The Rx and Tx antennas were mounted at the bottom and top of the unit cell. The antenna performance of a transmitarray unit cell is critical because these antennas directly receive and reradiate the signal from a source antenna. Therefore, the antenna should have a high gain and a proper radiation pattern. In this work, both Rx and Tx antennas were designed as stacked patches, considering the limited size of the unit cell (0.4λ₀ × 0.4λ₀). Fig. 2 shows the geometries of the proposed Rx and Tx antennas. The antennas had improved bandwidth due to a three-layer of stacked patches and proximity coupled feeding [25, 26]. When an antenna is designed on a high-loss substrate, its radiation efficiency declines due to interactions with the loss of substrate. This reduction directly affects the overall performance of transmitarrays. To address this issue, the implementation of a three-layer patch results in a more centralized electric field for the antenna compared to a conventional patch antenna, thus improving antenna directivity. It reduces the interaction between the field and the high-loss FR4. As a result, the loss of the antenna is minimized, leading to an increase in gain.

The current distribution of the Tx and Rx antennas is illustrated in Fig. 3. Although the Rx and Tx antennas have similar stacked patch structures, there is a significant difference in antenna polarization. The Rx antenna is designed for single linear polarization, whereas the Tx antenna supports dual-linear polarization for polarization selection. As a result of this difference, the Tx antenna features two orthogonal distributions, while the Rx antenna has a single distribution. These surface current distributions provide insights into the resonance mode of Tx (TM₁₀, TM₀₁) and Rx (TM₁₀) antenna. Fig. 4 shows the S-parameters and the fabricated Rx and Tx antennas. Each antenna was measured in an anechoic chamber, as shown in the inset of Fig. 4. The −10 dB impedance bandwidth of each antenna was sufficiently broad to cover the design target of Fs ± 0.05 GHz. The measured port isolation level between H-pol and V-pol was higher than 20 dB. Both the Rx and Tx antennas had broadside radiation patterns, as shown in Fig. 5. The co-pol and cross-pol gain values of the Rx antenna were measured to be 3.8 dBi and −21 dBi, respectively, at the operation frequency. The measured co-pol and cross-pol gain values of the Tx antenna were 4.25 dBi and −18 dBi, respectively, at the operation frequency. Certain discrepancies were observed between the simulation and measurements, as depicted in Figs. 4 and 5. During the fabrication process of the Rx and Tx antennas, three radiation layers were interconnected using bolts and nuts, which was challenging to precisely assemble by hand. As a result, small gaps occur between the layers, leading to subtle differences. However, despite these discrepancies, the simulations and measurements exhibited good agreement. Consequently, based on the aforementioned results, we confirmed that the antennas worked well as a Tx/Rx antenna.

At the bottom of the unit cell, as depicted in Fig. 1(c), there was an Rx antenna along with the control circuit. The control circuit comprised two main components: the phase shifter and the SPDT switch. Given the smaller size of the unit cell, careful consideration is required for the design of the phase shifter. To address this, this paper presents a miniaturized reflection-type phase shifter, as illustrated in Fig. 6(a). The proposed phase shifter consists of inductively loaded transmission lines for system miniaturization and two reflection loads for a broad phase-shift range of 0°–345° [27]. Two varactors were connected in series, and a 90° transmission line was resonated at the target frequency to improve the phase-shift range, as shown in Fig. 6(b). The varactor (MACOM MA46H120 [28]) was controlled by a DC bias voltage of 0–10 V. The reflection coefficient of the load is shown in Eqs. (1) and (2) [27]:

where Z_o is 50 Ω and X_L is equal to ωL_s – 1/ωC_v. The high-frequency current entering the bias line was blocked by the RF choke, and the DC current was blocked by the DC block. A conventional reflection-type reflector using a branch line coupler was miniaturized by an inductively loaded microstrip transmission line to integrate it with an Rx antenna in the same plane [29]. The designed miniaturized full-range reflection-type phase shifter was 36% smaller than a conventional reflection-type phase shifter. Fig. 6(c) shows the measured relative phase shift in degrees and the insertion loss (IL) of the proposed phase shifter. As the bias voltage changed from 0–10 V, the corresponding relative phase shift range varied between 0°–345°. In addition, the measured IL varied between 1.8 and 5.5 dB, depending on the voltage change. The IL variations were a result of the resonant properties of the series-connected varactors and the quarterwavelength T-line.

For polarization selection, the phase-shifted signal was directed to the V- or H-pol ports of the Tx antenna using an SPDT PIN diode switch, as shown in Fig. 7(a). The SPDT switch consisted of two PIN diodes (MACOM MA4AGP907 [30]), which were turned on and off exclusively. For example, 1.35 V and −5 V were applied to P3 and P2, respectively, to direct the input signal to P2. Fig. 7(b) shows the measured S-parameters of the designed SPDT switch. In this case, V_V = −5 V and V_T = 1.35 V (D1: on, D2: off). The measured IL of the switch from 0–20 dBm input power was approximately 0.7 dB, and the isolation level was approximately 27 dB. We experimentally verified that the proposed SPDT RF PIN diode switch satisfied all the design requirements for polarization selection, such as low loss, power-handling capability, and high isolation.

This section describes the operation principle of the proposed unit cell verification method using a TE₁₀ mode rectangular waveguide. In general, multifunctional unit cells are required to simultaneously control the polarization and phase of the incident wave [31]. The proposed waveguide-fed unit cell verification method confirmed the efficiency of the unit cell, polarization conversion, and phase shift. The efficiency of the unit cell (η_cell) is the ratio of the accepted power (P_a) by unit cells ((1-|Γ|²)·P_in) to the total radiated power (P_r), where P_in is the input power and Γ is the reflection coefficient. Thus, the efficiency of the unit cell was calculated as η_cell = P_r/[(1-|Γ|²)·P_in]. The calculated unit cell efficiency was approximately 13.2% because the loss values of the Tx/Rx antenna, switch (SW), and phase shifter (0 V) at Fs GHz were about 2.5 dB, 0.7 dB, and 3.1 dB, respectively. The polarization conversion and phase-shifting capability were easily measured by observing far-field patterns. Whereas the polarization conversion was easy to measure, the measurement of the unit cell’s phase-shifting performance was challenging. The phase-shift performance was measured indirectly by observing the radiation patterns of a waveguide-fed 1 × 2 unit cell array.

A shunt inductive iris was connected in parallel to match the 1 × 2 unit cell array to the designed 120 mm × 50 mm rectangular waveguide. A linear waveguide taper was designed to match the standard WR284 coax-to-waveguide transition to the designed rectangular waveguide. In this work, linear taper and rectangular waveguide structures were made of aluminum. The linear waveguide taper shown in Fig. 8(a) was designed to transfer a dominant TE₁₀ mode wave from the standard WR-284 to the designed 120 mm × 50 mm rectangular waveguide while suppressing mode conversion and higher-order mode excitation [32, 33]. The linear tapered matching waveguide section was designed for TE₁₀ single-mode operation using coupled-mode theory. The coupled-mode theory based on a cross-sectional method is a powerful design procedure for waveguide matching networks; thus, the total E- and H-fields along a general non-uniform waveguide structure can be calculated [34–36]. When a given waveguide structure operates in a single mode, the complex amplitude of forward (a⁺) and backward (a⁻) traveling waves associated with the propagation modes can be determined by a simple coupled-mode equation system, which is a strong function of the coupling coefficient (K) between those waves [35]. The K value of a rectangular waveguide for TE₁₀ single-mode operation having a propagation constant of β₁₀ was calculated analytically, as shown in Eq. (3) [37].

where a(z) and b(z) are the width and height along the z-direction, respectively. Eq. (3) clearly shows that the rate of change in the width or height profile along the wave propagation direction is an important design factor for the waveguide taper design. In this study, a single-mode rectangular waveguide taper with simultaneous variations in height and width was designed to match the 120 mm × 50 mm rectangular waveguide to the WR-284 standard. An inherent port mismatch level (ρ₀), which is a function of the difference between the cross-sections, was set to a value less than −15 dB (ρ₀ < −15 dB). The length of the linear waveguide taper was 50 mm, as shown in Fig. 8. The designed 120 mm × 50 mm rectangular waveguide had three available propagation modes of TE₁₀, TE₂₀, and TE₀₁ around the operation frequency (Fs), but the mode conversion level (TE₁₀-to-TE₂₀, and TE₁₀-to-TE₀₁) was less than −31 dB.

The fabricated waveguide-based feeding structure for the 1 × 2 unit cell array was 50 mm × 120 mm × 100 mm, as shown in Fig. 8(c). The designed rectangular waveguide structure was compatible with the WR-284 standard. A shunt inductive iris matched the 1 × 2 unit cell array to the coax-to-WR-284 waveguide transition. The measured reflection coefficient (|S₁₁|) of the designed rectangular waveguide-fed 1 × 2 unit cell array is shown in Fig. 9. We observed a discrepancy between the measurements and simulations, which can be attributed to fabrication errors in the customized waveguide. However, despite this discrepancy, both the measured and simulated data exhibited a similarity below −10 dB at the operating frequency, suggesting tight coupling between the Rx antennas and the waveguide. The rectangular waveguide-fed 1 × 2 array was measured in the anechoic chamber like Tx/Rx antennas, and the measured η_cell was 14.7% (−8.4 dB) with an input power of 30 dBm (P_in = 1 W). This was calculated by taking the ratio of the total radiated power to the power delivered to the array, as discussed in Section III-1. The measured total radiated and delivered power were 142.0 mW and 968.4 mW, respectively. This was very close to the theoretical efficiency of 13.2% of the unit cell discussed in Section III-1. As previously mentioned, the input power was tightly coupled and delivered to the unit cell without being reflected. Therefore, the measured efficiency can be regarded as the element loss (EL) of the full transmitarray antenna. The loss of the transmitarray antenna (ε_loss) is calculated as shown in Eq. (4) [31]:

where η_spill and EL_avg are spillover efficiency and element loss, respectively. η_spill is the ratio of accepted power to total power radiated by the feed, and it was calculated to be −2.3 dB using Ansys High-Frequency Structure Simulator (HFSS). The proposed verification was compared with the losses per component, as shown in Table 1. The total transmitarray loss (ε_loss) of the simulation was −11.1 dB, and the proposed method was −10.7 dB. The measured ε_loss values of a fabricated 8 × 8 transmitarray antenna composed of the presented unit cell were −10.75 dB [24]. A comparison of the measured ε_loss with the simulated losses showed good agreement, but the loss of the proposed method (−10.7 dB) more strictly predicts the measured value (−10.75 dB). Therefore, the proposed verification method provides useful information on the design and optimization of a full-sized multifunctional transmitarray or metasurface.

The simulation setup for unit cell verification is depicted in Fig. 10(a) and performed using HFSS. The standard WR-284 waveguide, excited by the waveport of a coaxial cable, was included in the model. A customized waveguide was positioned between the 2 × 1 array and the modeled WR-284, and the entire model, enclosed within a radiation box, was simulated. Fig. 10(b) and 10(c) show the radiation patterns of the waveguide- fed 1 × 2 array when there were no phase differences between the unit cells. The simulation and measurement data agreed very well. The polarization conversion capability was clearly observed, and the measured cross-pol suppression levels for both the V-to-V-pol and V-to-H-pol conversion cases were about 15 dB. The rectangular GND (ground) shapes had a difference between Fig. 10(a) and 10(b) because the GND lengths of the Tx antennas for V-pol and H-pol were asymmetric. Fig. 11 shows the measured beam steering capability for V-to-V and V-to-H polarization conversions. The ± 25° beam steering was achieved when the bias voltage for the phase shifter was 1.9 V, corresponding to a 90° phase shift. These results show that the proposed unit cell verification method successfully characterized all electrical properties, such as polarization conversion, phase shifting, and efficiency of a unit cell or a full-sized multifunctional transmitarray.

Table 2 displays a performance comparison between the proposed unit cell and the literature-reported transmitarray antennas. The results show that our proposed unit cell features a smaller size and higher phase resolution compared to other literature-reported transmitarrays. This confirms the superior performance of our proposed unit cell in terms of size reduction and phase control. In addition, conventional verification using a double waveguide structure and the proposed verification were compared, as shown in Table 3. Notably, all electrical properties of a multifunctional unit cell, such as unit cell loss, polarization conversion, relative phase shift, and far-field radiation patterns, can be experimentally verified through a one-time measurement. It is also possible to measure cross-pol suppressing levels. The rectangular waveguide structure of conventional verification methods supports only a single dominant mode at the operating frequency. The conventional closed waveguide structure for unit cell verification has limitations in measuring polarization conversion capability.