide aperture demonstrator. (c) Sketch of LEO AESA including aperture and beamforming network.

Figure 7

Comparison between full-wave simulations of an embedded element in an infinite array and the measurements of the central element of the fabricated array. The normalized realized gain is shown for the planes (a) $\varphi =0^{\circ }$ and (b) $\varphi =30^{\circ }$, and for the lowest and highest frequencies of the operation band. Co- and cross-polarization correspond to LHCP and RHCP: (-) LHCP (–) RHCP, (-) measured (-o-) simulated.

Figure 8

Measured realized gain of the central element of the demonstrator for several relevant planes and frequencies: (a) 10.7 GHz, (b) 11.2 GHz, (c) 11.7 GHz, (d) 12.2 GHz and (e) 12.7 GHz. Co- and cross-polarization correspond to LHCP and RHCP (continuous/dashed lines, respectively).

The experimental validation has consisted in measuring the radiation pattern of the central element with all other ports in the array terminated in a 50-ohm load. Figure  7 shows the comparison between the full-wave simulations and the measurements. Both results are in very good agreement, considering the pattern shape and also the levels of cross-polarization. The differences increase when moving towards high $\theta$ angles, and they can be attributed to the finite nature of the array. Additional results are shown in Fig. 8, which considers radiation patterns at other $\varphi$ cut planes and more frequencies. Due to symmetry, by studying the structure within a 60$^{\circ }$ angular range, one can deduce the properties over the entire 360$^{\circ }$ range, and measuring the $\varphi$ planes at 0$^{\circ }$, 30$^{\circ }$, and 60$^{\circ }$ is sufficient to characterize the array. The additional measurements at 120$^{\circ }$ and 150$^{\circ }$ provide redundancy and help confirm the symmetry and accuracy of the data. The patterns are quite flat within a cone of ± $40^\circ$ at all $\varphi$ planes and frequencies. For higher angles, a gain drop may appear, which is associated to the cutoff frequency of the GL. It should be noted that such harmonic is radiating away of the Earth FoV in all cases, and such excitation does not create scan blindness. For a fixed frequency, the gain drop appears at a different $\theta$ angle when changing the $\varphi$ plane. This is coherent with the theory²⁸, since the cutoff frequency depends on both $\theta$ and $\varphi$, and the lowest value corresponds to the $\varphi =30^\circ$ plane. Finally, it is also worth noting that the quality of the CP polarization generated by the tri-ridge apertures is very good, since the levels of cross-polarization are low, specially at high frequencies. The cross-polarization discrimination is comparable to 15 dB in most angles within the user coverage.

The measured realized gain at boresight goes from 4.5 to 6.3 dBi along the frequency range. This value includes 0.4 dB of setup losses, which are associated to the connector and other cables and transitions that are needed for the characterization. The simulated values for the realized gain are 5.2 and 6.7 dBi, respectively for $f_{low}$ and $f_{high}$, and assuming perfect electric conductor (see Fig.  5e). The conductivity associated to 3D printing and plating finish is $2.5\times 10^7$ S/m⁵². When combining the test setup uncertainties (evaluated in ± 0.25 dB) with all the above information, it is possible to experimentally estimate that the losses in the radiating element are below 0.2 dB. The array characterization also included return loss of the central element, and coupling to all other ports in the array. The active return loss was retrieved from such data and the resulting values remained below $-10$ dB within the FoV and the operation frequency bandwidth.

To the best knowledge of the authors of this paper, this experimental demonstrator represents both the first waveguide AESA front-end⁵, and the first AESA targeting LEO SATCOM payloads in the scientific literature. Indeed, in space payloads there is a strong interest to minimize the number of elements in order to reduce the size, weight and DC power consumption of the antenna, which goes in favor of maximizing the radiation efficiency. In addition, in space applications, the need for waveguide technology is even more relevant than in other domains because dielectric materials might suffer or degrade due to the space environment, which would reduce the life-time of the antenna. The tri-ridge antenna technology presented in previous sections enables compliance with all the previous requirements. The only potential disadvantage of this technology is its geometry complexity; however, existing additive manufacturing (i.e. 3D printing) technologies allow for the fabrication of complicated shapes in a monolithic and cost-competitive way.

Conclusion

This paper shows the importance of maximizing radiation efficiency in AESAs in order to minimize complexity, cost and power consumption. A solution is proposed based on the use of waveguide aperture radiators with 120º rotational symmetry. Firstly, this paper provides a modal analysis to justify how this type of symmetry allows to generate CP, considering triangular apertures. To comply with wide-angle scanning, miniaturization of the aperture with ridges is analyzed. Secondly, the scan-blindness phenomena appearing in CP waveguide apertures is analyzed, considering a well established benchmark from the literature. The origin of such spurious phenomena is explained by the excitation of higher order Floquet harmonics, and higher order modes of the aperture. Thirdly, a new solution is presented, based on tri-ridge apertures arranged in a triangular lattice. Its scanning performance is analyzed, showing that miniaturization of the aperture allows for scanning within a broad frequency range. This new, almost unexplored approach allows to achieve a significant increase in single-mode bandwidth and avoids the phenomenon of scan blindness, becoming a solution that overcomes the current phased array bottleneck. Finally, the proposed concept is validated experimentally in the context of SATCOM from LEO satellites. The measured results confirm the promising full-wave predictions, as well as the strong potential of the tri-ridge aperture for next-generation of AESAs.