II. DESIGN OF THE PROPOSED MS POLARIZATION CONVERTER

The proposed structure designed for both left- and right-handed rotation is shown in Figs 1(a) and 1(b). As seen in Fig. 1, the structure consists of parallel microstrip lines with a thickness of “a”, which are equally separated and enclosed by a frame with the same thickness. The proposed structure also includes a diagonal line drawn between the opposite corners of the outer frame to provide cross-polarization component. The diagonal line is drawn from the upper left corner to the bottom right corner to obtain right-handed polarization converter (RHPC). For left-handed polarization converter (LHPC) application, the diagonal line is drawn from the upper right corner to the bottom left corner of the structure. The detailed information about the dimension of the unit cell of the MS model is shown in Fig. 1(c). The MS structure used in this work consists of 16 unit cells arranged in a 4 × 4 layout, which is quite appropriate when considering the electrical length of the frequency range we are working with. Table 1 shows the detailed dimensions of 4 × 4 layout.

Fig. 1. Layout of (a) the linear to left-handed circular (LHPC design), (b) the linear to right-handed circular polarization (RHPC design), (c) unit cell of the MS polarization converter.

Table 1. The dimensions of the unit cell (mm).

RT5870 Substrate with a thickness of 1.575 mm is used as the base of the structure. The relative permittivity and the loss tangent of the substrate are 2.33 and 0.012, respectively.

III. NUMERICAL ANALYSIS OF THE PROPOSED MS

As the title stands, the purpose of this design is to control the polarization of a linearly polarized incident wave. The designed structure can convert the polarization from linear to circular, which is either LHCP or RHCP wave. This feature can be shown by examining the AR of the wave passed through the structure. AR is the ratio of the cross-polarized wave to the co-polarized wave, so for a perfect circular polarization conversion, the AR should be equal to one. The AR can be written as:

(1)$${\rm AR}({\rm dB}) = Mag(20\,\log ({\rm AR}(\omega ))),$$

where AR(ω) represents;

(2)$${\rm AR}(\omega ) = \displaystyle{{T(\omega )({\rm cross}-{\rm polar})} \over {T(\omega )({\rm co} - {\rm polar})}}$$

and

(3)$$T(\omega ) = Mag(S_{12}).$$

Simulation of the proposed design is performed by using a commercial 3D software, CST Microwave Studio based on Finite Integration Technique. In order to plot the AR, the floquet port modes are chosen and the co-polarized and cross-polarized waves are obtained by adjusting the ports.

A 3 dB point can be considered as the reference point for polarization conversion. When the values of AR are smaller than 3 dB, the structure can be used as a polarization converter. On the other hand, the frequency range for which the AR value is smaller than the reference value is called bandwidth of the structure. As seen in Figs 2 and 3, the designed structure has considerably large bandwidths. For the optimization, the effects of microstrip lengths (tx, ty, and tz) are analyzed. Furthermore, the cell size and the strip thickness are also changed for the parametric study and the results are shown in Figs 2 and 3. As seen in Fig. 2, the cell dimension affects the behavior of the structure significantly while the microstrip thickness only shifts the working band. The variation of x dimension affects both the bandwidth and the magnitude of the AR. The best value range is obtained between 3.9 and 4.4 GHz with a maximum AR of 2.4 dB for x = 14.2 mm. Since the AR of upper and lower x dimensions does not provide proper polarization conversion, the structure acts as a resonator only for the middle value of x. The vertical dimension of MS also affects the AR. The resonance is obtained at y = 17.396 mm. The change of the vertical component only breaks the resonance and the structure cannot behave as a polarizer. The thickness variation of the strip slightly affects the AR. The increment of the “a” does not change the bandwidth and maximum AR but shifts the resonance band to the higher frequencies. The unit cell can be assumed as the combination of an L-C circuit. The capacitance of two strips is inversely proportional with the distance between them. Therefore, increment of “a” decreases the interval between adjacent strips and increases the capacitance. The resonance frequency also increases due to its direct proportionality with the capacitance value. In Fig. 3, it can be concluded that the effect of the horizontal strip lengths on the AR is small; however, the effect of diagonal line is remarkable. Therefore, the length of the diagonal line has to be chosen carefully.

Fig. 2. Effect of the unit-cell width, height, and strip thickness on axial ratio: (a) unit-cell width, (b) unit-cell height, (c) strip thickness.

Fig. 3. Effect of horizontal strip length (a) tx, (b) ty and crossed strip length, (c) tz on axial ratio.

To elaborate the study, the electric field distribution within the layout is obtained and plotted for various incident angles as shown in Fig. 4. When the incident angle is 0 or 180°, the horizontal lines become more sensitive and scatter most of the incident E-fields. On the other hand, the vertical lines become more active when the incident angle is 90 or 270°. These results are already expected due to the polarization axis. In other words, when the incident wave is vertically polarized, the vertical side of the structure will respond and emit most of the E-field and vice versa. Hence, the proposed structure can be used as a circular polarizer for both horizontally and vertically polarized waves.

Fig. 4. E-field distribution of the proposed MS structure at various incident angles (a) 0°, (b) 90°, (c) 180°, (d) 270°.

AR and phase difference between two perpendicular field components are shown in Fig. 5. For the efficient AR range, the phase difference is in quadrature as it is expected [Reference Ranga, Thalakotuna, Esselle, Hay, Matekovits and Orefice33]. The phase difference is between 130 and 70°, which is around quadrature value in the frequency range of 3.8–4.4 GHz.

Fig. 5. Axial ratio and phase difference between two perpendicular field components.

IV. EXPERIMENTAL RESULTS OF THE PROPOSED MS

Experimental study is also carried out by using a vector network analyzer and two horn antennas for the specific frequency range. The manufactured structure consisting of 64 unit cells arranged in an 8 × 8 layout is shown in Fig. 6(a). Horn antennas are calibrated first in the free space to obtain more accurate results and the structure is placed between them. The co- and cross-polarized incident waves are then obtained by simply changing the orientation of one of the horn antennas (Fig. 6(b)). Both transmitting and receiving antennas are kept vertically polarized to obtain co-polarized response but just one of them is rotated 90° in order to get the cross-polarized response. As we expect from the numerical studies, we have successfully obtained both co- and cross-polarized responses, which means that the structure converts the incident wave from linear- to circular-polarized waves.

Fig. 6. (a) Fabricated MS structure sample; (b) experimental setup.

In short, the operating principle of MS is based on the transformation of half of the incident energy propagating along horizontal (vertical) direction into the signal traveling along vertical (horizontal) direction. This wave power division is due to the geometrical properties of the MS. As seen in Fig. 7, the incident wave is divided into two nearly equal perpendicular waves, which result in linear to circular polarization conversion without incident angle dependency. Beside this, it is well known that the frequency-selective surfaces (FSSs) are mostly used to filter microwave signals, i.e. FSSs can be used as low, high, band pass or band stop filters in free space applications. The fundamentally used types of FSSs are patch and aperture. The patch type of FSSs behaves as band stop filter, the aperture shapes act as band pass filter. Hence, these types of structures are named as MS if it involves in both MTM and FSS [Reference Ortiz, Baena, Losada, Medina, Marqués and Quijano26]. Furthermore, the proposed MSs are exactly used to convert linear wave to circular one. As it can be seen from the measurement results, approximately 95% of the incident wave is transmitted with an AR below 3 dB. This high transmission coefficient is due to standard intrinsic impedance matching characteristics of the MS, and high-level radiation properties of the metallic lines.

Fig. 7. Numerical and measured (a) co-polar transmission results, (b) cross-polar transmission results, and (c) AR results of the proposed MS polarization converter.

For a better understanding, the transmission values and AR results are also measured, plotted and compared with the numerical ones as shown in Fig. 7. The bandwidth of the AR is between 3.87 and 4.38 GHz. This means that the AR values are below 3 dB at this frequency range. Hence, the structure can be used as a circular polarizer with a bandwidth of 510 MHz range. Besides, almost all co- and cross-polar transmission values are around 0.5. In other words, half of the linearly polarized wave is co-polar directed and other half part is cross-polar directed with respect to the incident wave. One can see from Fig. 7 that we almost have the same bandwidth with the simulation except for some small oscillations that are probably caused by the noise on the cables and laboratory conditions. It should also be stated that the frequency range is high enough to be affected by small quality and length changes in the cable and the environmental laboratory conditions. As a result, the experimental results agree well with the numerical ones.