Packable and Readily Deployable Tightly Coupled Dipole Array (TCDA) With Integrated Planar Balun

To overcome the limited payload dedicated to onboard antennas for CubeSat/SmallSat applications, this paper presents a novel, foldable, dual-polarized Tightly Coupled Dipole Array (TCDA). In comparison to previous vertical feeds used for TCDAs, this array integrates a planar microstrip balun feed to enable foldability. The proposed array attains 5.4:1 (0.6-3.20 GHz) impedance bandwidth with VSWR < 3 at broadside and scans down to 45° at all azimuth planes. Notably, a substrate-integrated in-plane folding mechanism based on Lamina Emergent Torsion (LET) joints is employed to achieve foldability coupled with a simple fabrication process. The average simulated radiation efficiency was 95% across the band. A <inline-formula> <tex-math notation="LaTeX">$5\,\, \times8$ </tex-math></inline-formula> prototype array was fabricated and tested to verify the finite array’s foldability, bandwidth, and gain performance. This prototype achieves 80% one-dimensional size reduction, yielding 60% overall volume reduction. Consequently, the array prototype can be folded and stowed in a compact volume of 1.4U (25 cm<inline-formula> <tex-math notation="LaTeX">$ \times7.2$ </tex-math></inline-formula> cm<inline-formula> <tex-math notation="LaTeX">$ \times9$ </tex-math></inline-formula> cm).


I. INTRODUCTION
T HE DEVELOPMENT of wideband, high-gain antennas plays a key role for several space technologies, including remote sensing/imaging, inter-satellite links (ISLs), and high data rate ground-satellite communications. Additionally, modern space and satellite communications require broadband coverage and high reliability with reduced size, weight, power, and cost (SWaP-C). Furthermore, for next generations satellite applications, phased arrays can increase gain and mitigate interference with beamsteering capability. Consequently, space technologies are eyeing conformal and thin arrays operating across a wide bandwidth. However, the implementation of traditional antenna arrays is restricted by limited launch volume [1]. Notably, for space applications the arrays must be compactly stowed to accommodate the limited allocated storage [2].
Recently, a rigid-flexible TCDA (RF-TCDA) design demonstrated foldability of a TCDA using a combination of a flexible Kapton substrate with rigid FR4 [16]. This RF-TCDA design uses a tapered transmission line balun at the dipole layer to achieve foldability. However, the employed parallel-strip tapered transmission line balun in [16] prevented dual-polarization operation and introduced higher cross-polarization.
To enable foldability of the spaceborne antennas, numerous techniques were presented in the literature. In [17], [18], mesh-surfaced reflector antennas were presented to achieve the foldability and deployability of the spaceborne antennas. However, this kind of antenna suffers from undesired surface roughness, leading to poor gain performance. In [3], [19], flexible membrane-based deployable antenna arrays were presented. The latter were very lightweight and compact for space applications. But their deployment process is complex and often causes damage to the membrane causing creases at the folding joints.
A rigid foldable technique was proposed in [2], [16], [20], where the rigid parts of the antenna were joined with a flexible substrate material (Kapton Pyralux) at the folding locations. This approach is challenging for fabrication, particularly when a continuous metal connection is needed across the flexible substrate. In comparison to the abovementioned approaches, an in-plane Lamina Emergent Torsion (LET) joint was considered in [21], [22]. Monolithic fabrication of this joint from a rigid substrate simplifies the fabrication process while still ensuring the rigidity of the design.
This paper addresses the existing issues associated with the design and fabrication of spaceborne, foldable, UWB antenna array and their feed. Specifically, a foldable tightly coupled dipole array (TCDA) is presented that employs a planar microstrip-to-slotline balun feed on the same plane as the antenna array. This is done by printing the microstrip balun on the flip side of the dipole substrate, thus preventing the feed from causing undesirable cross-polarization. Another notable advantage of this design is the co-phase-centered dual-polarization operation which can not be realized using the traditional egg-crate TCDA configuration. Notably, a substrate-integrated in-plane folding mechanism based on LET joint is employed to achieve foldability, which requires a simple and low-cost fabrication process. Consequently, the dipole and ground plane layers can be folded to significantly reduce array volume by 60%. The presented TCDA delivers a 5.4:1 (0.6 GHz to 3.2 GHz) bandwidth at broadside and scans down to 45 • off-broadside at all azimuth planes. It is noted that the array retains its performance even after several folding and unfolding cycles. Also, the total height of the TCDA is quite thin, with only a thickness of λ low /14 (where λ low being the wavelength at the lowest frequency of operation). This paper is organized as follows. Section II details the design of the proposed TCDA. The prototype fabrication and measurements are given in Section III. Conclusions and future work plans are presented in Section IV.

II. ARRAY DESIGN AND SIMULATION
The proposed dual-polarized array is shown in Fig. 2. This model was designed and simulated using Periodic Boundary Condition (PBC) set-up using commercial software Ansys HFSS [23]. As depicted in Fig. 2, the unit cell is comprised of two main sections: the array layer and ground plane layer. These layers are separated by a distance of d = 39 mm. This distance was carefully optimized to achieve a ground plane reactive impedance of Z GP = jη 0 tan(βd), where η 0 is the free-space impedance, and β is the propagation constant [10]. Hence, the ultra wide-band operation of the TCDA is achieved by counterbalancing the ground plane inductance with the capacitive coupling between the neighboring array elements. The array stack-up comprises four metal layers. Top to bottom these are: 1) frequency selective surface (FSS) superstrate, 2) microstrip feed for horizontal (H) polarization, 3) microstrip feed for vertical polarization (V), and 4) dipole layer. The key design detail and layer stack diagram of the TCDA unit cell is shown in Fig. 2. The final design parameters of the unit cell are given in Table 1.
Below, Sections II-A-II-D discuss the design guidelines of each section for the proposed dual-polarized TCDA.

A. ORIGAMI-BASED FOLDABLE UNIT CELL DESIGN
The unit cell design is depicted in Figs. 2 and 3. The dipoles are etched on the bottom-most layer of the two-layer stack-up. Rogers RT/duroid 5880 ( r = 2.2) was used as the substrate and the dipoles were designed for a maximum separation of λ high /2, where λ high is the wavelength at the highest frequency of operation viz. f high = 3.2 GHz. Doing so, scanning at low angles is achieved without grating lobes. Notably, bow-tie-shaped dipoles were chosen for improved bandwidth. The dipoles were designed and optimized to have a narrow tip at the feed to accommodate dual polarization concentrically. As noted, no overlapping pads were used between neighboring (dipole) elements since the focus was folding not the bandwidth beyond 7:1 [4], [24], [25]. This folding approach facilitated the use of LET joints placed at the dipole edges.
To enable folding, the unit cell was placed on an origamibased lattice as proposed in [16]. Referring to Fig. 2(c), θ F is the folding angle relative to the flat plane. Specifically, θ F = 0 • refers to the unfolded (flat) case, and θ F = 90 • refers to the fully folded case. From geometry, the unit cell lattice parameters can be expressed as follows: In the above, a and b are the side lengths of the unit cell substrate. Therefore, the total volume of the unit cell at any folding state is given by:

B. PLANAR MICROSTRIP BALUN FEED DESIGN
A planar microstrip-to-slotline balun was used for feeding the dipoles, as depicted in Figs. 2 and 3. This feeding facilitated the transition between the unbalanced microstrip and the inherently balanced slotline. This type of microstrip-toslotline balun feed has been previously used for archimedean spirals [26], [27], [28]. Notably, the antenna arm serves as a ground plane for the microstrip line [29]. As depicted in Fig. 2, the center pin of the 50 connector excites the microstrip line at the end of each dipole arm. Subsequently, the microstrip line travels toward the center of the dipole and transitions to a balanced slot. At the center of the dipole, the microstrip crosses the dipole gap to terminate at a radial stub.
The strategic positioning of the radial stub improves impedance matching. Additionally, perforations were added across the dipoles arm. These perforations control the capacitance between the microstrip and the ground plane (dipole arms) to achieve improved impedance transition. To further improve impedance matching, both the microstrip feed line and the perforations were meandered and tapered along the length of the dipole [13]. The evolution of the aforementioned feed design is depicted in Fig. 3.

C. FREQUENCY SELECTIVE SURFACE (FSS) SUPERSTRATE
In earlier TCDA designs, dielectric slabs were used to improve scanning performance [10]. However, these dielectric slabs added weight while providing low-angle scanning. To circumvent dielectric superstrate drawbacks, vertical FSS superstrates were presented in [4], [13], [30]. Instead, here we consider a planar FSS superstrate to enable the folding of the antenna array. Also, by optimizing the dipole to FSS superstrate height (t), width (w FSS ), and spacing (g FSS ) the scanning impedance is improved. That is, the FSS superstrate is tuned with the rest of the array to improve scanning across the operating frequency range. The final dimensions of the FSS superstrate are given in Table 1.

D. LAMINA EMERGENT TORSION (LET) JOINT
The proposed TCDA incorporated a compliant joint mechanism (surrogate hinge) using lamina emergent torsion (LET) FIGURE 4. LET joint folding mechanism is shown with design parameters. joint to achieve folding and self-deployability. Unlike the conventional hinge mechanisms, the LET joint achieves the required movement of folding through deflection of the material at the joint area. Thus, with the combination of bending and twisting, the LET joints provides rotational motion out of the plane to accommodate folding [31]. Notably, flexibility at the joints is achieved by reducing the stiffness through optimization of the LET joint cuts/slots. As depicted in Fig. 4, the LET cuts/slots are milled on a monolithic substrate to build out the LET joint. These LET joints employ parallel and series arrangements of torsional bars to form an m×n array. Here, m and n refer to the number of elements in parallel and series, respectively. The stiffness of a torsional  bar or element is given [21], [32] as In the above, k b and k t refer to the bending and torsional stiffnesses, respectively. Further, for a joint comprised of m parallel and n series elements, the overall stiffness of the hinges along the folding direction can be expressed as This expression shows that the overall stiffness of the hinge decrease with n, viz., the number of torsional bar connected in series.
Overall, optimal LET joint folding can be achieved through the combination of LET joint cut optimization and befitting materials properties, e.g., tensile modulus, E, Poisson's ratio, ν. As such, LET joint optimization and the material selections are discussed in the fabrication section.

E. INFINITE ARRAY SIMULATIONS
Using the optimized unit cell element in a periodic array, an infinite array simulation was carried out. As depicted in  Henceforth, taking advantage of the symmetry of the design, and since the H-pol has the high VSWR, only simulation and measurements for this polarization will be discussed for the rest of the paper. When scanning in both E-and Hplane down to 45 • , the unit cell still maintains VSWR< 3 across 0.6 GHz to 3.2 GHz. As expected, higher VSWR is observed in the H-plane due to the 1/cosθ impedance variation [33]. Notably, the employed planar FSS superstrate of square patches serves to retain VSWR<3 during scanning. However, VSWR for scanning in both E-and H-planes can be further improved by increasing the thickness/gap, t (see in Fig. 2) between the dipole and FSS superstrate.
Simulated port-to-port isolation for the two polarizations (H-and V-pol.) are given in Fig. 6, showing > 30 dB isolation, on average, between the ports at broadside. At 45 • scanning, the isolation is > 25 dB, on average. The simulated radiation efficiency, on average, was achieved to be ∼ 95% across the operating frequency band.

A. FABRICATED PROTOTYPE
A 5 × 8 dual-polarized array prototype was fabricated to verify the design. Specifically, the dipoles were printed on the bottom layer of a Rogers RT/duroid 5880 ( r = 2.2) substrate (thickness 10 mils). Notably, for the H-polarization, the microstrip feed line was printed on the top of the same 10 mil substrate. Further, the microstrip feed line for the other polarization was printed at the bottom layer of a separate Rogers RT/duroid 5880 substrate 60 mils thick and placed above the 10 mils dipole substrate. Both substrate layers were bonded together using Rogers 2929 bondply ( r = 2.9). Finally, the FSS elements were printed on top of the 60 mils substrate. The complete stack-up of the antenna array layers are given in Fig. 2(a).
The ground plane layer of the antenna array was printed on a separate Rogers RT/duroid 5880 substrate (thickness = 60 mils). Holes were also drilled in the ground plane. These holes were used to insert the coaxial cables for feeding the microstrip lines. Notably, a ferrite choke was used near the feeding point to prevent backward current density leaks at the outer conductor of the coaxial cable, and suppress common mode radiation. Further, to ensure the quality of the fabricated prototype, commercially printed PCB manufacturing guidelines were followed throughout the entire design process. That is, minimum metal trace widths and a conductor separation of 0.15 mm (6 mils) were maintained.

B. LET JOINTS OPTIMIZATION AND MINIMIZING THEIR EFFECT ON ANTENNA'S RF PERFORMANCE
A key aspect of the presented design is its robustness along with folding flexibility at the joint using LET cuts. From Fig. 8, it can be observed that Rogers RT/duroid 5880 and Rogers DiClad 880 substrates were a good fit to achieve the required robustness of the design with high tensile modulus [34]. Eventually, the low dissipation factor ( tanδ = 9 × 10 −4 ) of Rogers RT/duroid 5880 substrate makes it attractive for low-loss RF applications.
Optimization of the LET cut/slot array was carried out to minimize the Von Mises stress (resultant stress of torsional and bending stress) at the LET joint. This was done while minimizing the footprint of the joint area. We note that the torsional stress is proportional to W t and inversely proportional to L t . Also, bending stress is inversely proportional to W b [31], [32]. We remark that the minimum constraint for W t was the fabrication tolerance and the robustness of the torsional bars for multiple folding cycles. An additional constraint relates to maximum L t and W b was the width of the dipole end, W dipole (see Fig. 9) to ensure that the LET cut does not split the dipole arm. Furthermore, the stiffness of the joint is inversely proportional to the number of series torsional bars, n. However, a larger n increases the LET joint's footprint. As such, the number of series elements, n, should be chosen to optimize stiffness with a smaller LET joint footprint. Notably, the maximum Von Misses stress at the joint can be expressed as [32] In the above, t is the total thickness of the substrate, θ F,max is the maximum folding angle, and Q refers to a geometry-dependent parameter, Q = w t 2 t 2 3w t +1.8t . Further, the total length of the LET joint in the transverse direction, L trans (perpendicular to the joint axis) is given by With the above relations in mind, we adopt the following cost function for optimizing the LET joint [35] In the above, σ target = 840 MPa, and L target = 13 mm are the desired stress and LET joint transverse lengths, respectively. Also, w i1 and w i2 are normalized weights with w i1 + w i2 = 1 and w i1 & w i2 > 0. Genetic algorithm was used to optimize the parameters and minimize the cost function in (8). The optimized parameters of the LET joint are given in Table 1, with the Von Mises stresses at the LET joint being 850 MPa. The optimized LET joint allows for 360 • bending about its axis of symmetry. Concurrently, this folding approach achieved high compactness.
The effect of the LET joint cuts was examined to assess their impact on the antenna array's RF performance. Surface current density (J surf ) is depicted for a single polarization in Fig. 10, where the arrows shows the array expansion. As depicted in Fig. 10, the dipole currents near the coaxial feed have a stronger J surf and this was to be expected. Consequently, the LET cuts were placed at the end of the dipole arms and opposite to the feed location where J surf is comparatively weaker. Simulated VSWR of the infinite array with and without LET joint cuts are depicted in Fig. 11. It was observed that the LET joint cuts have minimal effect on the antenna's performance at broadside and while scanning at 45 • off-broadside.

C. ARRAY FOLDING MECHANISM
As depicted in Fig. 12, the antenna array and ground layers are kept together and at a constant distance using nylon screws. Further, a semi-circular disk was introduced (see Fig. 7) to facilitate the folding of the array along with the nylon screw. This disk can rotate 180 • up/down with respect to the horizontal plane for complete folding of the antenna array. The folding states of the antenna array are depicted in Fig. 12. It is remarked that the radius of the semi-circular disk (see Fig. 12 ) limits the maximum folding angle to θ F,max = 80 • . At this folding angle, the TCDA achieves an 80% size reduction along one dimension, implying an overall volume reduction of 60%. That is, this 5 × 8 array prototype of 40 cm × 25 cm × 3.9 cm in volume is reduced to 1.4 U at its maximum folding state. However, depending on the RF connectors and cable assembly choices, the maximum folding angle may need to be relaxed by 5 • to 10 • . Specifically, when SMPM connectors are used, the maximum folding should be relaxed by 7 • (see Fig. 13).
The robustness of the proposed foldable design was tested through a fatigue test for numerous folding cycles and antenna's RF performance was also measured for those cycles. For this test, only the center element port was excited while other neighboring ports were terminated with matched load. The results of the VSWR measurements are depicted in Fig. 14. It was observed that the array's RF performance was not affected significantly after the 200 folding cycles.

D. ACTIVE IMPEDANCE MEASUREMENTS
The fabricated array in Fig. 7 was used for measuring the active impedance using a Keysight N5222B Vector Network Analyzer (VNA). To compute the impedance of the desired element (center element), the mutual coupling with neighboring ones was measured, and combined with the measured reflection co-efficient of that element. While measuring the element impedance, non-excited ports were terminated with a 50 load at the microstrip balun input. The active reflection coefficient at the desired element (p,q) was evaluated as follows [15], [36]: In the above, u = k sinθ cosφ and v = k sinθ sinφ, k defines the free-space wavenumber, and S mn,pq refer to the Sparameter between the mn th and pq th elements of the array. As usual, M, N and D are the number of elements along the x and y directions and the array lattice spacing.
During measurements, the edge effect due to the finite size of the array were also taken into account in calculating the active reflection coefficient. Notably, due to reduced element to element coupling, the edge elements of a finite array  have slightly different impedance characteristics than the center element [37]. Therefore, to achieve optimum matching across the 5 × 8 array, power tapering was adopted [37], [38], [39]. Specifically, a cosine amplitude tapering was implemented as depicted in Fig. 15, represented by V mn in (9).
Using the amplitude tapering, the active VSWR of the center array element of the fabricated prototype was measured at broadside and at 45 • off-broadside scanning. Semi-infinite array simulations were also carried out using the set-up discussed in [4], [16] to assess the effects of amplitude excitation tapering. The simulation model depicted in Fig. 16 employs eight array elements in the x-direction and uses periodic boundary conditions (PBC) along the y-direction. Also, the finite element mesh absorbing boundaries were placed at λ low /4 distance from the array perimeter.
The measured results are given in Figs. 17 and 18 for the finite array vs the semi-infinite array and infinite array simulations. As seen, the simulations and measurements are in good agreement, giving a VSWR < 3 from 0.6 to 3.2 GHz at broadside. Specifically, the semi-infinite array simulation corroborates measurements at the lower band.   It is also observed that the active VSWR while scanning agrees well with simulations. It is remarked that the measured performance of the proposed array, specifically at the low band, could be improved by fabricating a larger array.

E. FAR FIELD MEASUREMENTS
Gain measurements were performed using a near-field anechoic chamber from 0.65 GHz to 3.2 GHz. At frequency range from 0.6 to 0.65 GHz, outdoor gain measurements were carried out using an LPDA to set the reference gain. Further, to avoid nearby and ground reflections, the measurements were conducted in an open field using an elevated antenna as depicted in Fig. 19. The separation of the antenna and the transmitter were set to 3 m with the antennas placed at a height of 3 m from the ground. The measured broadside gain for both the co-and cross-polarizations is given in Fig. 20 from 0.6 to 3.2 GHz. It is compared to the theoretical aperture gain = 4π A/λ 2 . As seen, the measured and simulated gain are in close agreement with some losses at the high band attributed to the ferrite choke [40] placed at the coaxial cable feed. Notably, the measured average cross-polarization level is approximately below 20 dB. This cross-polarization measurement can be improved further using careful fabrication and a better-suited reference antenna.
Gain patterns for both E-and H-planes are depicted in Fig. 21 and Fig. 22, respectively. It is observed that finite array pattern measurements are in good agreement with simulations. Minor distinctions observed between E-and H-plane patterns resulted from the unequal number of elements along E-and H-plane directions.

IV. CONCLUSION
This paper focused on presenting a foldable and readily deployable TCDA for low-angle scanning. Specifically, to achieve foldability with dual-polarization operation, a planar microstrip-to-slotline balun feed was adopted. The presented array operates across 0.6-3.2 GHz with dual-linear polarization and scanning down to θ = 45 • . The array provided average simulated radiation efficiency of 95% across the operating band.
A key aspect of the design is the array's foldability using LET joint as the in-plane folding mechanism, build-out monolithically on the substrate. This approach simplified the fabrication as well as folding and deployment. A 5×8 prototype was fabricated and measured to verify folding and array gain and bandwidth performance. Folding of the array leads to an overall volume reduction of 60%. The measured data at broadside and at θ = 45 • from broadside showed close agreement to simulations. The presented one-dimensional foldable array with the planar feed can be implemented for two-dimensional folding like Miura-Ori for further volume reduction.