Meshed patch antenna integrated into car windscreen

G. Clasen and R.J. Langley

An investigation into the performance of a microstrip patch antenna integrated into a laminated glass windscreen for a vehicle is presented. It is fed using a coplanar waveguide feed printed on the innermost layer of the glass avoiding the need for a contacting feed. The patch and ground plane are meshed for manufacturing in the glass.

connector (not shown but the pad is visible at bottom of the Figure). This allows the positioning of the patch antenna anywhere on the windscreen. Both patch and ground plane are meshed; only the relatively small area of the centre strip of the coplanar waveguide is solid metal. Fig. 1 also demonstrates the degree of transparency that can be achieved using a meshed patch antenna. The letter A is clearly seen through the antenna.
patch

glass

Introduction: With the growing number of telematic systems used

in cars there is an increased need for the integration of antennas into the structure. Microstrip patch antennas which are lightweight and low in cost can be integrated into different parts of the car body. Previous work has examined their performance with glass substrates and superstrates [l, 21. Our aim in this Letter is to demonstrate how a patch antenna can be integrated into a car by making it part of the windscreen. Integrating the patch antenna into the windscreen poses two practical problems: (i) It is not possible to screen print large solid metal areas (e.g. a patch at 1.5 GHz) because this would distort the windscreen during the heating process by which it is formed. (ii) It is not possible to drill holes through the glass and it is therefore impossible to use a conventional coaxial probe feed. Nor is it possible to feed the patch directly using a microstrip line. The first problem is overcome by replacing the solid metal areas of the patch by a mesh structure [3, 41; this gives the additional advantage that the antenna gains a degree of transparency. The second problem is tackled by using a coplanar waveguide feed structure printed on the inside of the screen glass for simple connection. Otherwise a two layer external feed must be used.

glass

ground plane

co-planar line

1624/21
Fig. 2 Layer geometry of patch in windscreen

Fig. 3 Principle plane radiation patterns for patch in windscreen

(i) E-plane, copolar (ii) E-plane, crosspolar (iii) H-plane, copolar (iv) H-plane, crosspolar

Measurements: Fig. 3 shows the normalised radiation pattern

Pig. 1 Meshed patch antenna over ground plane with coplanar waveguide feed printed on glass

Antenna is placed on piece of paper with letter 'A' printed

Geometry: Fig. 1 shows the patch antenna structure. The patch, which is 6" 0 in diameter, is in the centre region over the larger ground plane. It resonates at 1.32GHz. The mesh geometry is printed for the TM,, mode in this instance [4] and follows the mode currents. Fig. 2 shows the layer geometry for the proposed antenna. The windscreen consists of two pieces of glass held together with a plastic interlayer material (PVB). During manufacturing the whole structure is heated and the PVB melts to adhere the two pieces of glass together. The meshed patch is printed onto the lower side of the top glass layer and it is thus protected from environmental influences by the glass. The meshed ground plane is printed onto the bottom side of the lower glass; this geometry allows access to the ground plane and feed from the inside of the car. Alternatively the patch can be printed on the same inner glass to reduce problems with glass alignment during manufacture. To feed the patch antenna, a coplanar waveguide (CPW) feed is etched into the ground plane and connected to a surface mount

measured in the E- and H-plane, respectively. Previous results showed that meshed patch antennas have a reduced resonant frequency and therefore an effectively reduced ground plane size. Circular meshed patch antennas also have a very low crosspolar component [4]. Fig. 3 shows that the meshed patch antenna integrated into the windscreen has indeed a low crosspolar component in the principle planes. The beam in the E-plane is shifted by a few decibels; this was found to be due to radiation from the CWP feed. Backward radiation is quite high at -7dB. This is partly due to the relatively small ground plane but also due to radiation through the holes in the meshed ground plane. The gain was -3dB lower than that in a conventional patch antenna, accounted for by losses in the glass and meshing of the structure. The return loss was measured giving a -10dB bandwidth of 2%. Conclusions: In this Letter we have demonstrated how a patch antenna can be integrated into a car windscreen. Feeding via a coplanar waveguide feed and a surface mount connector allows it to be positioned anywhere on the windscreen and avoids the need to drill holes. Creating the patch from a mesh structure allows the antenna to be printed onto the glass and gives a degree of transparency. Performance is adequate for many applications.

ELECTRONICS LETTERS

27th April 2000

Vol. 36

No. 9

781

6 March 2000 0 IEE 2000 Electronics Letters Online No: 20000614 DOI: IO. 1049/e1:20000616 G. Clasen and R.J. Langley (Electronic Engineering Laboratory, University o Kent, Canterbury, Kent CT2 7NT, United Kingdom) f

References
LOWES, P., DAY, s.R., KOROLKIEWICZ, E., and SAMBEL, A.: Performance of microstrip patch antenna with electrically thick laminated superstrate, Electron. Lett., 1994, 30, pp. 1903-1905 2 ECONOMOU, L., and LANGLEY, R.J.: Circular microstrip patch antennas on glass for vehicle applications, IEE Proc. Microw. Antennas Propag., 1998, 145, pp. 416-420 3 wu, M.-s., and ITO, K.: Meshed microstrip antennas constructed on a transparent substrate,IEICE Trans., 1991, E74, pp. 1277-1282 4 CLASEN, G., and LANGLEY, R.J.: Gridded circular patch antennas, Microw. Opt. Technol. Lett., 1999, 21, pp. 311-313

!. ... ..E.. ..

I
..-________...----_

/G .
Ix

,
lY

.. 4

~.....~..~~-~~~.......-~

Modelling of bow-tie microstrip antennas using modified locally conformal FDTD method
J. George
An analysis of bow-tie microstrip antennas is presented based on the use of the modified locally conformal finite-difference time-

The algorithm given in [7] is re-arranged and slightly modified as follows. To begin with, the computational volume is divided into a collection of Yee cubes and then the metal free face areas (A,) and segment lengths (1, and I at the antenna plane are com, ) puted for those cells through which the slanting edges pass. Now, for all the fields associated with them, flag matrices are set up for direct use in the field update equations [7, 81. For instance, the rearranged H-field update equation along the z-direction with the corresponding flag matrix (FhJ is given by

domain (FDTD) method. Thu approach enables the number of cells along the antenna length and width to be chosen independently of the antenna central width, which helps to keep the number of cells required in those directions to a minimum. The analysis results are compared wt experimental results and ih good agreement is observed.
Introduction: Bow-tie microstrip antennas have recently attracted a

large amount of interest [l - 41. In present day communications scenarios, bow-tie microstrip antennas are attractive candidates owing to their compact nature [l] compared to rectangular microstrip antennas. However, only very few attempts have been made to analyse these antennas [5, 61. One method, also based on the f~te-difference time-domain (FDTD) method, can be used to perform such an analysis, but it requires a variation in the cell size with the central width (w)of the antenna to ensure that the slanting metallic edges of the patch pass exactly through the diagonals of the cell faces. This requirement in turn requires an increased number of cells, both along the antenna length (L) and width (b), than the corresponding rectangular patch antenna of the same length and width. The other method, proposed by the author and colleagues, based on a segmentation technique is capable of performing the analysis only for bow-tie configurations that can be divided into segments, the Green functions of which are known. In this Letter, the mentioned inconveniencesilimitations in the analysis of bow-tie microstrip antennas are overcome by using the modified locally conformal FDTD algorithm [7]. The algorithm takes into account the slanted metallic edges of the patch by using slightly moditied field update equations, alone or in conjunction with a backward weighted averaging scheme depending on the degree of distortion of the cell face. This method provides the user with the flexibility to determine the number of cells required in different directions independently of w. Simulation results have been compared with the corresponding experimental results to demonstrate the effectiveness of this approach.
Antenna design details and summary of algorithm: The general

where Fh,(i,j, k ) = 2 if A,,

< 0.075 or

a 12 > AZ(z,.i,k)

12

Fhz(z,j,k ) = 1 otherwise
A, represents the distorted cell face area normalised with respect to the undistorted cell face area and li,, denotes the square of the maximum side length of the distorted cell face. Use of the flag matrix, which is generated in the meshing routine as in [7], directly in the field update equation makes this implementation faster than the method suggested in This is because the implementation of the update equation itself does both necessary decision making and backward weighted averaging. The other relevant E and Hfield components are also updated in a similar manner.

[I.

geometry of the bow-tie antenna and the relevant dimensions are shown in Fig. 1. The Figure also includes a hypothetical version of the FDTD grid at the antenna plane suitably drawn to demonstrate the algorithm (a magnified section of the grid through which the slanting edge passes is also shown in the Figure). All the antennas were fabricated on a substrate of dielectric constant E, = 4.4 and thickness h = 0.16cm. The feed point (probe feed) was specified in terms of the distancefd from the radiating edge along the central line as shown in the Figure.

Numerical and experimental results: The bow-tie antenna configurations used in this investigation were categorised depending on whether the length to width ratio (Ub)is equal to, greater than, or less than one. In each category, three different cases of central width (w = 0.3b, 0.5b and 0.7b) were considered. The fKst category considered (Ub = 1) was with L = 3.5, and b = 3.5cm. The cell dimensions along the x- and y-directions were chosen in such a way that an even number of them exactly matched with Land b as shown in Fig. 1. The cell dimension along the z-direction was selected so as to make an integral number of cells fill the substrate thickness. The numbers of cells used along the x-, y-, and z-directions for the three cases of central width were n, = ny = 60, and n, = 25, respectively, and the corresponding cell dimensions used were A, = Ay = 0.0875, and Az = 0.04cm. In all the cases investigated in this Letter, excitation was carried out by using a z-directed Gaussian pulse of source resistance 5 0 n [9] located just above the ground plane. The number of time steps used in all the cases was 16384 and the size of time step chosen in each case was 75% of the respective Courant limit. Fig. 2a shows the simulated and experimental variations of return

782

ELECTRONICS LETTERS

27th April 2000

Vol. 36

No. 9