366
IEEE ANTENNAS AND WIRELESS PROPAGATION LETTERS, VOL. 14, 2015
Design of a Planar UWB Dipole Antenna With an Integrated Balun Using Surrogate-Based Optimization Slawomir Koziel, Senior Member, IEEE, Stanislav Ogurtsov, W. Zieniutycz, and A. Bekasiewicz
Abstract—A design of an ultrawideband (UWB) antenna with an integrated balun is presented. A fully planar balun configuration interfacing the microstrip input of the structure to the coplanar stripline (CPS) input of the dipole antenna is introduced. The electromagnetic (EM) model of the structure of interest includes the dipole, the balun, and the microstrip input to account for coupling and radiation effects over the UWB band. The EM model is then adjusted for low reflection over the UWB band by means of fast simulation-driven surrogate-based optimization. This approach allows us to obtain the final design at low computational costs and at a high-fidelity level of structure description. Measurements of the manufactured optimal design validate the use of the balun as well as the design approach. Index Terms—Microstrip-to-coplanar-stripline (CPS) transition, numerical optimization, radial line stub, simulation-based model, surrogate-based optimization, ultrawideband (UWB) antenna, UWB balun, UWB dipole.
I. INTRODUCTION
P
ROPER yet simple interfacing of balanced and unbalanced transmission lines is critical for ultrawideband (UWB) circuits, in particular for UWB antennas [1]. A typical unbalanced line is a microstrip, and a typical balanced line is a coplanar stripline (CPS), which is commonly used as an input to the dipole antenna (balanced element). A balun element interfaces such lines to each other, e.g., as shown in Fig. 1. Different balun geometries for UWB antennas have been introduced so far with various levels of complexity [2]–[6]. Among these, balun structures delivering acceptable performance, structural simplicity, as well as compact footprint, all at the same time, are preferred for UWB antenna applications, e.g., [7]. The UWB band of interest dictates to use full-wave analysis not only for validation of the final design, but also through the design adjustment process to account for EM interactions within the antenna structure. It is also desirable for the electromagnetic (EM) model of the structure to include the antenna ele-
Manuscript received August 18, 2014; accepted October 14, 2014. Date of publication October 17, 2014; date of current version February 04, 2015. This work was supported in part by the Icelandic Centre for Research (RANNIS) under Grant 141272051. S. Koziel, W. Zieniutycz, and A. Bekasiewicz are with the Faculty of Electronics, Telecommunications and Informatics, Gdansk University of Technology, Gdansk 80-233, Poland (e-mail:
[email protected];
[email protected]). S. Ogurtsov is with the Engineering Optimization and Modeling Center, Reykjavik University, Reykjavik 101, Iceland (e-mail:
[email protected]). Color versions of one or more of the figures in this letter are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/LAWP.2014.2363932
Fig. 1. UWB dipole antenna with a balun: layout.
ment, balun, and microstrip input to reliably account for coupling and radiation effects over the UWB band. Manual adjustment of such EM models via simulation sweeps with one parameter active at a time is tedious, time-consuming, and hardly feasible even with few design variables. Therefore, we adopted automated numerical optimization to conduct the design. In particular, we utilize surrogate-based optimization (SBO) [10] to conduct design at a high-fidelity level of description yet at low computational costs. This letter is organized as follows. The antenna geometry including the proposed balun, EM models utilized in the design process, and the design task are described in Section II. Section III outlines the SBO design process. Section IV presents optimization and measurement results. Section V concludes the work. II. ANTENNA GEOMETRY, DESIGN TASK, AND EM MODELS The UWB antenna, shown in Fig. 1, should be matched within the UWB band of 3.1–10.6 GHz. The antenna comprises a planar dipole with a CPS input of length , a balun of length , and 50 microstrip input. The balun is a microstrip-to-CPS transition with a ground edge of a linear profile and an open radial microstrip stub, both as shown in Fig. 1. The radial stub element was adopted because it allows, in general, more broadband operation than the microstrip stub; additionally, it has two degrees of freedom. The radiating element is an elliptical dipole. The substrate is a 0.76-mm-thick Taconic RF-35 layer [8]. Metallization is with 70 m copper. The vector of design variables contains dimensional parameters of the dipole, CPS section, and the balun, namely . Other parameters shown in Fig. 1 are fixed as follows: , and , all in millimeters.
1536-1225 © 2014 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
KOZIEL et al.: PLANAR UWB DIPOLE ANTENNA WITH INTEGRATED BALUN USING SURROGATE-BASED OPTIMIZATION
367
Fig. 2. UWB dipole antenna with a balun at a certain design: typical differand the high-fidelity one ences between the low-fidelity EM model (---). It can be observed that the major types of discrepancies are the frequency shifts and the vertical misalignment. Therefore, frequency scaling and additive correction are utilized as surrogate modeling techniques.
Fig. 4. UWB-dipole antenna with balun: manufactured final design.
Fig. 3. Simulated reflection response: the initial design (---) and the final design without connector ( ) and with connector (—) as in Fig. 4.
The structure is modeled in the CST MWS environment [9] and simulated with the MWS transient solver. Two discrete EM models are defined: the high-fidelity model and the low-fidelity model . Models and are utilized in the automated design process as described in Section III. The mesh density (the number of meshes per characteristic wavelength) of the high-fidelity model is set with preliminary numerical experiments ensuring that no substantial changes in the reflection response occurs with further increase of the mesh density. At the initial design, the model comprises 11 180 680 hexahedral cells and simulated in 53 min, while the model is with 574 175 cells and simulated in 1 min 40 s. The coarse model is much faster than the high-fidelity model . However, it is less accurate, so to be reliably used in the optimization process, should be corrected relatively , e.g., as described in Section III. It was found with preliminary numerical experiments executed in the vicinity of the initial design that lateral extensions of the finite substrate have no noticeable influence on both reflection and radiation responses of the discrete EM models if dielectric layer extends more than 10 mm beyond metallization of the dipole. In addition, if the finite dielectric layer of a particular design extends more than 10 mm beyond metallization of the dipole, then its reflection response is essentially the same as that of the same design defined at the substrate modeled with infinite lateral extends, and the maximal difference of the radiation responses of such models stays within 0.5 dB. Therefore, dielectric substrate of the EM models is modeled as finite and extending 15 mm beyond metallization of the dipole. It was assumed that, in actual applications, the UWB antenna under design should be excited through a microstrip line
Fig. 5. Reflection response of the final design: simulated with connector (—) and measured (---) shown in Fig. 4.
of the printed board, i.e., there should be no connector in the close proximity of the antenna. Consequently, the discrete EM models to be used in the optimization process were defined with the waveguide port defined at the microstrip input. At the same time, an SMA connector interfaces the manufactured antenna with a vector network analyzer (VNA) in measurements. Therefore, a high-fidelity model with an SMA connector had been also defined for verification of the final design and its comparison to measurement data. III. DESIGN OPTIMIZATION PROCEDURE The design task can be formulated as a nonlinear minimization problem of the form (1) where is an objective function encoding the design specifications, here minimizing in the UWB frequency range. Perhaps the most serious bottleneck in solving (1) is the high computational cost. In order to make the design optimization process feasible in terms of the CPU time, most operations are performed on a corrected low-fidelity model, a so-called surrogate model The optimization algorithm is an iterative procedure that yields a sequence of approximations , of the optimum design [10] (2)
368
IEEE ANTENNAS AND WIRELESS PROPAGATION LETTERS, VOL. 14, 2015
Fig. 6. Normalized power pattern of the final design: simulated with connector (—) and measured (---) at selected frequencies: (a) 3, (b) 4, and (c) 5 GHz. 90 on the left corresponds to the normal direction above the antenna. 180 corresponds to “to-the-connector.”
where is the surrogate model at iteration . The surrogate model is constructed by suitable correction of the low-fidelity model . Here, we use the following two types of techniques: 1) frequency scaling, and 2) additive response correction. The reason for this choice is the fact (cf. Fig. 2) that the major types of misalignment between the responses of and are frequency shifts and vertical discrepancy. Let denote the explicit dependency of the low-fidelity model on the frequency ( is the set of frequencies of interest at which the model is evaluated). The surrogate model is defined as (3)
Fig. 7. Normalized power pattern of the final design: simulated with connector (—) and measured (---) at selected frequencies: (a) 6 GHz, (b) 7 GHz, (c) 9 GHz, and (d) 10 GHz. 90 degree on the left corresponds to the normal direction above the antenna. 180 degree corresponds to ‘to-the-connector’.
being the affine frequency scaling (shift and scaling) [11]. The frequency scaling parameters are obtained as (6)
with (4) and (5)
i.e., to minimize the misalignment between the high- and the scaled low-fidelity model response at . Although the models are evaluated at a discrete set of frequencies, the information at other frequencies can be obtained through interpolation. The misalignment is further reduced by the additive response correction term (output SM) (4) that ensures zero-order consistency
KOZIEL et al.: PLANAR UWB DIPOLE ANTENNA WITH INTEGRATED BALUN USING SURROGATE-BASED OPTIMIZATION
(i.e., ) between the surrogate and [12]. The algorithm (2) working with the surrogate model (3)–(6) typically requires only a few iterations to yield an optimized design.
369
computational costs at a high-fidelity level of structure description. Measurements of the reflection and radiation responses of the manufactured optimal design validate the use of the balun for UWB-dipole antenna well as the design approach.
IV. NUMERICAL RESULTS AND MEASUREMENTS REFERENCES
The initial design is , where is in degrees and other variables are in millimeters. The optimum, , has been found in four iterations of the optimization procedure that was described in Section III. Each iteration required about 120 low-fidelity model evaluations and one evaluation. The reflection response of the final design is shown in Fig. 3. The total numerical cost of obtaining this design corresponds to about 19 simulations of the UWB antenna high-fidelity model. A photograph of the manufactured design is shown in Fig. 4. The UWB antenna under test was excited through an edge-mount SMA connector [13]. Radiation and reflection responses of the fabricated designs have been have been measured at the anechoic chamber of Gdansk University of Technology, Gdansk, Poland, using a setup with a dual-polarized horn antenna [14] and E5071C ENA Network Analyzer [15]. Simulated and measured reflection responses are shown in Fig. 5. The discrepancies between both responses are the result of manufacturing inaccuracies. Simulated and measured radiation responses in the plane perpendicular to the antenna substrate at selected frequencies are shown in Figs. 6 and 7, from where one sees that the final design stays of omnidirectional radiation in this plane up to 7 GHz. V. CONCLUSION A fully planar balun configuration interfacing the microstrip input of the structure to the CPS input of the dipole antenna was introduced. Simulation-driven design of a UWB antenna with the balun had been performed using surrogate-based optimization. This approach allowed us to obtain the final design at low
[1] H. Schantz, The Art and Science of Ultrawideband Antennas. Norwood, MA, USA: Artech House, 2005. [2] J. W. Duncan and V. P. Minerva, “100:1 bandwidth balun transformer,” Proc. IRE, vol. 48, no. 2, pp. 156–164, Feb. 1960. [3] M. Karlsson and S. Gong, “Circular dipole antenna for mode 1 UWB radio with integrated balun utilizing a flex-rigid structure,” IEEE Trans. Antennas Propag., vol. 57, no. 10, pp. 2967–2971, Oct. 2009. [4] Y.-G. Kim, D.-S. Woo, K. W. Kim, and Y.-K. Cho, “Design of bow-tie type UWB antennas using an ultra-wideband balun,” in Proc. IEEE APS Int. Symp., 2007, pp. 1989–1992. [5] S. Kubota, X. Xiao, N. Sasaki, K. Kimoto, and T. Kikkawa, “Characteristics of UWB bow-tie antenna integrated with balun for breast cancer detection,” in Proc. IEEE APS APSURSI, Jun. 1–5, 2009, pp. 1–4. [6] G. Yang, Q.-X. Chu, Z.-H. Tu, and Y. Wang, “Compact printed dipole antenna with integrated wideband balun for UWB application,” in Proc. ICMMT, May 5–8, 2012, vol. 5, pp. 1–4. [7] T. B. Lim and L. Zhu, “Compact microstrip-to-CPS transition for UWB applications,” in Proc. IEEE MTT-S Int. Microw. Workshop Ser. Art Miniaturizing RF Microw. Passive Compon., Dec. 14–15, 2008, pp. 153–156. [8] Taconic, Petersburgh, NY, USA, “ORCER RF-35,” Data sheet, 2002 [Online]. Available: http://www.taconic-add.com/pdf/rf35.pdf [9] CST Microwave Studio. ver. 2013, CST AG, Darmstadt, Germany. [10] S. Koziel, J. W. Bandler, and K. Madsen, “A space mapping framework for engineering optimization: Theory and implementation,” IEEE Trans. Microw. Theory Tech., vol. 54, no. 10, pp. 3721–3730, Oct. 2006. [11] J. W. Bandler et al., “Space mapping: The state of the art,” IEEE Trans. Microw. Theory Tech., vol. 52, no. 1, pp. 337–361, Jan. 2004. [12] N. M. Alexandrov, J. E. Dennis, R. M. Lewis, and V. Torczon, “A trust region framework for managing use of approximation models in optimization,” Struct. Multidiscipl. Optim., vol. 15, no. 1, pp. 16–23, 1998. [13] Johnson/Emerson Connectivity, Bannockburn, IL, USA, “SMA connector, part no. 142-0701-881,” 2012, p. 1179. [14] Agilent Technologies, Santa Clara, CA, USA, “Agilent E5071C ENA network analyzer,” Data Sheet, 2013. [15] QPar Angus, Ltd., Leominster, U.K., “WBHDP0.9-18S dual polarized horn antenna,” “Horn antennas,” Product catalog, 2012.
