Abstract

In the paper, a wideband miniaturized impedance-transforming quadrature four-feed network with a flat output phase difference is presented and applied to the design of an active integrated GNSS antenna where no extra impedance matching circuit is needed. The features of impedance transformation and flat output phase difference are achieved by the proposed miniaturized rat-race coupler. When combining the proposed rat-race coupler with two trans-directional (TRD) couplers, a four-feed network with stable sequential quadrature phase shifts is obtained in the whole GNSS band. Since the quadrature four-feed network has the feature of impedance transformation, integration with a low-noise amplifier (LNA) can be realized without extra impedance matching circuits, which reduce the overall size and losses. For validation, a simple rectangular patch is applied as the radiator, and the active prototype is fabricated. Measurement results show that over the entire GNSS band from 1.164 GHz to 1.610 GHz, the miniaturized integrated antenna exhibits a return loss of more than 10 dB, an axial ratio of less than 3 dB axial ratio, and a gain of greater than 16 dBic.

1. Introduction

Nowadays, the global navigation satellite systems (GNSSs) have become indispensable in civil and military fields, which involved in the applications such as vehicle positioning, ship navigation, and disaster relief [1]. To maintain high accuracy in global positioning, the GNSS terminal should have the ability to receive the GPS, GLONASS, Galileo, and BDS signals, which indicates that the GNSS antenna operates in the 1.164–1.3 GHz and 1.55–1.61 GHz [2].

For GNSS applications, dual-band, multiband or wideband circular polarized (CP) antennas are proposed. In the designs of dual/multiband CP antennas, structures such as shorted annular ring [3, 4], concentric ring patches [5, 6], two-element conformal structure [7, 8], and reconfigurable technology [9] are presented. However, since the dual/multiband is realized by increasing the complexity of the radiator, the fabrication difficulty is increased. Besides, these structures are difficult to cover all GNSS bands.

Wideband CP antennas are demonstrated to be good candidates for GNSS applications, which can be compatible with different navigation systems in the simple structure. Common wideband CP structures include crossed dipoles [10, 11], Archimedean spirals [12, 13], slot antennas [14], and feed network-based antennas [1520]. However, metallic cavities are always applied in crossed dipoles, slot antennas, and spiral antennas for obtaining directional radiation, which increases the profile and cost. Thus, feed network-based antennas have been attractive in designing wideband CP antennas due to the allowed simple structure of the radiator. In [15], by using the dual-band power divider, the 90° phase shifter, and the 180° delay lines to form the feed network, a wide CP bandwidth of 72% is obtained. In [17], a GNSS antenna is designed with a wideband feed network consisting of one power divider and two types of wideband differential phase shifters. Similar structures of the feed network can also be found in [18]. However, these feed networks [15, 17, 18] are both based on the connections of three-stage circuits, which is large and complex. To reduce the volume, metamaterial transmission lines are used as a replacement for obtaining phase shifts [18]. In [20], a compact feed network composed of one rat-race coupler and two quadrature hybrid couplers is designed, where the phase shift is fused between the two components. However, since traditional rat-race couplers and three-branch line couplers are used, the size is still large.

Besides, most of the reported wideband GNSS antennas are passive antennas, where the low-noise amplifier (LNA) is not considered. In general, the passive GNSS antennas and the LNA are designed separately, and then directly connected to each other in applications [21]. For example, based on standard 50 Ω impedance matching, a microstrip antenna is integrated with a LNA in coplanar for application in the GPS receiver [22]. In [23], a metasurface-inspired cavity antenna is designed and connected with a commercial LNA through coaxial lines to form the active GNSS antenna. An average gain of 15 dBi is measured for the active antenna in the range of 1.38 GHz to 1.7 GHz. In [24], a two-stage LNA with more than 26 dB gain is implemented on the bottom layer of a GNSS antenna (covering the GPS L1 and GLONASS L1 bands). The port impedances of the antenna and LNA are both 50 Ω and a direct connection is used. Through the direct connection, disadvantages such as increased cost, poor loss, and large size appear. Besides, the performances of the devices via connection may be deteriorated. With the development of modern technology, high-integrated terminals are more preferred for low loss, reduced size, and improved performance. Thus, the integrated design of active GNSS antennas is the future trend. Here, the integration is different with direct connection, where the input port impedances of the passive antenna and the LNA are not 50 Ω. To the author’s knowledge, no report has been found for the effective integration of a passive GNSS antenna and the LNA.

In the paper, an active integrated wideband GNSS antenna is proposed with the contribution of an impedance-transforming quadrature four-feed network. By using the proposed rat-race coupler which features the characteristics of flat output distributions and impedance transforming, combined with the trans-directional (TRD) couplers, stable circularly polarized excitation with 0°, 90°, 180°, and 270° sequential phase shift over the wide GNSS band are obtained, as well as integration with LNA. Besides, the dimension of the feed network is reduced. After using a simple rectangular patch as the radiator, an active integrated antenna is fabricated and measured. Over the entire GNSS band, the antenna exhibits good impedance matching, CP radiation, and high gain.

2. Antenna Structure and Theoretical Analysis

2.1. Structure of the Proposed Active Integrated GNSS Antenna

Figure 1 shows the geometry of the proposed active integrated GNSS antenna. It is composed of a rectangular radiation patch excited through the microstrip line-based coupled feeding, a rectangular ground plane, and an active integrated quadrature four-feed network. Four microstrip lines with the same size are etched on the top of the substrate I (F4B, εr = 3.5, tanδ = 0.003, h = 1.5 mm, ). On the bottom of substrate 1, a rectangular patch with the dimension of is printed. The ground plane is etched on the top surface of substrate II (F4B, εr = 3.5, tanδ = 0.003, h = 1.5 mm,  = 120 mm). Four holes are inserted on the ground plane to protect the feeding probes. On the bottom of the substrate II, the active integrated quadrature four-feed network is placed.

As shown in Figure 1, the active integrated quadrature four-feed network is composed of a wideband impedance-transforming rat-race coupler, two TRD couplers [25], and an integrated LNA. In recent years, developments in rat-race couplers have included harmonic suppression [26], dual-band [27, 28], wideband [29], and reconfigurable [30, 31]. However, less research focuses on improving the output port distributions and phase differences of a rat-race coupler. For example, in [32], a compact transmission line equivalent to the LC circuit is used for the miniaturization of a rat-race coupler, as well as obtaining harmonic suppression. The 20% fractional bandwidth and up to the 7th harmonic suppression is obtained. But the output port phase differences which are an important performance for applications are not provided. In the proposed design, three aspects of the traditional rat-race coupler are developed. The first one is to use a three-period T-type transmission line and a short-circuited coupled line loaded with an open-ended stub to achieve size reduction. The second is to use an LC-loaded stub to improve the output port phase flatness, which enhances the wideband applications of traditional rat-race coupler. The third one is to use different impedances in the three-period T-type transmission line to achieve impedance transformation.

Since the TRD coupler has the advantages of wideband, small size, and compact structure, and it is used as the quadrature coupler connected to the output ports of the rat-race coupler. Thus, a four-feed network with stable sequential phase shifts of 0°, 90°, 180°, and 270° is generated to excite the radiation patch. The input port of the rat-race coupler has the impedance transforming feature, which can be integrated with the LNA. Thus, one of the impedance matching circuits can be reduced in the LNA. Since the active integration is contributed by the impedance-transforming rat-race coupler, the detailed design process of the proposed rat-race coupler is introduced in detail.

2.2. Theory of the Active Integrated Four-Feed Network

To process navigation signals conveniently in subsequent systems, the receiving antenna usually needs to be cascaded with an LNA. In order to solve the problems of high loss, large size, and poor performance when using a direct connection, an impedance-transforming quadrature four-feed network is proposed. It consists of an impedance-transforming rat-race coupler and two quadrature couplers.

Figure 2(a) shows the transmission line-based configuration of the proposed rat-race coupler. It consists of three types of transmission lines (TLs), named as TL1 (Z01, θ01), TL2 (Z02, θ02), and TL3 (Z03, θ03). Since the circuit is symmetric along the MM′ line, even-odd mode is used for analysis.

To simplify the calculation and maintain the basic structure of a rat-race coupler, the electrical lengths of the three types of TLs are θ01 = 90°, θ02 = 270°, and θ03 = 90°. Then, the even- and odd-mode ABCD matrices [Mo] and [Me] can be expressed as follows:

For impedance transforming, the impedances of ports 1 and 3 are defined as RS, whereas the impedances of ports 2 and 4 are defined as RL. Then, the S-parameters of the rat-race coupler can be obtained according to the following:where

Here, and are the reflection and transmission coefficients. By substituting (1) into (2), the values of Z01, Z02, and Z03 can be calculated as follows:

It is known that the traditional transmission line-based rat-race coupler has the disadvantage of being large size. Besides, the poor phase flatness between the output ports limits the wideband applications of the traditional rat-race coupler. To reduce size and improve the output port phase flatness, the TL1 and TL3 are replaced by LC-loaded three-period T-type transmission line, as shown in Figure 3(a). While the TL2 is implemented by a short-circuited coupled line loaded with an open-ended stub, as shown in Figure 3(b).

In general, the LC equivalent circuit with transfer function analysis is applied for modeling passive devices [33, 34]. In this design, to omit the conversion from the LC circuit to the transmission line, the transmission line equivalent method is directly utilized in the analysis. In the following, the ABCD matrices are applied to perform the equivalization.

According to Figure 3(a), the ABCD matrix [MT{1,3}] of the TL1 and TL3 can be derived.whereand

According to Figure 3(b), by multiplying the matrix of two-end short-circuited coupled lines with that of the open-ended stub-loaded transmission line, the ABCD matrix [MT2] of TL2 is derived.where

Based on (5)–(9), the S-parameters of the proposed equivalent circuits can be obtained as follows:

Besides, the phase difference between TL2 and TL{1, 3} can also be expressed as follows:

Since the circuits in Figure 3 are equivalent to the TL{1, 3}, and TL2 of the rat-race coupler, the phase conditions in (12) should be satisfied, where f0 is the center frequency.

Besides, to obtain wideband operation and flat output port phase differences, the conditions in (13) are also necessary.

Here, fi is the sampling frequency within the assigned bandwidth, and N is the number of sampling points. Based on the relations in (12) and (13), the circuit parameters of the rat-race coupler can be obtained. However, it is hard to direct calculation. A better way is to use the optimization method [35] with specific tolerance limits.

3. Design and Implementation

3.1. Design of the Active Integrated Four-Feed Network

The diagram of the active integrated four-feed network is shown in Figure 4. The input port impedance of the rat-race coupler (RS) is assigned to be equal to the input impedance of the LNA (ZLNA), while the output port impedance of the rat-race coupler (RL) corresponds to the input impedance of the TRD coupler (ZTRD).

First, the LNA operating within 1.09–1.7 GHz is designed to obtain the input impedance ZLNA. Figure 5 shows the circuit schematic of the designed LNA. Here, the transistor ATF-54143 is utilized and modeled using the ADS software. After analysis, the DC feature of the transistor, the gate and drain voltages (VGS and VDS) are selected as 0.5 V and 3 V, respectively. Since the DC voltage is 5 V, the components in the bias circuit can be calculated. In the design, the bias circuit is composed of three resistors, where RB1 = 68 Ω, RB2 = 240 Ω, and RB3 = 27 Ω.

Then, the stability coefficient is simulated. In order to increase stability and broaden the operating bandwidth, the series connection of one capacitor (CF1 = 0.5 pF) and one resistance (RF1 = 820 Ω) is applied as the feedback circuit. To connect with the output port (50 Ω), the output matching network is constructed by one simple transmission line with a characteristic impedance ZM1 of 107.5 Ω, and an electrical length of θM1 of 37.9°. While no matching network is needed to be inserted due to the use of an impedance matching rat-race coupler. Here, the input port impedance of the LNA (ZLNA) is obtained as (29.2 + j0.02) Ω. Since the imaginary part of the ZLNA is small, the input port impedance of the rat-race coupler (RS) is directly assigned as 29.2 Ω. In other cases, if the imaginary part of the ZLNA can’t be ignored, an open-circuited stub can be utilized to remove the imaginary part. Figure 6 gives the results of the designed LNA simulated using ADS. It is observed that both of the input and output impedance matchings are larger than 10 dB from 1.09 GHz to 1.7 GHz. At this frequency band, the gains of the LNA are larger than 15 dB.

Second, a quadrature coupler is designed. Here, the TRD coupler in [25] is selected for the merits of wide bandwidth, flat output distributions, and small size. Figure 7 shows the schematic of the TRD coupler. According to [25], the parameters are obtained as follows: Ct1 = 1.2 pF, Ct2 = 2.2vpF, Zte = 126.7 Ω, Zto = 88.6 Ω, and θtc = 23.4°. Figure 8 shows the theoretical results of the designed TRD coupler according to the calculated circuit parameters. It is shown that the return loss is better than 10 dB in the frequency ranges 1.03 GHz to 1.7 GHz. The phase difference keeps 90° ± 2° from 1.11 GHz to 1.81 GHz, with an amplitude imbalance of less than 1 dB.

For connecting with the LNA and the TRD coupler designed above, the input and output port impedances (RS and RL) of the rat-race coupler are 29.2 Ω and 50 Ω, respectively. According to (4), the values of Z01, Z02, and Z03 are first calculated to be 41.3 Ω, 70.7 Ω, and 54.0 Ω, respectively. Second, the parameters of the proposed equivalent circuits of the rat-race coupler are calculated. As Section 2 introduces, optimization technology is preferred for fast calculations. Here, the particle swarm optimization [35] is applied, and the tolerance limits in this design are assigned as given in following equations:

After optimization, one group of circuit parameters is obtained, as shown in Table 1. Figure 9 shows the theory results of the designed rat-race coupler. It is observed that the return loss and isolation are more than 10 dB and 35 dB, respectively, ranging from 1.13 GHz to 1.73 GHz (42.9%) for both port 1 and port 3 excitation. From 0.6 GHz to 2.2 GHz, the amplitude imbalance is less than 0.5 dB, and the phase differences error is within 1° for both excitations. Especially, the amplitude imbalance keeps less than 0.2 dB from 0.6 GHz to 2 GHz.

After combining the designed LNA, the two TRD couplers, and the impedance-transforming rat-race coupler, the integrated active four-feed network is constructed, as shown in Figure 1. Since the even- and odd-mode characteristic impedance ratio in the rat-race coupler is large (tight coupling). A vertically installed planar (VIP) [36] structure is used. Since the space for the shunt LC stubs is limited, the LC-loaded stub with halved characteristic impedance, halved inductance value, and doubled capacitance value is served, as a replacement. A similar method is also applied at the points with four stubs. Figure 10 shows the detailed layout of the rat-race coupler. After simulation using HFSS, the final dimensions of the four-feed network are obtained, as shown in Table 2. Figure 11 shows the simulated results of a wideband four-feed network. It can be observed that the return loss is more than 10 dB from 1.13 GHz to 1.86 GHz (48.8%). In the GNSS band from 1.164 GHz to 1.610 GHz, the output port amplitude imbalances of the feed network are less than 1 dB, and the output port phase difference errors are within 5°.

3.2. Design and Implementation of the GNSS Antenna

Due to the prominent performance of the wideband integrated feed network, a simple rectangular patch is applied as a radiator. To broaden the operating bandwidth of the antenna, a layer of air with a thickness of h_air is added between the two substrates. Furthermore, to compensate for the high inductance induced by long metallic probes, the method of coupled feeding is applied by inserting four microstrip lines with a length of lou. It is noted that improvements in AR beamwidth and antimultipath performance were not considered in designing the CP antenna. Since the proposed four-feed network is universal, it can be applied to feeding the newly designed CP antenna with good radiation performances [3741].

Figure 12 shows the simulated gain and axial ratio (AR) of the designed passive antenna. It is seen that the AR is less than 3 dB from 0.84 GHz to 1.73 GHz. At the higher band in GNSS (1.55 GHz∼1.61 GHz), the gains of the passive antenna are in the range of 4.3 dBic∼6.1 GHz. While at the lower band in GNSS (1.164 GHz∼1.3 GHz), the gains are from 0.8 dBic to 5.6 dBic. The gain drop at the lower operation band is mainly due to the single-mode resonate of the CP antenna. To deal with it, the multimode resonated CP antenna can be used as a replacement. Besides, the dropped gains can also be compensated by the integrated LNA. Figure 13 shows the current distribution of the designed passive antenna at 1.561 GHz. It is seen that the surface current vectors at 0, T/4, T/2, and 3T/4 are directed in a counterclockwise direction, which indicates the radiation of right-hand circular polarization (RHCP).

Figure 14 shows the fabricated integrated wideband GNSS antenna with a whole size of 120 × 120 × 23 mm3. The S-parameters are measured with an Agilent N5230A network analyzer, and the far-field features are measured in an anechoic chamber. Figure 15 shows the simulated and measured results of the active integrated GNSS antenna, including the |S11|, AR, and gain. Under the criterion of |S11| < −10 dB, the measured bandwidth is in the range of 1.07 GHz to 1.73 GHz, yielding a fractional bandwidth (FBW) of 47.1%. As seen from Figure 15(b), over the whole GNSS band from 1.164 GHz to 1.610 GHz, the AR is less than 3 dB and the active gain is more than 16 dBic.

Figure 16 gives the simulated and measured normalized radiation patterns at 1.207 GHz, 1.228 GHz, 1.561 GHz, and 1.602 GHz, which respectively correspond to Galileo E5b, GPS L2, BDS B1, and GLONASS L1. It is seen that in the main direction, the polarization is RHCP with symmetric radiation patterns. Besides, for the RHCP radiation, the simulated and measured results in the broadside (−60°∼60°) agree well except for a tilt observed at 1.602 GHz. While in this angle range, the trends for LHCP radiation are similar between the simulated and measured results. The differences observed are mainly due to the fabrication error, the error in the substrate permittivity, and the error in the applied commercial components. To mitigate these differences, substrates with more accurate permittivity (for example, the Rogers substrate) and commercial components with small errors can be used as a replacement. The machine welding will induce less fabrication error than the manual welding.

Table 3 illustrates the comparisons between the proposed antenna and several GNSS antennas. It is seen that the proposed antenna exhibits a compact size compared with those in [10, 12]. Besides, the reported antennas in [10, 12, 13] are all passive antennas though they own the ability for GNSS operation. Antennas in [38, 39] show compact dimensions and high gain, but their overlapping working bandwidths between the IBW and ARBW are 14.7% (1.38 GHz∼1.60 GHz) and 5.7% (1.55 GHz∼1.64 GHz), respectively, which are narrower than 45% (1.07 GHz∼1.7 GHz) of the presented antenna. In other words, the antenna in [38, 39] can only operate at the higher band in GNSS, while the proposed antenna can operate at the whole GNSS bands. Besides, the gain of the antenna in [23] is around 15 dBic. While the minimum gain of the designed antenna is 16 dBic. The maximum gain reaches 21 dBic. Since the proposed design is an active antenna, more transistors can be cascaded to increase the gain of the LNA, and at the same time, higher active antenna gain can be achieved. Using the proposed antenna as a unit to design an antenna array can also increase the antenna gain. Better efficiency can be obtained by using techniques to suppress edge scattering. The loading of a backed cavity is an effective method. In summary, the proposed integrated antenna has the advantages of the wide operation band, high gain, and an integrated active structure, which can make it a good candidate for GNSS applications.

4. Conclusion

In the paper, the design of an active integrated wideband GNSS antenna is performed based on the impedance-transforming quadrature four-feed network. As an essential component in the four-feed network, the proposed rat-race coupler features the characteristics of flat output distributions and impedance transforming, which is beneficial for providing stable circularly polarized excitation in the GNSS band and realizing integration with the LNA. A prototype is designed by using a simple rectangular patch as the radiator. Measurement results show that the proposed antenna has wideband impedance matching, good RHCP radiation, and high gain over the whole GNSS band with a reduced size, which can make it an attractive candidate for GNSS applications.

Data Availability

The data used to support the findings of this study are included within the article.

Conflicts of Interest

The authors declare that they have no conflicts of interest.

Acknowledgments

This work was supported in part by the National Natural Science Foundation of China under Grant 51809030, in part by the Liaoning Revitalization Talents Program under Grant XLYC2007067, in part by the Young Elite Scientists Sponsorship Program by CAST under Grant 2022QNRC001 and in part by the Fundamental Research Funds for the Central Universities under Grant 3132023246.