Projects:2021s2-63132 Multi-port Multi-mode antennas for sub-6GHz 5G MIMO applications
Abstract here
Contents
Introduction
In the project, a wideband dual-port dual-mode 3D printed cylindrical dielectric resonator antenna is designed and fabricated. The antenna has a broadside pattern and an omnidirectional pattern that work at the same frequency band, which has 26.73% overlapping bandwidth. Proposed designs are suitable for sub-6 5G spectrum. All proposed designs have low reflections and high isolations between each port at the operating frequency band, which can work with multi-input and multi-output technology to provide fast and stable 5G data transmission.
Project team
Project students
- Tianchang (Vincent) Ma
- Shenhua Zhou
Supervisors
- Dr Nghia Nguyen-Trong
- Prof Christophe Fumeaux
Advisors
Objectives
Two ports excite two modes at the same frequency band
Work for sub-6 5G spectrum
Low reflection for each port < -10 dB
High isolation between each port < -15 dB
Small size (easy to integrate for the small facilities)
Dual-port: wideband 3D printed design
Dual-port: narrowband planar design
Background
The new generation of telecommunication 5G allows to achieve higher data rate than Long Term Evolution (4G). The antenna plays an important role to support 5G, which is required to improve performance to match with higher data rate and stability. One solution is using multi-port multi-mode antenna with multiple input multiple output (MIMO) technology. Thus, the project will focus on multi-port multi-mode antenna design, which only have one physical structure (volume is same as traditional antenna), but the electromagnetic performance is same as combination of multiple antennas. The design should have low reflection, high isolation and wide bandwidth features for sub-6 5G frequency. When the design passes the simulation, which will be tested by vector network analyser (VNA) and anechoic chamber.
Method
The wideband 3D printed dual-port dual-mode cylindrical dielectric resonator antenna is designed by pattern diversity.The proposed designs use slot feed and probe feed. The slot feed (feed by port 1) can excite the HEM11δ mode for the dielectric resonator (higher mode HEM12δ is involved in the high frequency band). The probe feed (port 2) excites the TM01δ mode (TM02δ mode is excited in the high frequency band). HEM11δ and HEM12δ¬ modes refer to broadside patterns. TM01δ and TM02δ modes refer to omnidirectional patterns. Although 4 modes are involved in the working frequency band, based on types of radiation patterns, the design still can be called dual-port dual-mode CDRA. Different modes work in different frequency bands.
Different modes have different radiation patterns, E field’s distribution, and working frequencies. HEM11δ mode refers to the broadside radiation pattern. TM01δ mode refers to the monopole radiation pattern. The E field of the HEM11δ mode is mainly located near the edge of the cylindrical dielectric resonator. In contrast, the E field of the TM01δ mode is mainly concentrated at the center of the cylindrical dielectric resonator. Different E fields’ distributions can guarantee high isolations. Formulas to calculate HEM11δ and TM01δ mode resonant frequencies are shown in equations (1) and (2) [1]. In equations (1) and (2), parameter a refers to the radius of the CDR. The parameter h refers to the height of the CDR. ε_r represents the dielectric constant of the CDR. C is the speed of light. Based on Equations (1) and (2), under the same dielectric constant value and physical dimensions, the k_0 a (resonant frequency) of TM01δ mode will be always higher than HEM11δ mode [1]. ((k_0 a))_TM01δ=√((3.83)^2+((πa/2h))^2 )/√(ε_r+2) (where 0.33≤a/h≤5) (1) (k_0 a)_HEM11δ=6.324/√(ε_r+2) (0.27+0.36(a/2h)+0.02(a/2h)^2 )(where 0.4≤a/h≤6) (2) f= (k_0 a*c)/2πa (3)
The dielectric resonator with multiple dielectric constants can be fabricated by 3D printing with ABS1500 filament. By varying the 3D printing infill percentage, the dielectric constant can be changed following the Bruggeman model. The dielectric resonator’s constant for the inner core, middle layer, and outer layer is 12.3, 10, and 6.85 referring to 100%, 87, and 67% nominal infill percentages. A higher dielectric constant at center can reduce the TM01 mode resonant frequency. The middle layer can lower peak reflections in the working frequency band, ensuring the ability against tolerance.
After finishing the preparation work in section three, the antenna front layer is shown on the left. As it shown, the antenna is made up of five patches in a square shape. Two ports are distributed on the square patch (shown in orange) in the middle. The edge length of the centre patch is w1. The distance between the port and the patch center is d1. The distance between the middle patch and the surrounding patch (shown in yellow) is gap1. The four patches surrounding the central patch are rectangular patches with a shape similar to a square. Their length and width are w1 and b1 respectively, where the length of the patch is set to match the side length of the central patch. The distance between the side patch and the edge of the antenna is gap2. The length of the whole antenna is w.
The antenna back layer is shown on the right. From the bottom, there is nothing special about the antenna design. Except for a place for the mounting port, the rest is made of copper.
Figure shows the S-Parameter of S11, S22 and S21. Since the antenna is symmetric, S11 and S22 overlapped. The bandwidth of the antenna is 2∗(2.51−2.37)/(2.51+2.37) = 5.74%. In such bandwidth range, the isolation value stays below -15dB.
Results
Dual-port Dual-mode wide band 3D printed cylindrical dielectric resonator antenna By measurements, in the frequency band from 2.82 to 3.69 GHz, two ports reflections are lower than -10 dB with about -30 dB isolations. 26.73% consistent with simulations.
Since the fabrication issues, 3D printing is not high precision and an air gap existed between DR and the substrate. when testing the gain and radiation patterns, the prototype working is not stable. Especially for port 1. The maximum gain for the broadside pattern excited by port 1 is much lower than the simulation value in the most working frequency bands. The gain for the omnidirectional pattern in -40 and 40 deg is basically consistent with simulation values, it has about 2dB variation.
Through checking the radiation pattern excited by port 1, it can see that patterns are still broadside patterns. The shape is consistent with simulation. But have lower gain and higher cross-polarization. For the Y-Z plane, the pattern is not systematic. Meanwhile, at 3.69 GHz, the main beam does not gather well since the higher mode is involved.
The co polarization of Omnidirectional radiation patterns excited by port 2 is basically consistent with simulations. However, the measured cross-polarization is higher than the simulated data.
In the future, after solving existed fabrication issues, one more port can be added in the orthogonal direction to build a three-port three-mode design through polarization diversity. The overlapping feed line can be solved using an air bridge design with insulator tape.
Conclusion
For this project, there are two designs.
For the dual-port dual-mode 3D printed wideband CDRA, simulation results show the overlapping bandwidth is about 27% (from 2.82 to 3.7 GHz), which fully spans Australia deployed Sub 6 5G frequency band. In the working frequency band, the reflection for each port is lower than -10 dB. Meanwhile, isolations between ports are lower than –15 dB, capable with MIMO application. The dimension of the proposed designs is 120*120*11.45 mm, which is relatively easy to be integrated. By simulations, the gain of the broadside pattern is about 1.5 dB at the high frequency band since the higher mode is involved. The gain is about 3.8 to 4.8 dB for the lower frequency band. The maximum gain of the omnidirectional pattern is about 2 dB at -40 and 40 deg. By measurements, the prototype can achieve 26.73% relative bandwidth. However, since the fabrication issue, measured reflections are not stable. Not stable reflections lead to a low realized gain of the prototype.
For second design, a square patch antenna with two probe feeds works at 2.45GHz has been designed. This thesis states two different methods to increase the BW, and both methods can achieve the expansion of bandwidth. The material for substrate has changed from expensive one Duroid 5880 to less expensive one DiClad 880 which has similar relative permittivity as Duriod 5880. There are also some problems in the antenna design process. For example, the value of gain becomes smaller in the effective operating frequency of the antenna, but fortunately, a solution was found: the problem of realized gain that will suddenly dwindle can be solved by using a thicker substrate. The final antenna model is excited by two ports and operate at TM01 and TM10 modes respectively with a resonance at 2.45GHz. The simulated outcomes and measured outcomes are quite similar. At 2.45 GHz, the constructed prototype with an overall size of 137mm * 137mm * 3.18mm has a 10 dB return loss with a 6.1% bandwidth and a gain of 7.5 dB in the 2.45GHz band [10], whereas the simulation result for the gain of antenna model is close to 10dB. Furthermore, both ports can achieve an isolation which is greater than -15dB [10]. Based on all properties and evidence from measurement and simulation, the proposed antenna meets all requirements of this final year project. Meanwhile, when checking the progress of the project in section 3.1, the planning made in last semester was quite consistent with the completion of this semester. It can be said that the project is progressing smoothly without any trouble.
References
[1] a, b, c, "Simple page", In Proceedings of the Conference of Simpleness, 2010.
[2] ...