Improved Design of an Ultra Wideband Universal Serial Bus Device Mounted Antenna Based on Comparative Radiation Efficiency Measurements Nuno Pires∗† , Marco Letizia† , Stephen Boyes‡ , Yang Lu‡ ,Yi Huang‡ , Anja K. Skrivervik† , Ant´onio A. Moreira∗ ∗
Instituto de Telecomunicac¸o˜ es, Instituto Superior T´ecnico Av. Rovisco Pais 1, 1049-001 Lisboa, Portugal
[email protected],
[email protected] † Laboratoire d’Electromagn´etisme et d’Acoustique, Ecole ´ Polytechnique F´ed´erale de Lausanne STI-IEL, Station 11, CH-1015 Lausanne, Switzerland
[email protected],
[email protected] ‡ Department of Electrical Engineering and Electronics, University of Liverpool, L69 3GJ, United Kingdom
[email protected],
[email protected],
[email protected]
Abstract—The present paper reports a planar, microstrip-fed Ultra Wideband (UWB) monopole antenna designed to operate inside a Universal Serial Bus (USB) dongle. The antenna was dielectrically loaded with a sandwich fabricated in commercial substrate. This design improves on a coaxial fed monopole presented earlier by showing a better matching through the UWB [3.1 - 10.6] GHz band. Wideband efficiency measurements were conducted, using the reverberation chamber technique and a cavity-based approach, called the source-stirred method. The dielectric loading implementation described here has a small impact on antenna radiation efficiency, with the loaded prototype having a figure higher than 78% over the UWB band.
II. C OAXIAL FED U LTRA W IDEBAND U NIVERSAL S ERIAL B US DEVICE MOUNTED ANTENNA WITH SANDWICH DIELECTRIC LOADING
A coaxial fed version of the Ultra Wideband Universal Serial Bus device mounted antenna with sandwich dielectric loading was designed and presented earlier [5]. This antenna was fabricated in standard FR4 substrate with εr = 4.4 and 0.8 mm thickness. It was fed by a semi-rigid 50 Ω coaxial cable with the sleeve connected to the ground plane and the tip soldered to the antenna. The USB connector was soldered to the non-used portion of substrate metallization (ground plane). The built configuration and simulation model are shown I. I NTRODUCTION in Fig. 2, and the corresponding monopole dimensions in Fig. 3. Measurements of the |S11 | parameter confirmed that this The integration demands of new consumer devices would revision covers the whole UWB bandwidth with |S | < −6 dB 11 benefit from a single antenna that operates within the require- in free space. ments of all the available wireless applications. The main issue The influence on matching of the USB casing was also undermining the approach is the wide range of frequencies that investigated and it was found to have a non-negligible effect. A the different communication standards use. As examples, Ultra slight shift towards the lower frequencies was observed when Wideband (UWB) works on the [3.1 - 10.6] GHz band in the the case was fitted, leading to a slight bandwidth increase. These USA on the [6 - 8.5] GHz band in Europe, while Bluetooth, LTE and 802.11 operate in lower frequency bands. Dielectric loading of Ultra Wideband (UWB) antennas has been presented as a technique to improve low frequency matching of a coaxial fed UWB antenna inside a Universal Serial Bus (USB) dongle. The procedure has also shown to increase the matching resilience upon connection to a laptop computer [1]. Both improvements have to be weighted against an expected decrease of antenna efficiency. This paper introduces a microstrip fed improvement to a previous design, as shown in Fig. 1. The prototype enhancements took into account antenna efficiency measurements that resorted to the reverberation chamber technique [2], [3]. Also, a novel cavity measurement approach called the source-stirred method [4] that is currently in development at the University of Liverpool was tested. Fig. 1. Improved, microstrip fed, UWB antenna, shown with USB casing and dielectric loading sandwich.
III. U LTRA W IDEBAND EFFICIENCY MEASUREMENTS
(a)
(b)
(c)
Fig. 2. Initial, coaxial fed, revision of the UWB antenna, shown (a) bare; (b) mounted inside the USB dongle; and (c) the simulation model with dielectric loading sandwich.
differences can be explained by a dielectric loading effect of the casing on the antenna as the plastic casing has an electric permittivity greater than air, considered to be εP V C ' 4.5, in the simulations. The dielectric loading effect of the case was reinforced by padding both sides of the radiating structure with two dielectric layers – called sandwich – with high relative electric permittivity. Implementations of the technique on UWB monopoles have been discussed in [6], adding to other available miniaturization techniques discussed in the literature such as [7], [8], for instance. Several materials and thicknesses of the dielectric sandwich were simulated and the best configuration was selected. It consisted of a pair of equal sized 60 mil (1.524 mm) thick RO3003 sandwich slices with εRO3003 ' 3. The sandwich was fixed to the antenna substrate using photographic aerosol glue. The |S11 | measured and simulation results showed that the technique further decreases the lower antenna operating frequency, while introducing a mismatch around 10 GHz. This prototype covers the [3.006, >12] GHz band with |S11 | < −6 dB. With |S11 | < −10 dB the antenna lowest operational frequency is 3.231 GHz.
Essential to the usefulness of the UWB dielectric loading sandwich technique discussed in the last section is its impact on antenna efficiency. To assess this, the efficiency of the coaxial fed dielectric loaded UWB monopole was measured. Reverberation chamber measurements were conducted at the University of Liverpool’s High Frequency Engineering Research Group. A novel and simple procedure of UWB efficiency estimation, called the source-stirred method, being developed at University of Liverpool was also used. The measurement campaigns were conducted at Instituto de Telecomunicac¸o˜ es/Instituto Superior T´ecnico (IT/IST) of the Technical University of Lisbon and at University of Liverpool. The prototype discussed before provides a good range of measurement configurations to test the novel method, applied to small UWB antennas, as several configurations can be selected: with or without sandwich as well as with or without USB casing. This research effort tries to contribute to the ongoing discussion on the extension of the narrow-band efficiency concept to UWB antennas [9]–[12]. A. Reverberation chamber The measurement of efficiency in the reverberation chamber uses two antennas besides the antenna under test (AUT). One of these, the reference antenna (REF), has a known radiation efficiency. A Vector Network Analyzer (VNA) measures the scattering parameters in a transmit-receive system in which the REF is first measured and then substituted by the AUT. The radiation efficiency can be calculated by Eq. 1. In this mathematical expression, the brackets hi mean an average of measurements from a mode stirred ensemble. ηAUT =
2 2 hS21AUT i 1 − S22REF × × ηREF 2 2 hS21REF i 1 − S22AUT
The setup for this measuring campaign included an Anritsu 37369A VNA and a Satimo SH2000 Dual Ridge Horn Antenna as reference. The chamber is constructed from individual Point A B C D E F G H I L M N O P Q
(a)
(b)
Fig. 3. Dimensions (in mm) of the UWB antenna: (a) front view; (b) detail of the antenna radiating element with the corresponding path point coordinates shown in TABLE I.
(1)
x 1.5 2.51 2.58 8 9 11.35 14.35 14.25 12 11.5 11.75 3.5 0 3.1 0
y 8.5 4.45 2.21 0.29 0.29 1.31 6.11 8.75 9.5 10.77 12.5 12.5 0 13 9.9
TABLE I C OORDINATES OF THE POINTS ( IN MM ) THAT DEFINE THE BUILT PROTOTYPE RADIATING ELEMENT EDGE , AS SEEN IN F IG . 3 A .
Radiation efficiency/%
aluminium panels and its dimensions are 3.6×5.8×4 m3 (W×L×H). It is fitted with two principle sets of mechanical s cap 2 free |2 stirring paddles; one is mounted about a rotational shaft from |S11 | − |S11 (2) ηˇ = cap free |2 + |S free 2 2 the floor to the roof and configured for vertically polarised wave 1 − 2 × |S11 11 | × |S11 | stirring; the other set is configured for horizontally polarised The UWB antenna radiation efficiency is then estimated by waves mounted about a rotational shaft from the front to the considering only the joint peak values of all the calculated back wall at ceiling height. The 200 independent samples η ˇ (f ) functions. This approach will ideally cancel the plot dips generated when using 1 degree mechanical stirring for each caused by the several resonant modes excited in the cavities. (10 MHz) frequency point results in an expected uncertainty The source-stirred method was applied using two metallic of approximately ± 5%. This figure should be improved by a caps available at IST/IT [13]. The cavities are pictured in post-processing frequency stirring, i.e. a smoothing algorithm Fig. 5b, along with measured prototypes. As shown, the cavities on the frequency data. are cylindrical, with dimensions 2alarge = 14 cm, dlarge = The results, after processing are shown in Fig. 4 as radiation 20 cm and 2asmall = 9 cm and dsmall = 15 cm, for the large efficiency (in %) vs frequency plot. As expected, the loaded and small case, with 2a and d indicated in schematic of Fig. 5a. prototype presents a lower efficiency at higher (>'6.5 GHz) The measurements were done in three locations inside each frequencies. It is interesting to note that the phenomenon is cavity, by using two different coaxial extensions added to the reversed in lower frquencies (
