A Disruptive Approach to mmWave for Wireless Telecom Applications
This article will discuss a disruptive antenna technology that is based on dielectric waveguides that travel along a medium, traditionally known as a polyrod. In contrast to widely used electronically steerable phased arrays, polyrods do not require per-element phase shifters, leading to a drastic reduction in cost, design complexity and power consumption. In fact, the polyrod, dielectric waveguide antenna (DWA) element allows for beam pointing rather than beam forming. This paper provides some historical context, introduces the technology and finally discusses different use cases and applications.
Traditionally, the advantages of a polyrod were light weight, low cost, reduced volume, ultra-wide bandwidth and stable radiation patterns. The ideal use cases were satellite communications, radar systems, chip-to-chip communications and mmWave imaging. ED2 re-invented the approach and realized its true potential for 5G and beyond.
In engineering and product design, the solution is always a consideration of many comparisons and trade-offs. For a system, DWA-based or phased arrays are no different. This article does not suggest that DWAs will replace phased arrays. The intention is to highlight that DWAs enable low cost, low power and provide ultra-wide bandwidth. DWAs can be an alternative to phased arrays for appropriate applications.
Bell Labs Polyrod Technology
The “polyrod antenna” was developed by Bell Labs from 1941 to 1944. A rod or wire of dielectric material is a well-known transmission line and shown in Figure 1. A portion of the energy travels on the outside of the rod. Irregularities in the surface of the rod cause a radiation of the signal. This “leaky” transmission line can be used as an antenna structure. ED2’s polyrod antenna has achieved approximately 20 dBi gain and supports either dual linear or circular polarizations.
Bell Labs described it this way: The tendency toward radiation inherent in the dielectric guide is turned to advantage, however, in a new form of radio antenna. Here the objective is to encourage radiation from all parts of the dielectric rod. In progressing along the rod, therefore, power is gradually transferred from within the dielectric to the space outside. At a point where the transfer has been effectively completed, the rod can be terminated abruptly. By proper design, this radiating structure is an end fire antenna. Since it has been most often fabricated from polystyrene, it has become known as the polyrod antenna. It is especially useful for microwaves.1
New Challenger for mmWave Phased Arrays: DWA
In contrast to electronically steerable antenna arrays that are commonly used in 5G base stations and UEs, ED2’s patented DWA elements enable new equipment that will beam point. In its basic form, the DWA is a cylindrically or cone-shaped dielectric rod that acts as a waveguide, radiating a beam whose width and gain is proportional to the RF frequency and rod length.
As stated before, DWAs offer a few advantages over electronically steerable arrays, including lower cost, lower power consumption and more consistent beam characteristics. In terms of beam characteristics, DWAs produce symmetric antenna patterns, back lobe suppression and zero scan loss.
ED2 has patented linearly and circularly-polarized antenna elements, commercializing these products through partnerships. The DWA can easily be modified to support any mmWave frequency used by telecom. At 28 GHz center frequency, DWA designs that achieve 20- to 90-degree beamwidth (depending on polyrod length), with 20 dBi gain along the antenna boresight. These antennas also benefit from being dual polarized. DWA can achieve uniform signal coverage over a targeted area making it a great fit for mobility applications.
In 2018, it became clear that a DWA alternative to phased arrays would provide benefits to the 5G ecosystem. This alternative antenna should achieve a few specific requirements or goals. The new DWA requires a feed that will induce two linearly polarized signals into the rod that are orthogonal to each other. This new DWA will act as an end fire antenna when excited by a conductive resonator and generate a highly directive narrow beam with no phase shifters needed.
First Design Attempt
ED2 took a fresh approach to phased array design and concluded that the trade between circular and linear polarization was a toss-up. For the first concept, the team decided to design a circularly-polarized antenna element.
The circularly polarized antenna element would transition to a waveguide in organic substrates. The team leveraged experience with passive surface-mount waveguide components to make the waveguide launch successful. Next, they needed to integrate the waveguide launch with a rod using an organic material. This led to an intensive CNC and lathing process that proved to be expensive and labor intensive. By the end of this first version’s development, they had not only created 3D spiral exciters with a rod that can change in gain by adjusting its length, but also achieved the predicted performance. The exercise led to what is used today and taught significant lessons to the team.
Although the polyrod concept has been around for some time, its use in 5G communications had not been explored before. The gain and beam characteristics of a polyrod antenna depend on its geometry and the operating frequency. Specifically, for a cylindrically shaped polyrod of electrical length Lλ (where λ is the wavelength), the half-power beamwidth (HPBW) is . As an example, producing a 20-degree HPBW requires a polyrod of length Lλ = 9λ; this is slightly more than 9 cm at 28 GHz frequency (see Figure 2). Antenna length and geometric properties can be adjusted to achieve any desired beamwidth at any target mmWave frequency.
Expensive and Labor Intensive
The team learned that the initial design was not cost-effective to manufacture for a number of reasons. The material in the stack-up had an overall thickness of approximately 280 mil, and most board shops cannot handle stack-ups that large. Manufacturing required a shop that could handle an overly thick board and build per our manufacturing drawings. Only a single supplier was available, at a high cost. The cost of the PCB rod represented a majority of the overall cost of the antenna, regardless of manufacturing quantities. The new design incorporates a significantly thinner stack-up, expanding the number of outsourced PCB shops that could support a lower cost antenna (see Figure 3).
Besides high costs, the original design presented manufacturing challenges. The material was brittle, thin and long, making post-processing difficult; and, it was prone to failure from vibration. Concentricity was difficult to repeat. A 10 percent failure rate during machining was experienced. Tight design tolerances and machining difficulties made it unfeasible to build in a timely manner and suggested that fixtures and new machines were needed. With the high labor requirements, machining this part became 20 percent of the overall cost. And once built, the polyrod cap and connector required further assembly, adding additional fabrication time and cost.
The team decided to redesign to reduce labor, complexity and cost. The antenna needed a symmetrical directional pattern to meet requirements. Based on our understanding of market direction, the team wanted to convert the design from a circular polarization to dual linear polarization, both horizontal and vertical. The gain of the first design fell short at 16 dBi, when 20 dBi gain was required.
The new design was reconceptualized as a much smaller organic board (see Figure 4) that receives two orthogonal linearly polarized signals and converts to a dual polarized wave that is launched into a dielectric rod (see Figure 5). The return loss performance is shown in Figure 6. The results were:
• Low Cost - The team went to a multi-layered organic board and dielectric rod, which led to lower cost. The antennas were designed for low-cost connectors.
• Easier to Manufacture - The new design requires assembly of only three components and the team created a low-cost test procedure.
The lower taper angle, the lower base diameter and the upper rod diameter were all adjusted for optimum return loss, gain and minimum side lobes. The total rod length was adjusted for the necessary antenna beam width, 22.5 degrees for the FreeStar and optimized for maximum gain for the repeater (see Figure 8). A dielectric rod length of 5.6 in. achieved a gain of 19.5 dBi. Figure 9 shows the improvements in gain and reduction of the side lobes.
A Comparison: Beam Pointing and Phased Arrays
mmWave communications are limited by the high attenuation of the signal. Additionally, high frequency does not propagate well through materials, so the blockages and blind spots are numerous. The best results are achieved with line-of-sight between transmitter and UE. The effect of rain attenuation adds another dimension of complexity to network planning with mmWave. Finally, significant doppler shift appears at high UE speeds (coherence time ∝ λ2).
Despite the challenges, current and projected spectrum requirements make it difficult to visualize a future without mmWave; the amount of spectrum required to support projected wireless data consumption points to mmWave. A key element to making mmWave a success is the antenna. Unlike applications below 6 GHz, mmWave antennas must be integral to the electronics due to inherent cable losses that greatly degrade system performance so the type of antenna is key to a mmWave system.
There are key features that an ideal 5G mmWave antenna should have. First, the mmWave antenna should lead to a low cost, with minimal complexity and small form factor, especially for CPE devices. The mmWave antenna should be low power and dissipate as little heat as possible. The beam activation and switching should be in the hundreds of nanoseconds. The antenna should be dual polarized with a consistent beam gain and beam width, irrespective of UE location. The mmWave antenna should have an adaptive beam width depending on distance and user distribution. It should also support SU-MIMO and MU-MIMO operation. Lastly, the mmWave antenna should support multi-lobe beamforming to reduce impact of blockage.
Most designs for mmWave 5G rely on electronically steerable planar arrays. The planar array is composed of many antenna elements and each element requires phase shifters and power amplifiers (PAs) among other circuits. Telecom solutions typically require multiple panels, each partitioned into several subarrays.
Planar arrays have some drawbacks. First, planar arrays are made up of (up to) hundreds of antenna elements to produce narrow beams. Each panel requires tens to hundreds of antenna elements, phase shifters and PAs along with the electronics to power and control them. Panels are densely packed, making mechanical alignment, thermal management and control electronics a challenge to manage. That leads to high-power consumption and expense. Coverage can be inconsistent because beam widths and gain vary with the relative orientation of the UE to antenna. The antenna backlobe gain is almost as high as the main lobe gain, as shown in Figure 10. Phased arrays are complex with lots of circuit content needed per element, while the DWA requires a basic amp with the antenna.
Finally, a planar phased array has limited support for single-user-MIMO and multi-user-MIMO functionality. To simplify design, each RF chain (MIMO stream) is typically connected to a specific subarray. This leads to limited flexibility in adapting MIMO streams based on spatial DOFs. The architecture leads to limited multi-lobe capability (one beam per RF chain). For MU-MIMO, more active UEs lead to a smaller subarray per UE and wider beam widths and lower antenna gain.
Empowering 5G Solutions Antennas
This DWA is a standalone, connectorized solution that can be used in any number of products. ED2 is partnering with world-class antenna manufacturers to guarantee performance and leverage cost efficiencies. Table 1 shows key performance parameters for DWA with a comparable mmWave phased array reference design offered by a leading antenna manufacturer.
In 2021, ED2 licensed its technology to Wilson Electronics with the intention of addressing the cellular repeater market. Together, the companies are producing a portfolio of repeater products that utilize ED2’s DWA technology in the FR2 space.
Advanced Antenna Systems (Freestar 5G)
In 2020, ED2 designed and produced an Advanced Antenna System for Freefall 5G Inc., with the intent to address a variety of markets that take advantage of mmWave capability, like defense, radar, telecom and point-to-point or point-to-multi-point broadband.
The Freestar 5G antenna utilizes 16 individual cards, each with four DWAs, arranged circularly. The array is driven by a switch-matrix capable of connecting to four independent radio elements, any of which can be output to whatever desired antenna coverage patterns are needed for the application. Figure 11 shows a cutaway view of the Freestar 5G antenna.
The DWA offers significant advantages over state-of-the-art planar arrays. One DWA replaces a whole array/subarray, comprised of dozens of electronically steerable patch antennas, resulting in significant reductions in cost, power consumption (no phase shifters needed) and Si area. Furthermore, for use cases that involve servicing multiple users along different directions (e.g., street intersections), the incoming narrow-beam 5G signal from the gNB can be simultaneously fanned out by the repeater(s) through multiple DWAs. This switched-beam architecture, shown in Figure 3, does not require phase shifters. However, it requires a “switching matrix” that can turn on and off any subset of multiple beams at nanosecond speeds, and dynamically activate them in transmit or receive modes.
The team has been testing with Wilson Electronics and Mobile Network Operators beginning in 2021, with extremely successful outcomes (see Figure 12). Most of the detailed results are confidential and cannot be shared publicly, but in field trials the repeater product has been proven to extend the range of a gNb and bend coverage into areas shadowed from RF signals.
A series of desert range tests were performed in Tucson, Ariz. A FreeStar stack module, which uses the DWA elements, was used as the transmitting device. A low QAM OFDM signal was transmitted at the start point and an ED2 repeater device was used as the receiver. Several way points were taken along the path to verify signal integrity. A range of 2.91 miles (4.68 km) was achieved before the team reached the limits of the testing ground. The test team verified that no multipath issues were present on the nearly 3-mile stretch of desert dirt road while maintaining a 15 to 20 dB SNR with a QPSK signal.
The DWA antenna approach can work across the entire mmWave spectrum. As of this writing, the 60 GHz spectrum is receiving a lot of attention, in part because of Meta. The DWA is a good solution for 60 GHz band as the interface would be a direct waveguide to a much smaller rod. Another potential application is the use of multi-frequency band DWA’s pointing at a particular fixed point. These and other ideas are part of the company’s research and development efforts.
Electronically steerable antenna arrays have been a popular choice for analog beamforming in 5G systems, due to the ease of integrating them with RFICs, the market advantages to Si providers who support the architecture and earlier deployments of phased arrays at lower frequencies. They are particularly suited for very narrow beamwidth applications under high mobility. DWA technology presents an alternative approach to phased arrays for use cases involving moderately narrow beamwidths (e.g., 20-degree nominal HPBW) and pedestrian speeds or fixed wireless applications. For these types of cases, phased arrays’ flexibility in beam steering becomes overkill, especially when considering the associated cost, complexity (tens to hundreds of antennas and phase shifters), high-power consumption, antenna pattern/gain inconsistency. The cost, performance and complexity advantages of DWA will enable them to play a role in delivering on the promises of 5G.
- G. Mueller and W. Tyrrell, “Polyrod Antennas,” Bell Labs, October 1947, Web: www.bell-labs.com/institute/publications/bstj26-4-837/#gref.
- “Samsung Announces 5G Data Breakthrough,” May 2013, Web: https://phys.org/news/2013-05-samsung-5g-breakthrough.html.