Antenna description

The antenna is simulated using CST Microwave Studio and it is depicted in Fig. 1. A square low profile ARMA antenna was built containing 25 pixels [Reference Siblini, Taam, Jecko, Ramma, Bellion and Arnaud15] with 0.5λ ₀ spacing. Only nine are fed by the experimental device. The other ones, located at the edges of the antenna and connected to 50 Ω impedances are not fed. They are only used to maintain a good axial ratio for the whole matrix antenna. Each pixel includes a microstrip patch with four feeding ports, associated with a 90° phase shift, to obtain a circularly polarized field. The polarization circuit is printed on a TMM10i substrate (0.381 mm height) designed to generate these phase shifts. The main challenge to tackle this design is the small area behind the pixel surface (0.5λ ₀ × 0.5λ ₀ = 18.29 mm × 18.29 mm). It must be enough to integrate three hybrid couplers, three resistor pads, and a connector pad (Fig. 1(d)). Each of these three hybrid couplers needs one port connected to a 50 Ω resistor with grounded pads. These pads are designed and optimized separately to minimize their dimensions. SMP connectors are used. The connector pad for grounding and the microstrip line to solder the circuit-connector pin were also optimized to reduce the size. The patches are printed on the top of another substrate (RO6002 laminate, 1.524 mm). They are connected to the outputs of the polarization circuits by plated through holes (PTH). The ground plane is common for both the polarization circuit and the feeding patch substrates. Both substrates are stuck together by a Cuclad 6250 bonding film after removing the bottom metallization of the patch substrate. To connect the metallic walls and the ground plane, metallic plates and PTH are made on the patch layer. The other parts of the metallic walls (2 mm height, 1.5 mm thickness) are connected from the top of the patches substrate to the bottom of the FSS substrate (RO6002 laminate, 1.524 mm).

The manufactured planar antenna is therefore a multilayered structure built with four layers. Figure 2(a) shows the TMM10i substrate used to print the circuit with the nine elements of the RF polarization circuits. The feeding patches are printed on the top of the RO6002 laminate (Fig. 2(b)). Figure 2(c) shows the grid to obtain metallic walls. The FSS printed on the radome is presented in Fig. 2(d).

Fig. 2.

Multilayer structure: (a) PCB with the nine polarization circuits [Reference Siblini, Jecko and Arnaud14], (b) feeding patches printed on the dielectric substrate, (c) the grid, (d) FSS printed on the radome.

Scattering matrix measurement setup

A measurement test bench for multi-element antenna, previously described in [Reference Menudier, Arnaud, Thevenot, Fezai, Oueslati, Chevalier, Reynaud and Monediere16], has been used as shown in Fig. 3. It is equivalent to a multiway VNA (up to 49 ways).

Fig. 3.

Experimental setup for S _ij measurements.

S _ii parameters

The frequency evolution of the nine S _ii parameters obtained in simulation (Fig. 4(a)) is compared with the experimental one in Fig. 4(b).

Fig. 4.

S _ii parameters evolution as a function of the frequency: (a) simulation and (b) experimental results.

In both cases, the curves show the S _ii parameters lower than the expected “−10 dB” value on the whole bandwidth, illustrating the capacity of the ARMA solution to work on a wide frequency band [Reference Jecko, Arnaud, Abou Taam and Siblini13]. However, the S _ii minima are shifted toward high frequencies in the experimental results.

Coupling effects

An important property of the ARMA is that the couplings between the pixels are very small: usually smaller than the ones between elementary antennas in an array [Reference Jecko, Arnaud, Abou Taam and Siblini13]. This property is well illustrated in Fig. 5 which shows the modulus of the coupling coefficients between the central port (P5 in Fig. 3) and the other ones for all the frequency range. The experimental results (Fig. 5(b)) are lower than the theoretical ones (Fig. 5(a)) because the matching impedance losses are increasing this phenomenon, as expected.

Fig. 5.

Mutual couplings between the central pixel and the other ones versus the frequency: (a) theoretical result, (b) experimental result (modulus comparison).

Experimental device description

To measure the directivity, gain, axial ratio, and efficiency, the experimental device includes a power divider function with nine active outputs associated with a table of weightings [Reference Siblini, Jecko and Arnaud14, Reference Siblini, Taam, Jecko, Ramma, Bellion and Arnaud15]. The amplitude ratio of all the ports surrounding the central one (level = 1 and phase = 0°) is approximately A = 0.15 with a 180° phase. It was not possible to obtain a suitable power divider to check the previous specifications. As a whole, an experimental device was designed for measurements (Fig. 6).

Fig. 6.

Experimental device for radiation pattern measurements.

This device is built with “commercial off the shelf components” (COTS). It is made of an equi-amplitude and equi-phase power divider with 18 ways. Only nine outputs are used. The other ones are connected to 50 Ω loads, as shown in Fig. 6.

The eight ports corresponding to the pixels surrounding the central one are connected to attenuators to obtain an output level ratio close to A = 0.15, as given by the theoretical approach [Reference Siblini, Jecko and Arnaud14, Reference Siblini, Taam, Jecko, Ramma, Bellion and Arnaud15].

The central pixel is fed by the central port called P6 through a phase shifter to obtain a 180° phase shift with the surrounding ports. The weighting table obtained experimentally with the device is presented in Table 1 and compared with the theoretical one. The ratio between the central port and the other ones is slightly different (A = 0.17) due to the COTS attenuators resolution.

Table 1.

The weighting function

Measured scattering matrix of the experimental device

The previous table of weightings is verified in linear modulus (Fig. 7(a)) and phase (Fig. 7(b)) for the whole frequency range.

Fig. 7.

Input–output parameters versus frequency: (a) modulus and (b) phase.

The matching parameters S _ii are compliant with our −10 dB objective as shown in Fig. 8(a). It is the same for the S _ij coupling parameters in Fig. 8(b).

Fig. 8.

(a) S _ii and (b) S _ij parameters evolution over the frequency range.

Isoflux radiation patterns

Radiation patterns (directivity, IEEE gain, and realized gains) are measured in our anechoic chamber (Fig. 9). An aluminum thin sheet allows simulating the nanosat platform. Absorbers were installed behind the antenna to minimize electromagnetic interferences of the motor. Therefore, the back radiation will be minimized resulting in a slight increase of the antenna directivity.

Fig. 9.

XLIM anechoic chamber used for radiation patterns measurement : (a) side view, (b) back view.

The 3D RHCP radiation pattern theory-measurement comparison is presented at the central frequency of 8.2 GHz in directivity. In Fig. 10(a), the isoflux shape is illustrated on the 3D radiation pattern for theoretical and experimental results. In both cases, the circular symmetry is disturbed by the edges diffraction. Figure 10(b) gives the RHCP radiation pattern in directivity at the same frequency for some φ cut-planes (0°, 44°, 90°, 134°) and Fig. 10(c) shows the axial ratio versus the θ site angle: the results are not really compliant with the isoflux characteristics around the axial direction. However, it is not a problem for a spatial bi-mode application [Reference Siblini, Jecko and Arnaud14, Reference Siblini, Taam, Jecko, Ramma, Bellion and Arnaud15] because the isoflux solution is used only for angles higher than 30°.

Fig. 10.

(a) 3D radiation patterns, (b) RHCP radiation pattern in φ = 0°, 44°, 90°, and 134° cut-planes, (c) axial ratio for the same cut-planes.

A good agreement is obtained between the theoretical and experimental results.

Frequency dependence comparison and power balance interpretation

Figure 11 shows the theoretical and experimental evolution with the frequency of the directivity and associated maximum realized gain. The theoretical and experimental results are in good agreement on the whole frequency band.

Fig. 11.

Theoretical and experimental directivity and maximum realized gain versus frequency.

The directivity is obviously higher than the realized gain (about 7 versus −10 dBi) due to the losses of our experimental device. This one introduces 14.4 dB losses, as shown in Table 1: 10 log(0.17² + 8 × 0.03²). It remains 2.6 dB due to the losses in the antenna itself (50 Ω resistances, dielectric substrates, transmission lines, connectors, etc.) and due to the experimentation process (±0.8 dB).