An Optimum Design of Paired Balanced Antipodal Vivaldi Antennas With Mirror-Imaged Symmetric Architectures for Ultra-Broadband Characteristics From Microwave to Millimeter-Wave Frequency Ranges

This paper presents a balanced antipodal Vivaldi antenna design to produce ultra-broadband radiations from microwave to millimeter-wave frequencies. In particular, this work relaxes the design constraints to optimize the geometrical parameters to broaden the operational frequency bandwidth, where 147% relative bandwidth has been achieved. Moreover, in this optimized design, the beam squint and cross-polarization (X-pol) discrepancy at high frequencies are improved by pairing two mirror-imaged antenna structures to form an integrated one. The resulted X-pol level can retain good performance in the entire band. The beam squint can be minimized by taking advantage of the 3dB gain increase provided by the array factor. HFSS full-wave simulations first examine the proposed work, further validated by measurement from a basic antenna prototype. The beam squint is less than 3 degrees in the major middle part of the achieved band, and the X-pol level is less than −30dB for the paired antennas.

summarizes the frequency bands for the applications 23 mentioned above, located in the broadband of 17 ∼ 86 GHz. 24 The associate editor coordinating the review of this manuscript and approving it for publication was Chinmoy Saha . Such applications may be integrated into a single device with 25 different combinations. 26 It is desirable to develop a single antenna set to accom-27 modate these frequency bands to simplify system implemen-28 tation cost simultaneously. It is challenging to retain good 29 radiation characteristics in such broad bandwidth and cre-30 ate broadband reflection coefficients. In this case, antipodal 31 Vivaldi antennas (AVA) [5] were considered a good candi-32 date for wide frequency bandwidths. Various efforts have 33 been dedicated to improving radiation characteristics, where 34 radiation mechanisms. Section III presents the numerical 91 investigations by HFSS full-wave simulations and parametric 92 studies for broadband operations, highlighting the behaviors 93 at the high part of mmW frequencies. The experimental 94 validation is shown in Section IV examines the radiation 95 characteristics, where comparisons to previous related works 96 are also summarized. A short conclusion is presented in 97 Section V. The proposed antenna architecture is illustrated in Fig. 1, 103 where detailed descriptions of operational principles are 104 available in [17]. We focus on the characteristics of high 105 mmW frequency band extension by optimization. The 106 antenna is a sandwiched structure formed by two dielectric 107 substrates to produce three interface layers and implement 108 three metal Vivaldi radiating flares on the interface surfaces. 109 This antenna is in a dipole format with two essential arms (the 110 metal flares) to radiate the electromagnetic (EM) fields from 111 their gap. The goals of low XPLs and broad bandwidth are 112 achieved by using the following mechanisms. First, the dipole 113 arms are formed by a symmetric configuration. In particular, 114 the positive arm, indicated by the red color flare in Fig. 1, 115 is implemented on the middle interface sandwiched by the 116 two dielectric substrates. 117 FIGURE 1. Proposed antenna structure and geometrical parameters. The yellow region is the substrate cutout, and the via-hole grounding is also shown.
On the other hand, the opposite arm is split into two 118 parallel ones, indicated by the blue color flares in Fig. 1, 119 which are implemented on the two outer surfaces of the 120 two dielectric substrates. Metal vias short them to form an 121 effective arm. This configuration of dipole arm architecture 122 can minimize the XPLs at low frequencies. The two blue 123 arms can be paired with the red arm to form two dipole 124 sets in a mirror-imaged relationship to the middle interface 125 where the red arm is placed. Thus, the X-pol components 126 polarized along the cross-section dimension of the antenna 127 can be canceled regardless of frequencies to suppress the 128 VOLUME 10, 2022 if the radiations are mainly produced from these two arms.

130
To increase radiation efficiency and match the free space impedance, the dielectric substrates on the gap between the 132 two Vivaldi arms are cut out to form the electric lines in free 133 space directly.

134
On the other hand, the broadband operation is first achieved 135 by modifying the Vivaldi arms' shapes to eliminate the 136 diffraction effects and current flows from the external edges 137 back to the feeding structure. It is further suppressed by 138 cutting slotted edges to suppress the outer edge currents, 139 the return paths back to the feeding structure. Moreover, 140 a broadband microstrip-to-stripline transition is designed.    The design equations are optimized to control the radiation 179 patterns and, therefore, the gains because it is challenging to 180 retain a broadband gain performance. The basic equations for 181 the inner and outer curved edge profiles of the copper flares 182 are given by [17] 183 where W S is the width of the feeding stripline at the origin of 185 the copper flares and (α 1 , α 2 ) are indices for controlling the 186 flare's profile. The design criteria are to narrow the opening 187 of the radiating flares for high-frequency radiation control. 188 Thus, the gains at the lower frequency part are slightly smaller 189 than the results in [17] as a compromise of gains and band-190 widths. 191

2) CONSIDERATION OF BROADBAND TAPERED EDGE SLOTS 192
As illustrated in Fig. 1, the numbers and tapers of edge slots 193 are optimized for a broadband operation from a dual-scale 194 corrugation on the outer edges of the antenna's three copper 195 flares. It is noted that the high-frequency slot widths and their 196 microstrip lines cannot be too small due to the mechanical 197 limitations and the related distortions in fabrication. Thus,198 to increase the number of high-frequency slots for broadband 199 induced current suppression, the low-frequency slots were 200 reduced from four elements in [17] to three to accommodate 201 twenty high-frequency elements. The slope of the tapered 202 slots is decreased to provide a smoother transition for broad-203 band operation, which is particularly important for the high-204 frequency induced current suppression. On the other hand, 205 the decrease of low-frequency elements will only slightly 206 affect the radiation patterns, i.e., the symmetry of the E-plane 207 radiation patterns.

208
In the implementation, the shapes and orientation of the 209 high-frequency slots are rectangular and perpendicular to the 210 edges for easy fabrication, while these low-frequency slots 211 are in pencil shapes and slanted to reduce the depths to the 212 inner edges. These outer edge corrugations are essential for 213 low microwave frequency to middle mmW frequencies for 214 smoother impedance variations. At the higher mmW frequencies, the antenna's radiation 218 arises from the metal flares on the portion near the feeding 219 transition. To achieve broadband matching and avoid EM 220 emissions from the microstrip lines, metal vias are installed 221 to optimize the radiation and reflection coefficients. Full-222 wave simulations optimize the number of metal vias and 223 their positions to achieve good performance. In this design, 224 five pairs of metal vias are implemented. It is noted that 225 the mechanical fabrication limitations of metal vias-holes 226 should be considered. A metal ring pad should be applied 227 to every vias. They should also be at a distance from the 228 stripline and the junction between the microstrip and stripline. 229 These vias are arranged symmetrically to retain good polar-230 ization because the lateral EM leakage produces X-pol fields 231 between the vias and stripline.

259
One first considers parametric studies by HFSS full-wave 260 simulations to examine the behaviors at high-frequency parts.

261
The studies are based on the optimum parameters in Table 2 to 262 discuss the parametric effectiveness.  Table 2 for good behaviors of reflection coefficients. A metal 270 ring of 0.65 mm in diameter is added to via holes for a com-271 plete structure. The parameters under study are the separation 272 distances to the stripline and the junction of the microstrip-273 to-stripline transition, as illustrated in Fig. 3 (a). One first 274 examines the impedance matching behaviors. The reflection 275 coefficients are shown in Fig. 3(b) and (c), respectively, 276 as expected that these two parameters mainly affect the high-277 frequency parts. The offset to the transition junction between 278 the microstrip and stripline affects the reflection coefficients 279 at frequencies beyond 48 GHz, while the separation to the 280 stripline affects the behaviors beyond 64 GHz. In particular, 281 when the vias are not used, the reflection coefficients at the 282 lower frequency part will also be affected in addition to the 283  One also considers an important factor in altering the radi-300 ation patterns, especially the main beam pattern forming to 301 result in beam squints to frequencies, which is the width of 302 the dielectric substrate leftover after the dielectric substrate 303 cutout in Fig. 1. Fig. 5 (a) illustrates the width of the left-304 over dielectric substrate on the side of the twin blue flares.

305
This leftover was intended to balance the radiation due to 306 the asymmetric metal flares in the initial design. Fig. 5(b) 307 first shows the influence on the reflection coefficients, where 308 relatively stable results have been obtained. The E-plane 309 radiation patterns at 72 GHz are shown in Fig. 5(c). It is seen 310 that this parameter will alter the beam direction and gain.

311
The gain difference can be as large as 2 dB, resulting from  68 GHz bandwidth), or 133.3% fractional bandwidth with an 323 up-down frequency ratio of 5. The gain performances and 324 their corresponding antenna radiation efficiency are shown 325 in Fig. 6(b). It is seen that the gains are larger than 10dB 326 in most of the frequency bands. The maximum gain appears 327 at 63 GHz by 13 dBi. However, they decrease from this 328 maximum gain frequency at 63 GHz, remaining a relatively 329 stable variation. The simulated efficiency is approximately 330 94%, which is relatively stable in the achieved band.

331
The beam squints are shown in Fig. 6(c), where the maxi-332 mum deviation appears at low frequencies. In most of the fre-333 quency bands, the beam peak deviation is less than 3 degrees. 334 The large beam squints at low frequencies do not cause diffi-335 culties because of their large beamwidths with tiny gain devi-336 ations for very small gain drops at low frequencies, as shown 337 by the patterns in Fig. 7(a)-(d) at 28, 38, 60, and 72 GHz. 338 These beam squints exist in the single antenna radiations, but 339 their impacts will be reduced in an array situation because 340 the array factor may lessen this effect. The XPLs in Fig. 8 are 341 words, the mirror-imaged architectures of the Vivaldi flares 356 may well suppress the XPLs at low frequencies, which may 357 not properly suppress the high-frequency XPLs because the 358 major radiation parts of the antenna are very close to the feed-359 ing transition region to lose the symmetric natures. Indeed, 360 such asymmetric structures near the feeding transition signif-361 icantly produce X-pol components at mmW frequencies. The 362 XPLs in [17] were approximately larger than -30dB and can 363 be as large as -10dB for frequencies larger than 40 GHz. The 364 measured results were even larger than -20dB toward -17dB 365 at 40GHz.

366
The E-plane XPLs and beam squint comparisons between 367 the original BAVA antenna and the paired ones are shown in 368 Fig. 8 and Fig. 6(c), respectively. It is noted that, in princi-369 ple, the X-pol components on the E-plane should be exactly 370 canceled because they are in the opposite directions of equal 371 magnitudes in the directional beam applications. It is seen 372 from Fig. 8 that the XPLs in the paired antenna case are very 373 low on the E-plane radiation patterns, which is much superior 374 to the XPLs in the original antenna structure case. Moreover, 375 the beam squint impact has also been improved, as seen by 376

381
Experimental validation considers a single BAVA prototype, 382 as shown in Fig. 10(a), which was measured in an anechoic 383 chamber by the measurement setups in Fig. 10   connector is limited to 67GHz. The reflection coefficients 412 beyond 67 GHz are not accurate, which can be verified by 413 the measured gains in Fig. 6(b). The gains drop very rapidly 414 when the frequencies are larger than 67GHz. Based on these 415 examinations, the valid frequency range is concluded to be 416 13-85GHz. The measured radiation patterns are shown in 417 Fig. 11