Approaching the 5G mmWave Filter Challenge

In the world of LTE, developers are very familiar with the available filtering technologies that work, namely surface acoustic wave (SAW) and bulk acoustic wave (BAW) filters. These filters cover a range of frequencies up to 6 GHz, come in small sizes and offer good performance-to-cost tradeoffs, making them the dominant off-chip approaches in mobile devices today. Unfortunately, analogous filtering options for the mmWave spectrum have issues regarding viability, performance, size and availability, while research teams helping to write 5G standards have yet to provide information on what filters will be required, where they need to be placed in the base station and what performance metrics they must meet.

With the promised delivery date of 5G wireless communication fast approaching, the world is waiting to see if this next-generation network will hit its ambitious goals of 10 Gbps peak data rates, less than 1 ms latency, ten times greater energy efficiency, and more. In past decades, each generation of mobile systems – from 1G analogue systems to 2G digital standards to 3G mobile broadband capabilities to 4G LTE and LTE-Advanced networks – has overcome a unique set of challenges. Leaps in technology are necessary to enable these advancements in performance.

To tackle the first challenge of increased data rates, we can return to the Shannon-Hartley Theorem that explains the maximum amount of error-free digital data that can be transmitted over a channel of a given bandwidth in the presence of noise:

where C is channel capacity, M is the number of channels (e.g., the MIMO order), B is the bandwidth, and S/N is the signal-to-noise ratio. To increase channel capacity (or data rates), one of the primary factors of focus for 5G is bandwidth, as well as how much bandwidth is available in a given part of the spectrum.

Nowadays contiguous bandwidth of 100 MHz is hard to find in the LTE spectrum landscape. Given that the 5G peak data rate goal is 10 Gbps, the increase in channel capacity must come from somewhere. A key innovation at the heart of 5G is utilizing new spectrums where increases in bandwidth are easier to obtain. Therefore, the millimetre wave (mmWave) spectrum – which technically corresponds to Extremely High Frequencies (EHF) from 30 GHz to 300 GHz but generally also includes 20 GHz and higher centimetre-scale wavelengths – is of great interest to designers because it offers the most dramatic increase in available bandwidths compared to available spectra below 3 GHz.

As we enter the unfamiliar territory of the mmWave spectrum, one of the main questions that arise is filtering. In the LTE generation, developers are very familiar with the available filtering technologies that work, namely surface acoustic wave (SAW) and bulk acoustic wave (BAW) filters. These acoustic filters cover a range of frequencies up to 6 GHz, come in small sizes, and offer an excellent performance-to-cost tradeoff that makes them the dominant off-chip approach in mobile devices today.

Unfortunately, analogous filtering options for the mmWave spectrum seem to have major issues in viability, performance, size, availability, and so forth. Even the research teams helping to write the 5G standards have yet to provide information on what filters will be required, where they need to be placed in the base station, and what performance metrics they need to meet.

Considerations

The obstacles to and advantages of using the mmWave spectrum are both widely-publicized and well-understood. High frequencies suffer from range limitations and path loss through the air, objects, and buildings. However, mmWave signals require much smaller antennas, which can be tightly packed together to create single, narrowly focused beams for point-to-point communication with greater reach. Frequency bands around 28, 38 and 72 GHz are the main candidates for 5G mmWave, having demonstrated directional antenna, beamforming and beam tracking performance in multipath environments.1-2

The focus has now shifted to solving the practical issue of how to build an actual mmWave-capable base station and implement high-performance RF filtering. Fortunately, mmWave technology has been employed for decades in various fields and functions. For example, it has a long history in military, aerospace and SATCOM applications such as K-Band inter-satellite communication and ranging and Ka-Band high-resolution radar.

In other industries, the automotive field is using both 24 GHz in short-range and 77 GHz in long-range radar-based advanced driver assistance systems (ADAS) to scan a vehicle’s environment for driver support or automated driving. In the U.S., 36 to 40 GHz is currently licensed for high-speed microwave data links between a cellular base station and a base station controller; and, the unlicensed 60 GHz band is used in short-range data connections and IEEE 802.11ad WiGig for high bandwidth streaming.

• Size and Packaging: In phased arrays, the antenna elements must be sufficiently close together to avoid generating grating lobes; a half-wavelength spacing for mmWave frequencies amounts to only a few millimetres. As shown in Figure 1, phased arrays often used in mmWave applications use a plank architecture, in which the gold areas indicate the antennas mounted on a printed circuit board (PCB) (green area) and the blue areas indicate the circuit “planks” extending 90 degrees from array (where space for filters is already minimal). Base station manufacturers, however, are now looking into flat-panel architectures where the circuitry is implemented on the backside of the antenna PCB, requiring even denser spacing for filtering and other functional blocks.
• Temperature Stability: To make efficient use of available bandwidth, the filter must meet its specifications over a range of temperatures. Small-scale systems may be deployed in exposed environments that experience extremes in temperature and temperature variation. Further, overall size reduction in systems leads to densely populated boards, where heat from surrounding components can affect filter stability.
• Power Handling: The ability of the filter to withstand large amounts of transmitting RF power is mostly a concern in traditional macro-cell use cases below 3 GHz. In mmWave 5G use cases, the transmit power is spread over the individual array elements, and the transmit range is reduced as well.
• Filter Placement: As shown in Figure 2, there are several practical locations: at the antenna element (Position 1), behind the amplifiers closest to the antenna (Position 2) and on the high-frequency side of the mixers (Position 3). It should be noted that the dashed lines represent potential 1: N branching in a beamforming architecture, and Figure 2 shows a simplified single path from the mixer to an antenna. The advantage of the Position 1 implementation is that linearity and bandwidth design constraints for the subsequent blocks are eased, potentially reducing overall system cost. Filtering placed at Position 1 suppresses noise far from the channel of interest across a wide frequency range, making wideband suppression and shallow insertion loss key performance features. Given that each sub-array has its filter, small size and low cost are essential.

Solutions

SAW and BAW filters have long dominated the off-chip filtering market in mobile devices because of their excellent performance specs, tiny footprints, and affordable prices compared to other options. Unfortunately, the nature of their design – using interdigital transducers (IDT) to process signals as acoustic waves – exhibits degradation in selectivity at frequencies greater than 6 GHz, making SAW and BAW technology unfeasible for mmWave applications. Nevertheless, it is worth noting their performance metrics (see Table 1) as a benchmark for potential mmWave filter solutions.

Emerging SAW/BAW Alternatives

Given the successful use of SAW and BAW technologies at lower frequencies, researchers are naturally looking into developing an acoustic equivalent for mmWave applications.

The film bulk acoustic resonator (FBAR) filter is a form of BAW filter that can reportedly operate from 5 to 20 GHz, three which is applicable for LTE but still below desired mmWave ranges. Resonant Inc. is developing a so-called XBAR™ technology that seeks to outperform FBARs, but currently, it is only available as licensable intellectual property for manufacturers to complete development.

The use of substrate integrated waveguides (SIW) offers another approach where researchers seek to create small waveguide cavity filters for use in PCB and on-chip applications. While attractive for its wideband properties, good isolation, and lower losses, challenges for mmWave include radiation leakage between plated through-holes (PTH), difficulties designing SIW transitions and dimensional variations in the PTH sidewalls (which are detectable at mmWave frequencies).4

Micro-Electro-Mechanical Systems (MEMS) offers the potential for tiny, tunable RF filters created using conventional semiconductor fabrication processes. The methodology is interesting due to its dimensional accuracy, high component density and low-cost at high volumes;5 however, current designs are still in the research phase or are limited to the sub-mmWave ranges.

Proven Approaches

Figure 3 summarizes the frequency ranges of today’s commonly available filters. At higher mmWave frequencies, acoustic filtering is not as practical, so many developers have turned to electromagnetic (EM) solutions. For mmWave applications at 20 GHz and higher there are dielectric and cavity waveguide, on-chip, and microstrip (or planar thin film) filters.

Figure 3 Frequency ranges of standard filter technologies.

Waveguide filters are hollow or dielectric-filled cavities within metal tubes that act as BPFs, blocking certain wavelengths while allowing others to pass. Characterized by high-power handling and low loss, waveguide filters are widely used for mmWave frequencies from 20 to 80 GHz in military, radar, satellite and broadcasting markets. Unfortunately, waveguides typically have dimensions in the centimetre range (see Figure 4). A λ/2 array element spacing at 28 GHz in free space is 5.35 mm. Until manufacturers can sufficiently reduce waveguide sizes while still meeting electrical performance requirements, this solution may not be practical for an array antenna system.

Challenges arise from various factors, including the physical characteristics of the semiconductor material and the cost of implementation. For example, GaN circuits are made as thin as possible to increase heat dissipation; however, filter Q is proportional to dielectric substrate thickness, so the high-power advantage of using GaN devices works in opposition to integrating filters with high Q. Also, a filter structure in GaN occupies valuable real estate on the wafer that would be better allocated to active devices. Building on-chip high Q filter structures to serve in a front-end application is currently impractical.

Microstrip filters have been considered for mmWave applications but are commonly dismissed for various performance issues. Note that there are at least three different form factors:

• Microstrip on PCB
• Microstrip is a multilayer; low temperature co-fired ceramic (LTCC) package
• Microstrip is a small form factor, single-layer package

Microstrip filters printed on PCB are appealing because of their simple construction. Still, high-performance PCB solutions generally reach centimetre-range sizes, which are much larger than the sub-wavelength dimensions required for mmWave antennas. Variations caused by the PCB manufacturing processes also limit performance, resulting in higher insertion loss and lower suppression values.

Another option is surface mount technology (SMT). SMT assembly has been long used in commercial systems and is now being adopted in mmWave military systems for potential cost savings. Unlike chip and wire solutions, SMT filters have standardized form factors to reduce overall assembly time, and require no post-tuning. Figure 5 shows the performance repeatability of 100 SMT BPFs, measured in the 26.5 to 29.5 GHz range without tuning.

Figure 5 Repeatability of Knowles Precision Devices 28 GHz SMT single-layer microstrip filters, using 100 samples without tuning.

LTCC filters come in SMT form factor and are similar to multilayer capacitors, in which multiple layers of very thin ceramic tape are printed with different passive elements and then stacked together to prevent substrate warping. Prototypes of LTCC technology for mmWave applications are being developed, potentially offering a method of including both filters and antennas on the same component with a very small footprint. Unfortunately, since the metal coatings are screen printed, dimensional precision is not as high as other thin film solutions and the unpolished substrate can lead to high losses.6 Plus, suppression is generally limited to 30 dB and below.

Source: Microwave Journal article, Knowles Precision Devices

References

1. T. S. Rappaport, “Millimeter Wave Cellular A Road to 5G,” IEEE International Conference on Communications,” June 2013, https://wireless.engineering.nyu.edu/presentations/icc2013.pdf.
2. K. Sakaguchi, T. Haustein, S. Barbarossa, E. C. Strinati, A. Clemente, G. Destino, A. Pärssinen, I. Kim, H. Chung, J. Kim, W. Keusgen, R. J. Weiler, K. Takinami, E. Ceci, A. Sadri, L. Xain, A. Maltsev, G. K. Tran, H. Ogawa, K. Mahler and R. W. Heath Jr., “Where, When, and How mmWave is Used in 5G and Beyond,” IEICE Transactions on Electronics, Vol. E100-C, No. 10, October 2017.
3. “5G NR Primer for Amplifier and Filter Design,” National Instruments, September 2018, www.awrcorp.com/news/article/5g-new-radio-primer-amplifier-and-filter-design-now-available.
4. J. Conrood, “Substrate Integrated Waveguides (SIW) Limits and Capabilities,” Rogers Corp., January 2018, www.youtube.com/watch?v=I9A66sUqaTk.
5. T. Pensala, “MEMS & Thin Films Manufacturing Platform for Mm-wave and 5G RF Components,” Eemeli Workshop, May 2015, 
6. N. Friedrich, “The Do’s And Don’ts Of LTCC,” Microwaves & RF, September 2013, www.mwrf.com/passive-components/do-s-and-don-ts-ltcc.