Design of Frequency Selective Windows for Improved ... - IEEE Xplore

IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. 54, NO. 6, JUNE 2006

secondly by allowing characterization of the scatterers’ phase response as a function of their length and of the incidence angle. In the second case, the numerical simulation of the scattering structure can be completely avoided, which is useful when the material properties of the layers are not known accurately. 2. In the prototypes used for validation, the H-plane angle of the incident signal on the array varied from 0 to 55 . It was found that taking into account the effect of this incidence angle in the design slightly improved the overall gain of the antenna. These conclusions apply to the reflectarray of dipoles printed on a thin membrane used in this paper. More validations are needed to generalize our results and validate the parameterization equations (8) and (9) for other types of scatterers. Two experimental conditions must be fulfilled to make the proposed technique applicable. Firstly, the condition given by (6) must be valid. This implies that the dielectric and conductor losses in the reflectarray structures need to be considered as negligible. Also, there should be no loss in grating lobes and the main beam should point in the direction of specular reflection of the incident illumination. These conditions are desired in practical reflectarray designs. Secondly, the near-field probe embedded in the scattering cell to extract its phase response should have no (or very weak) coupling with the fields of the incident and ground-reflected waves. In case of TE incidence used in this paper, this was efficiently achieved with a vertical monopole probe. Other probes (e.g., small loop) should be designed in case of other polarizations but this issue has not been addressed in this paper. The antennas shown in Section IV were designed for the sake of demonstration and had only 161 elements. In practical applications found in the literature, this number is typically increased by at least one order of magnitude. Given the small aperture size in our case, an offset feed configuration was used to minimize blockage. The effective aperture of the array projected in the plane perpendicular to the main beam direction is thus smaller than the physical aperture. Taking this into account, the efficiency values presented in Table II have been divided by cos max , as in [2]. ACKNOWLEDGMENT The authors are grateful to S. Fillipatos for providing MoL simulation results.

REFERENCES

1897

[9] K. Y. Sze and L. Shafai, “Analysis of phase variation due to varying patch length in a microstrip reflectarray,” in IEEE Antennas Propagation Soc. Int. Symp., Jun. 1998, vol. 2, pp. 1134–1137. [10] R. C. M. Li and A. A. Oliner, “Scattering resonances on a fast-wave structure,” IEEE Trans. Antennas Propag., vol. AP-13, no. 6, pp. 948–959, Nov. 1965. [11] F. Venneri, G. Angiulli, and G. Di Massal, “Experimental evaluation of the phase of the field scattered by microstrip patches for reflectarray design,” Microw. Opt. Technol. Lett., vol. 34, no. 3, pp. 163–164, Aug. 5, 2002. [12] K. Y. Sze and L. Shafai, “Formulation for reflection coefficient phase of infinite periodic array of microstrip patches,” IEE Electron. Lett., vol. 37, no. 3, pp. 142–143, Feb. 2001. [13] M. A. Tilston, “Thin-wire reciprocal multiradius implementation of the electromagnetic moment method,” Ph.D. dissertation, Dept. Elect. Eng., Univ. Toronto, Toronto, Canada, 1989. [14] É. Choinière and J.-J. Laurin, “Modeling of planar multilayered periodic arrays using the method of lines,” J. Appl. Computat. Electromagn. Soc., vol. 17, pp. 145–157, Jul. 2002. [15] K. Y. Sze and L. Shafai, “Substrate thickness in a microstrip reflectarray,” in Proc. Asia Pacific Microwave Conf., Nov. 30 –Dec. 3 1999, vol. 1, pp. 146–149.

Design of Frequency Selective Windows for Improved Indoor Outdoor Communication Mats Gustafsson, Anders Karlsson, António Pedro Pontes Rebelo, and Björn Widenberg

Abstract—The use of low emissivity windows degrades radio communication. This Communication presents design, manufacturing, and test measurements for an energy saving window that is transparent to GSM, GPS, and 3G radio wave frequencies. A frequency selective structure (FSS) is used in the metallic coating of the window to provide the desirable transparency in the frequency range from 900 MHz to 2 GHz. The periodic pattern used for the FSS is of the aperture type and the elements are hexagon loops. The FSS simulations are performed using the mode matching technique as well as the finite-difference time domain method. A frequency selective window was manufactured from a commercially available low emissivity glass. Measurements indicate that the frequency selective window has approximately 10 dB better transmission in the 900 MHz–2 GHz band than the original window. Index Terms—Frequency selective structure (FSS), radio channel.

[1] R. G. Malech, “The reflectarray antenna system,” in 12th Ann. Antenna Symp. USAF Antenna Res. Develop. Program, Urbana-Champaign, 1962, University Illinois at Urbana-Champaign. [2] D. M. Pozar, S. D. Targonski, and H. D. Syrigos, “Design of millimeter wave microstrip reflectarrays,” IEEE Trans. Antennas Propag., vol. 45, no. 2, pp. 287–296, Feb. 1997. [3] M. ElSherbiny, A. E. Fathy, A. Rosen, G. Ayers, and S. M. Perlow, “Holographic antenna concept, analysis and parameters,” IEEE Trans. Antennas Propag., vol. 52, no. 3, pp. 830–839, Mar. 2004. [4] F. S. Johansson, “A new planar grating-reflector antenna,” IEEE Trans. Antennas Propag., vol. 38, no. 9, pp. 1491–1495, Sep. 1990. [5] W. Menzel, M. Al-Tikriti, and R. Leberer, “Low-profile folded reflectarray antennas for communication applications,” presented at the Eur. Workshop on Integrated Radio Communication Systems, May 6-7, 2002, Chateau de Pignerolles, Angers, France. [6] M. R. Chaharmir, J. Shaker, M. Cuhaci, and A. Sebak, “Circularly polarized reflectarray with cross-slot of varying arms on ground plane,” Electron. Lett., vol. 38, pp. 1492–1493, Nov. 2002. [7] J. A. Encinar, “Design of two-layer printed reflectarrays using patches of variable size,” IEEE Trans. Antennas Propag., vol. 49, no. 10, pp. 1403–1410, Oct. 2001. [8] S. D. Targonski and D. M. Pozar, “Analysis and design of a microstrip reflectarray using patches of variable size,” in Antennas and Propagation Soc. Int. Symp. AP-S. Dig., Jun. 1994, vol. 3, pp. 1820–1823.

I. INTRODUCTION The use of a very thin metallic coating in modern window design is an extremely effective way to save energy. Acting as a filter, the coating blocks the electromagnetic radiation in the infrared region and is completely transparent to the visible part of the spectrum. Thus it rejects the heat transfer from outdoor to indoor during the summer and vice versa during the winter. These special windows are called low-emissivity (low-e) or energy saving windows since the metallic oxide layer reflects a significant portion of infrared energy [1]–[4]. The windows Manuscript received February 14, 2005; revised September 8, 2005. M. Gustafsson, A. Karlsson, and A. P. P. Rebelo are with the Department of Electroscience, Lund Institute of Technology, SE-221 00 Lund, Sweden (e-mail: [email protected]). B. Widenberg is with the Department of Electroscience, Lund Institute of Technology, SE-221 00 Lund, Sweden and also with Chelton Applied Composites AB, SE-580 13 Linköping, Sweden. Digital Object Identifier 10.1109/TAP.2006.875926

0018-926X/$20.00 © 2006 IEEE

1898


Fig. 1. Illustration of a frequency selective window. The infrared radiation (heat) is blocked by the window, but the indoor outdoor communication is possible. The visible part of the spectrum remains unchanged.

to microwave radio propagation since it is thin compared to the wavelength and its conductivity is extremely small. The panes are transparent for more or less all radiation with frequencies below UV light. Thus a window pane is transparent for visible light (390 to 770 nm), infrared (IR) light (770 to 2100 nm) as well as microwaves. Low-emissivity windows consist of a microscopically thin, practically invisible coating deposited on the surface of the window pane. One drawback with these energy saving windows is the degrading of the radio channel properties. For a commercial low-emittance window, the metallic oxide coating provides 20 to 35 dB of transmission damping in the frequency range 1–2 GHz. There are two common types of low-e windows: Hard coat low-e and soft coat low-e. Hard coat low-e, or pyrolytic coating, is a coating applied at high temperatures and is sprayed onto the glass surface during the float glass process. The coating is relatively durable, which allows for ease of handling and tempering. Soft coat low-e, or sputter coating, is applied in multiple layers of optically transparent silver sandwiched between layers of metal oxide in a vacuum chamber. This process provides the highest level of performance and a nearly invisible coating, but is a more expensive alternative than the hard coat low-e glass. III. NUMERICAL SIMULATIONS

Fig. 2. Hexagon geometry. a is the periodicity along the x and y axis, respectively, h is the height, w the width, and t the distance between elements of the hexagon.

are commercially available at large scale and used in many new buildings and vehicles. From a communication point of view, the windows involve a major drawback: the electromagnetic radiation in the microwave region is blocked by the coating and the blocking gets more intense as frequency increases. This means that wireless communication is severely restricted into and out from buildings. This can be an advantage for Wireless LANs since spectrum re-use can be increased by creating small isolated zones and privacy can be obtained by delimiting the space of use. However, for GSM, GPS, and UMTS transparency is crucial for the usage of these services inside buildings, see also Fig. 1. A solution to the radio communication problem is to create a frequency-selective structure (FSS) in the metallic coating of the low-e glass. The structure behaves like a band pass filter and is tuned to a bandwidth that covers the frequencies for GSM, GPS and UMTS, without degrading the thermal response of the window [1], [3], [4]. Hence the required bandwidth ranges from 900 MHz to 2 GHz. The FSS is in this case an array of periodic apertures in a conductive surface that, when illuminated by an electromagnetic wave, exhibits total transmission around the resonance frequency [1], [5], [6]. The planar two-dimensional periodic structure proposed in this paper is made of hexagonal loop elements, see Fig. 2. The hexagon elements were chosen due to their superior bandwidth and stability to different incident angles and polarization [6]. There are many numerical methods to simulate the response of an FSS. The methods used in this paper are the mode matching technique used in conjunction with the finite element method (FEM) [1] and the finite difference time domain (FDTD) method [9]. II. FSS LOW-EMISSIVITY WINDOWS The standard type of window pane is made of non-magnetic glass with a typical conductivity of = 10012 S/m, a relative permittivity r 4, and a thickness of l 4 mm. The pane is not an obstacle

Several simulations were made to investigate how the transmission curves depend on the parameters of the elements. These simulations are performed at different angles of incidence and different polarizations, cf., Fig. 3. Also the influence of the metallic-layer conductivity and the glass permittivity on the pass band was studied [10]. The parameters that can be used for the optimization of the transmission response in the microwave region, are directly related to the element itself. These include the element’s height, width and periodicity. For a better understanding of the simulations, these parameters are replaced by the height of the inner hexagon hi , the width of the loop w and the distance between hexagons a, cf., Fig. 2. For the typical soft coat window, a relative permittivity of r = 4 and a glass thickness of 4 mm were assumed. The elements are placed in a grid depicted in Fig. 2. The grid is infinite in the xy -plane and the z axis is orthogonal to the FSS surface. The elements were spaced periodically along the x axis with period a. The fraction of removed material, for the hexagon geometry, is given by 4w(w + 2hi ) : (2w + t + 2hi )2

(1)

The parameters of the FSS were determined by a parametric study where the width was kept small in order to minimize the transmission of infrared radiation [10]. The spacing between elements was t = 0:8 mm, the inner height hi = 9:3 mm and the thickness of the loop w = 0:2 mm. This design yields 3.8% of removed material and a bandwidth of B3 dB = 1:2 GHz, as shown in Fig. 3. There is a major improvement of the transmission compared to the window without FSS over the bands of interest, considering that the metal oxide coating provides 20 to 30 dB of attenuation. A further improvement can be obtained by centering the percentage of transmitted power so that the transmission loss is the same at 900 and 2 GHz. The conductivity of the structure can be approximated by a perfectly electric conducting (PEC) plate if materials with high conductivity like copper are used. If the material is not a metal with high conductivity, the FSS performance degrades and the PEC approximation is not applicable. To examine the significance of thickness and conductivity for the transmission of microwaves, two types of numerical simulations of an FSS with finite conductivity were carried out. The structure was a hexagon pattern with a thickness of one grid cell along the z -direction.


Fig. 3 Calculated transmission curves for the hexagonal aperture FSS design for incident angles = 1 ; 15 ; 30 ; 45 ; 60 . (a) TE case and (b) TM case.

The simulations were done with the FSS layer in free space in order to remove the influence of the window pane on the transmission. The frequency band for high transmission is then shifted but that is of less importance. In Fig. 4(a), transmission through a single layer is depicted for different values of the product d when the layer thickness, d, is fixed and the conductivity of the layer, , is altered. With a perfectly conductive surface (PEC), the FSS has a strong resonance around 2.4 GHz. The curve for a conductivity of = 104 S/m is almost on top of the PEC curve which indicates that PEC is a good approximation when the conductivity is high. However, as d decreases, the resonance fades away and the surface ceases to be frequency selective. Eventually, when the conductivity is extremely small, the structure behaves like a dielectric plate. In the second case simulations were performed for a constant d when both and d are altered. Due to the well defined points in space required by the FDTD method, the thickness of the conductive layer is altered by reducing the grid cell in the z -direction. The corresponding transmission curves are shown in Fig. 4(b). The transmission response tends to converge when the thickness of the conductive layer is considered thin. An extrapolation to zero grid size was performed to predict the response of the “infinitely thin” case. IV. MEASUREMENTS In order to perform measurements, a window pane with an FSS was manufactured. A 40 2 40 cm2 low-e glass soft coat window pane was chosen for this purpose. The hexagonal FSS was milled in the coating with a milling machine. The milling machine used a 36 mm long LPKF universal cutter for milling isolation channels and for engraving front plates [11]. This cutter can cut a line with a width of 0.2 mm. Fig. 5 shows that the hexagon elements can be visually observed since the

1899

Fig. 4 Dependence of the transmission response for a hexagonal FSS of the conductivity. (a) different d where d = 1 mm. As d decreases, the frequency selective behavior of the structure fades away. (b) Transmission response for constant d = 0:1 S.

thickness of the oxide layer is microscopically thin, and the machine removed glass material when performing the cutting. The result was far from perfect, however it was accurate enough to show the concept. A simple measurement was made on the window, using a shielded chamber with a 30 2 30 cm2 aperture. Two identical omni-directional antennas operating in the 1–12 GHz range were used as well as one network analyzer. The transmitting antenna was set 1.2 m away from the aperture in the chamber and the receiving antenna on the other side at 0.4 m from the window. Both antennas were at the same height and in line-of-sight as shown in the Fig. 5(a). Aluminum foil was used on the sides of the different glasses to improve the contact with the chamber. Measurements with five different devices were made: first the metallic plate that covers the chamber, then the soft coat window, then the soft coat window with FSS, then the hard coat window, and finally, just the aperture to normalize the measurements. The measurements were made twice to confirm the results. The data was collected in a laptop, that was directly connected to the network analyzer, and processed in MATLAB. The normalized results are depicted in Fig. 6. The fluctuations are mainly due to that the chamber is not anechoic, which causes a standing wave inside the chamber, that the size of the window is not large compared to the wavelengths, and the fact that the door was not completely closed to let the cable of the receiving antenna through the chamber. The short distance between the antennas and the window affects the curve and the finite size of the window may give rise to diffraction patterns. The improvement of the soft coat window with the FSS over the same window without the FSS is roughly 10 dB in the bands of interest (900 MHz–2 GHz). This is very satisfying considering that less than 3.8% of the infrared light is transmitted though the slits. The hard coat window has

1900


Fig. 5 (a) Measurement setup. (b) Soft coated window with the milled hexagons.

ulations are verified by measurements. The elements of the FSS are straightforward to design with the software used in this study. The size of the element relates to the resonant frequency, the width of the loop relates to the percentage of removed material and the inter element spacing relates to the bandwidth. The manufacturing of the hexagonal slits in the metallic coating of the glass suffered from several deviations from the original design. The engravings are much deeper than the actual thickness of the metallic coating, and the depth varies along the window. The soft coat window is also sensible to handling and can only be used in double glazing configurations. A more severe restriction in the measurements is the size of the window. Unfortunately it was not possible with the economical resources for this project to use a larger window. Nevertheless, the measurements show that there is a transmission improvement over the original window of 10 dB in the frequency band between 1 GHz and almost 2.5 GHz. It is expected that a full size window will give an even better result. The importance of the conductivity of the metallic coating was also analyzed. High conductivities provide good FSS performance, whereas materials with low conductivity degrade the spectral selectivity of the structure. The use of FSS is highly beneficial in the energy saving windows from an indoor outdoor communication point of view. A more precise manufacturing can give even better results and make the element geometry invisible, which is a requirement for successful commercial implementation. A general conclusion is that FSS technology seems to be an inexpensive solution to the problem of wireless communication in buildings with energy saving windows. With an increasing demand on wireless communication, simple designs like a bandstop space filter for Wireless LANs or bandpass filters like the one presented in this paper can lead to more efficient radio frequency management.

REFERENCES

Fig. 6. Four measurements are depicted: Soft coat windows with hexagonal apertures on the metallic coating provide an improvement of 10 dB in the transmission power over the same window without apertures. The plate measurements are relatively high due to the leakage contributions in the frame and the door of the shielded chamber.

a higher damping than the soft coat, especially from 1.4 to 1.9 GHz. This behavior was expected because the hard coat window was manufactured to work as a single pane configuration. There is a mismatch in the resonant frequency and the amplitude between the measured transmission and calculated transmission. That can be explained by the uncertainty in the value of the permittivity of the glass and of the finite structure in the measurements. The simulated curve is much smoother than the measured one. V. CONCLUDING REMARKS Design, manufacturing, and measurements of frequency selective windows are presented in this paper. The analysis and numerical sim-

[1] B. Widenberg, S. Poulsen, and A. Karlsson, “The design of windowpanes with high transmission at 900 MHz and 1800 MHz,” in Antenn 00, Nordic Antenna Symp., Lund, Sweden, 2000, pp. 185–190. [2] B. Widenberg and J. V. R. Rodríguez, Design of energy saving windows with high transmission at 900 MHz and 1800 MHz Lund Institute of Technology, Dept. Electroscience, Lund, Sweden, 2002, Tech. Rep. LUTEDX/(TEAT-7110)/1--14/(2002) [Online]. Available: http:// www.es.lth.se/ [3] C. Mias, C. Tsokonas, and C. Oswald, An investigation into the feasibility of designing frequency selective windows employing periodic structures 2002, Nottingham Trent University, Nottingham, U.K., Tech. Rep. AY3922. [4] M. Philippakis, C. Martel, D. Kemp, R. Allan, M. Clift, S. Massey, S. Appleton, W. Damerell, C. Burton, and E. A. Parker, Application of FSS structures to selectively control the propagation of signals into and out of buildings ERA Technology, Surrey, U.K., Tech. Rep., 2004. [5] B. Munk, Finite Antenna Arrays and FSS. New York: Wiley, 2003. [6] J. D. Kraus and R. J. Marhefka, Antennas, 3 ed. New York: McGrawHill, 2002. [7] B. Widenberg, “Thick frequency selective structures,” Ph.D. dissertation, Dept. Electroscience, Lund Institute of Technology, Lund, Sweden, 2003. [8] B. Munk, Frequency Selective Surfaces: Theory and Design. New York: Wiley, 2000. [9] A. Taflove, Computational electrodynamics: The Finite-Difference Time-Domain Method. Boston, MA: Artech House, 1995. [10] M. Gustafsson, A. Karlsson, A. P. P. Rebelo, and B. Widenberg, Design of frequency selective windows for improved indoor outdoor communication Lund Institute of Technology, Dept. Electroscience, Lund, Sweden, Tech. Rep. LUTEDX/(TEAT-7132)/1--14/(2005), 2005 [Online]. Available: http://www.es.lth.se [11] LPKF ProtoMat 95s/II Manual, 04 ed. Garbsen, Germany: LPKF Laser @ Electronics AG, 1999.