This paper presents a novel printed uniplanar antenna architecture for circular polarization. The structure consists of a single-fed slot-ring antenna with asymmetrically placed perturbations. The influence of different kinds of perturbations and substrates on the size of the antenna, its impedance bandwidth, and its axial ratio bandwidth is investigated. Various feed circuits based on coplanar waveguides (CPWs), coaxial line, and microstrip are investigated as well. Low-cost applications such as tagging antennas at 2.45 GHz are tested in combination with a coaxial line and CPW feed circuit. It was found that antennas achieving an impedance bandwidth over 60% and an axial ratio bandwidth up to 15% can be realized without the need of airbridges and rf-substrates.
 Printed planar antennas are attractive for a variety of applications due to their low profile and ease of fabrication. Two basic types of printed planar antennas are very popular, patch antennas and slot antennas. Patch antennas exhibit a one-sided radiation pattern and a narrow impedance bandwidth [James and Hall, 1989] whereas slot antennas radiate to both sides and have high impedance bandwidths. Printed slot antennas are usually fed by microstrip lines or by coplanar waveguides (CPWs). The advantage of the CPW-feed is that for uniplanar antennas the metal layer of both the feed line and the radiator is on the same side of the substrate. This eliminates the need for vias, potentially lowers the production costs and offers easy integration with active devices.
 Substantial work has been done on linearly polarized printed slot antennas [i.e., Daniel et al., 1989; Garg et al., 2001]. Today's applications, however, require for the most part circular polarization. Little work has been published on this topic up to now. Circular polarization results when two orthogonal linear modes are excited with equal amplitude and 90° phase shift between them. This can be achieved by either using a single feed or a dual feed [Haneishi and Suzuki, 1989; Karmakar and Bialkowski, 1999]. In the dual feed solution the feed network splits the signal in two equal parts with 90° phase shift, which are then fed to the antenna at 2 different locations. The feed networks consist either of a branch line coupler or a T-junction followed by two feedlines with a quarter wavelength difference (λ/4). The use of a branch line coupler yields a wide axial ratio bandwidth (ARB) but occupies also nonnegligible space on the printed circuit board. Furthermore, a branch line coupler is inconvenient for low-cost uniplanar technology (CPW) because several airbridges are needed. A T-junction requires less space but yields also a smaller ARB. In the single feed solution the circular polarization is produced by an asymmetry in the antenna itself. Similar to the use of the T-junction this allows a small feed network but reduces the achievable ARB. In the work of Qing and Chia  a dual microstrip-fed circularly polarized slot-ring antenna using a branch line coupler was presented. An ARB of 22% (defined at the 3 dB points) was achieved. The combined size of the radiating slot structure and the feed network was 0.3λ × 1λ. In the study by Qing and Chia  a single microstrip-fed circularly polarized slot-ring antenna was presented showing an ARB of 11%. In this case the size of the antenna was about 0.35λ × 0.35λ.
 Very few papers treating the topic of uniplanar circularly polarized slot antennas can be found in the literature. A CPW-fed circularly polarized rectangular slot antenna operating at 25 GHz and achieving an ARB of 9.9% was presented by Soliman et al. . The design utilizes a CPW-slotline T-junction in order to feed the antenna at two different points. The overall size of the antenna including feed network was about 0.31λ × 0.47λ. Matsuzawa and Ito  presented a CPW-fed circularly polarized antenna consisting of two circularly polarized radiating elements. The achieved ARB was only 3% and the overall size of the antenna was about 0.4λ × 0.52λ.
 In the following a variety of uniplanar antenna architectures and feedline techniques are investigated. Design rules are derived which allow to improve the performance of the above designs. As a result of this investigation four antennas were fabricated utilizing three different feeding techniques. Two of the antennas are uniplanar designs, which are especially suited for low-cost tagging applications in the 2.45 GHz ISM-band. The first one is built on a low-cost FR4 dielectric and utilizes a CPW-feed whereas the second one does not need any rf-dielectric nor any airbridges achieving an impedance bandwidth over 40% and an ARB of 15%. Although from a technical point of view single sided radiation is usually preferred in tagging applications [Fries et al., 2000] this antenna is very interesting because of its low fabrication costs and its large impedance bandwidth.
2. Antenna Design
 The antenna architectures investigated in the following can be divided into two parts: First the asymmetrically shaped circularly polarized slot-ring radiating element fed by a slotline and second the feed circuit, which matches the input impedance of the radiating element to the 50 Ω input signal. In a first step different radiating slot architectures on different substrate materials will be investigated in terms of ARB, size and input impedance. In a second step 3 different feed structures will be discussed. The simulations are performed with ENSEMBLE from Ansoft assuming an infinite groundplane.
2.1. Investigation of Antenna Architecture
 The linearly polarized slot-ring (also-called slot-loop) antenna is well known in rectangular [Greiser, 1976] as well as in circular form [Daniel et al., 1989; Chang, 1996]. To produce circular polarization with a single-fed slot-ring antenna an asymmetry with respect to the feed has to be introduced in the slot-ring geometry. The asymmetry of the investigated antennas consists of two perturbations in the slotline ring, which are located at 45° and at 225° with respect to the feed point. Different kinds of perturbations can be used. Their shape influences the size, the ARB and the input impedance of the antenna. Besides the shape of the perturbation these antenna parameters depend also on the substrate (εr, thickness) and the slotline width. In order to determine the optimal perturbation for a given specification four different kinds of perturbations (type B-E, Figure 1) are investigated together with a linearly polarized slot-ring antenna without perturbations (type A, Figure 1). The latter serves as a reference structure. The first and second perturbation (type B and C, Figure 1) consist of wider slotline sections. In one case widened toward the outer side of the slot-ring (type B) and in the other case widened toward the inner side of the slot-ring (type C). The perturbation of antenna D consists of a narrowed slot-ring section. In the fifth antenna (type E) the perturbation consists of a narrowed ring section that is bent inside the ring.
 Every antenna type was simulated with slot widths ranging from 0.5 mm to 5 mm and for two different substrates with a thickness of 0.635 mm and a permittivity of εr = 2.2 and εr = 10.2, respectively. For each slot width and substrate the perturbation was adjusted (dimension indicated by arrows in Figure 1) to achieve a maximum ARB.
Figure 2 shows the antenna diameter, which is the largest dimension of the antenna normalized to the wavelength versus the slot width. It increases with the slot width due to the larger guided wavelength. The antenna diameter decreases with the higher permittivity of the substrate due to the shorter guided wavelength. Compared to the linearly polarized antenna the antenna diameter is increasing for type B and decreasing for all other types. The smallest antenna size is achieved with the perturbation of type E where the electrical length of the ring is increased by bending a part of the line to the inside of the ring.
Figure 3 shows the ARB versus the antenna diameter. It can be seen that there is a tradeoff between the antenna size and the ARB. Antenna B outperforms clearly all other antennas in terms of ARB but at the expense of a larger size. The best ratio between the ARB and the diameter is achieved with Antenna E. Antenna diameters down to 0.2 wavelengths can be achieved.
Figure 4 shows the real and imaginary part of the input impedance at the operating frequency versus the slot width. The real part of the input impedance increases with the slot width. It is higher for the circularly polarized antennas than for the linearly polarized antenna A. Comparing the real part of the input impedance versus the ARB instead of the slot width it can be seen that for a given substrate and ARB the real part of the input impedance of the different antennas is similar except for antenna C. Contrary to the linearly polarized antenna A the imaginary part of the input impedance of the circularly polarized antennas is not zero at the operating frequency but depends strongly on the type of perturbation and the substrate. Antenna B shows a strongly negative imaginary part whereas the antennas C and D show a positive imaginary part on the low permittivity substrate.
 To conclude the investigation on the radiating element it can be stated that the single fed circularly polarized slot-ring antenna is a very flexible radiating element. It can be optimized either for small size (type E) by limiting the bandwidth or for a wide ARB (type B) by accepting a larger physical size. The widening of the slot width and use of a low permittivity substrate increases the ARB but also increases the input impedance. Therefore, the maximum achievable ARB is determined by the maximum input impedance that can be matched to a 50 Ω feed.
 The antennas fed by a slotline possess an input impedance ranging from 10 Ω up to 300 Ω with high imaginary parts. A suitable feed structure must be found to match this antenna to a 50 Ω input signal coming from an unbalanced feed. In this paragraph we will present three kinds of feed lines (Figure 5) that are potentially suitable for this task. The first feed is a microstrip slotline transition [Schuppert, 1988]. This transition is not uniplanar, but it allows an easy matching of impedances with real parts ranging from 50 Ω up to 100 Ω and the compensation of the imaginary part. The second transition is a CPW-slotline transition [Ma and Itoh, 1997]. This transition is uniplanar but not as flexible as the previous one and it requires an airbridge. Typically it allows to match a 50 Ω CPW feed to a 200 Ω antenna input impedance. Finally the third kind of feed matches the antenna to a coaxial feedline. It is uniplanar and in addition does not need any airbridges such as the previous transition. It allows to match impedances over the range of 150 Ω to 250 Ω to a 50 Ω coaxial feedline.
 Several antennas were simulated and built. Type B and D (Figure 1) were built on a 0.635 mm thick RT/Duroid 3010 substrate with a permittivity of εr = 10.2 and on a 0.508 mm thick RT/Duroid 5880 substrate with a permittivity of εr = 2.2, both shown in Figure 6. These antennas were fed by the microstrip feed shown in Figure 5. They allow to verify the simulation results of paragraph 2.
 In a second step the focus is on low-cost uniplanar antennas yielding a wide ARB. Therefore antenna B was realized on a 0.8 mm thick FR4 substrate with a permittivity of εr = 4.4 and on a 0.127 mm thick RT/Duroid 5880 dielectric with a permittivity of εr = 2.2 (Figure 7). For the antenna on FR4 a uniplanar CPW-feed and for the antenna on RT/Duroid a coaxial line feed was used.
 All antennas showed a −10 dB impedance bandwidth of more than double the ARB, which is due to the two resonances excited in the single-fed circularly polarized slot-ring antenna. These resonances lead to a significantly larger impedance bandwidth than for the linearly polarized slot-ring antenna.
 The antennas were simulated with ENSEMBLE from Ansoft and HFSS from Agilent. ENSEMLE is a 2.5-D full wave field solver, which is restricted to planar layered structures and infinite groundplanes. HFSS is a 3-D full wave field solver based on the finite element method. Because the simulated antennas have finite groundplanes the ENSEMBLE results showed some significant differences compared to the measurement. The finite groundplane size alters the input impedance as well as the radiation pattern. This was better accounted for in the HFSS-simulation, which showed better agreement with the measurement for the radiation pattern of all antennas and the input impedance of all antennas except the CPW-fed antenna in Figure 8.
Figure 6 shows the antenna of type B and type D fed by a microstrip line. The microstrip feed shows some advantages here because it allows matching a variety of impedances without interfering with the radiation characteristics of the antenna. This allows for easy verification of the results of paragraph 2. The antennas were built on a 0.635 mm thick RT/Duroid 3010 dielectric with a permittivity of εr = 10.2 and on a 0.508 mm thick RT/Duroid 5880 substrate with a permittivity of εr = 2.2. Table 1 shows the dimensions and measurement results of the different antennas. The diameter given in the table does not include the feed structure and thus allows direct comparison with the results of paragraph 2. The obtained ARB's agree well with the predictions of paragraph 2.
Table 1. Dimensions and Measurement Results of Microstrip-Fed Circularly Polarized Slot-Ring Antennasa
 This section introduces a low-cost uniplanar antenna design operating at 2.45 GHz with maximum achievable ARB and fed by a CPW-feed. It was shown in paragraph 2 that the lower the permittivity of the substrate the higher the ARB becomes. Unfortunately, the dimensions of the CPW are not practical for permittivities below εr = 4. Therefore a 0.8 mm thick low-cost FR4 substrate with a permittivity of εr = 4.4 and a tanδ = 0.025 was chosen. It was found in paragraph 2 that for a given dielectric, antenna B yields the highest possible ARB which is achieved by widening the slot width. At the same time a higher input impedance is obtained. Therefore the largest possible slot width, which could be matched by the CPW-feed (Figure 5) was chosen. On FR4 this turned out to be 220 Ω–j120 Ω, which leads to a slot width of 4 mm as shown in Figure 7. The width of the perturbation region was increased to 20 mm compared to the antennas in Figure 6 because it was found that this measure lowers the imaginary part of the input impedance. Figure 8 shows the simulated and measured return loss, axial ratio and input impedance. The antenna shows a very wide −10 dB impedance bandwidth which is better than 40%. The measurement agrees well with the simulation (ENSEMBLE). The ARB predicted by the simulation is 9.3% whereas the measurement yields 5.9%. Apparently this is due to the effect of the finite groundplane, which was not considered in the simulation. The antenna radiates RHCP to the front side and LHCP to the backside. The measured radiation pattern (2.45 GHz) agrees well with the HFSS simulation. The radiation pattern is asymmetric due to the asymmetric shape of the antenna. The simulated gain is 2.5 dBi and the measured gain 2.3 dBi. Although FR4 is quite a lossy substrate the losses of the antenna are small compared to a patch antenna because most of the field remains in air. The size of the antenna together with the feed is 0.4λ × 0.4λ.
 The best substrate in terms of high ARB and low input impedance is air. However, in order not to loose the advantage of printed circuit technology the next best substrate is the 0.127 mm thick RT/Duroid 5880 material with a permittivity of εr = 2.2. For slot structures with slot dimensions much wider than the thickness of the substrate, the electromagnetic field behaves similar as in an air substrate. Because the CPW is not practical for such a substrate a coaxial line-slotline transition has been developed shown in Figure 5. At the feed point the ground conductor of the coaxial line is connected to one side of the slotline while the inner conductor runs across the slot and is connected to the other side of the slotline. The ground conductor (outer diameter: 4 mm) is electrically attached to the groundplane. If the antenna were connected to an electronic device, the latter would be connected exactly at the same point where the coaxial line feeds the slotline. To achieve the maximum ARB antenna B was chosen with the largest slot width that can be matched with the coax to slotline transition. This leads to a slot width of 6 mm (Figure 8) and an impedance of 260 Ω–j120 Ω. Note that no air bridges are required and the slot width of the feed structure is not a critical dimension (0.4 ± 0.1 mm). Figure 9 shows the measured and simulated return loss, axial ratio and radiation pattern at 2.45 GHz. The return loss was simulated with HFSS because the coaxial line can not be simulated with ENSEMBLE. The agreement between simulation and measurement of the input impedance are not as good as for the antenna on FR4 but still reasonable. The −10 dB impedance bandwidth is more than 60%. While the simulated ARB of 15% agrees well with measurements, the agreement between simulated and measured radiation pattern is not that good but still reasonable (Figure 9). The size of the antenna is 0.54λ × 0.45λ. Front side radiation is now LHCP and, because the structure is mirrored, backside radiation is RHCP. The simulated gain is 2.6 dBi whereas the measurement yields 2.1 dBi. The potential for low-cost fabrication of this antenna is obvious: first of all the entire structure is uniplanar, Second, no rf-dielectric is needed, thirdly, airbridges are eliminated and, fourthly, the structure is largely insensitive to manufacturing tolerances. The electrical performance exhibits a wide impedance bandwidth and ARB, and the size of the antenna is also attractive for a large variety of applications.
 This paper has shown a new single-fed uniplanar circularly polarized slot-ring antenna architecture. The circular polarization is excited by two asymmetrically placed perturbations in the slot-ring. Four different types of perturbations and two different substrates have been investigated. Their influence on the antenna size, the axial ratio bandwidth and the input impedance of the antenna has been shown and experimentally verified. Two low-cost uniplanar circularly polarized antennas, which are attractive for tagging applications, have been demonstrated. The first one was built on a FR4 dielectric and fed with a CPW. The second one was fed by a coaxial line and does not need either an rf-dielectric nor any airbridges. They impedance bandwidth obtained is larger than 60% and the ARB 5.9% and 15%, respectively.