Array Design and Measurement

An array with five pairs of slots is simulated with dt varying. The results are shown in Fig. 3.6. Here the normalized radiated power is obtained by normalizing the radiated powers by the maximum among them, i.e. the one when dt = 18 mm. The normalized radiated power peaks when dt is about 18 mm and oscillates with the periodicity of 16 mm, which is close to half of a guided wavelength of the dominant CBCPW mode at 5.5 GHz. The simulated H-plane (φ = 90°) radiation pattern with this value of dt is plotted with the cross marks in Fig. 3.8. The figure also plots, with the circular marks, the same array but with the end of the transmission line terminated. When the line is terminated, the power tapering makes the pattern rather asymmetrical, with the first pair of the side lobes at -16.86 dB and -10.98 dB, respectively. When the forward traveling wave on the CBCPW is allowed to reflect back, its leakage also excites the side slots, but now it is the last (5^th) pair that is excited most strongly and decays in the reverse order. This compensates for the original tapering and shapes the pattern much more symmetrical, so the levels of the first pair of the side lobes become -13.03 dB and -12.06 dB, which is comparable to that of a uniformly-excited array.

To check the validity of our design, the above array was fabricated on the FR4 substrate with dielectric constant εr = 4.3, thickness h = 1.6 mm, and loss tangent tanδ = 0.02. The remaining parameters are as follows: ws = 2 mm, ls = 23 mm, dt = 18 mm, dy

= 31 mm, d = 15.5 mm, and ds = 13 mm. The dimensions of the feed-line are G = 0.8 mm and S = 2.5 mm, which correspond to a 50-Ω characteristic impedance. The return loss is measured using the Agilent E8364B network analyzer. The E- and H-plane radiation patterns are measured in an anechoic chamber, using the Agilent 8722ES network analyzer.

Fig. 3.7 shows the return loss versus frequency. Fig. 3.8 and Fig. 3.9 are the H- and E-plane (φ = 0°) patterns, respectively, with and without termination at 5.5 GHz. It can be seen that there are some notable discrepancies between the simulation and the measurement results. The main reason is that as the leakage wave impinges on the slots, it partly radiates, partly reflects back, and partly transmits to the side directions. In the simulation, because the infinitely large ground plane and substrate are assumed, these transmitted waves will never reflect back. However, in practice they radiate as well as reflect at the periphery of the structure and interfere with the antenna itself, thus resulting in the poor performance. We therefore design another larger array with three columns of slots on each side of the feed-line. The number of rows is still five. The offsets dx12 = 14 mm, dx23 = 12 mm, dy12 = 4 mm, and dy23 = 10 mm. The transverse

spacing dxmn is chosen to be about λgpp/2 according to the phase cancellation technique and the longitudinal spacing dymn is tuned to ensure the in-phase excitations of the columns of slots. Here λgpp denotes the guided wavelength of the parallel-plate TEM mode. The columns of slots are not at each other’s exact broadside due to the longitudinal offset dymn, but the phase cancellation can still be achieved in the transverse (x-) direction as [116] indicates. It is expected that with the increasing number of slots on the side, more power will radiate through the slots instead of radiating and reflecting at the ground edge. The photograph of the 5 x 6 array is shown in Fig. 3.10. The input return loss, the H-plane pattern, and the E-plane pattern are shown in Figs. 3.11, 3.12, and 3.13, respectively. The cross-polarization level is below -20 dB in all directions and is omitted here for clarity. The accuracy of the simulation improves considerably, especially in the patterns. This also confirms the above discussions. The main beam points towards broadside and the backside radiation level is below -30 dB. The levels of the first pair of the side lobes are -13.6 dB and -12.8 dB, respectively. The measured antenna gain is 10.84 dBi, which corresponds to the antennas efficiency of 33.6 %. Note that doubling the substrate height will drastically increase the efficiency to more than 50

%.

The array is capable of frequency-scanning in the H-plane. Fig. 3.14 plots H-plane

patterns at various frequencies between 5 GHz and 6 GHz. The amount of the beam scanning and antenna gains are summarized in Table II.

3.1.5 D ESIGN P ROCEDURE

The design procedure can be summarized as follows.

1. Start the design by listing several possible feed-line dimensions for the predetermined characteristic impedance. The lower and upper bounds may be, for example, the fabrication process limit and the dimension that is mechanically not matched to the connector, respectively. Choose one that has the medium leakage rate

2. For the design with a broadside main beam, make d and dy equal to 0.5 * λgCBCPW and λgCBCPW, respectively.

3. Choose ds to be about 0.5 * λgpp.

4. Choose ls and ws such that the maximum radiated power is obtained.

5. Place one column of slots on either side of the feed-line. Vary dt and choose the value that results in the maximum radiated power.

6. If the resulting H-plane radiation pattern is too asymmetrical, or if the input reflection is too strong, go back to Step 1 and choose another feed-line dimension accordingly.

7. Let dxmn be about a half of the guided wavelength of the parallel-plate TEM mode.

Adjust dymn such that all slots are in-phase, or nearly so, and a maximum antenna gain is achieved.

3.1.6 S UMMARY

A novel design of a longitudinal slot array antenna fed by a CBCPW has been presented. By longitudinally placing the slots and offsetting, the array radiates pure linearly-polarized broadside main beam. The attenuation characteristic of the CBCPW has been studied for suitable choice of the feed-line configuration. The dimensions and the arrangement of the radiating slots and the effects of the reflected wave have been discussed. A small array was fabricated but the performance suffered from the reflection and the radiation from the waves at the periphery of the structure. A larger array was therefore designed and fabricated to overcome this problem. The array exhibits wide impedance bandwidth and a highly unidirectional pattern capable of frequency-scanning.

This antenna, using only one layer of the dielectric substrate, can find applications where inexpensive frequency-scanning arrays on the CBCPW are demanded.