What you’ll learn:
- The basics of a modeling-based approach to RF-filter design.
- Why today’s IPD/LTCC filters are inadequate for 5G mobile devices.
- Where RF-filter design is heading and why.
Filters are a critical component—perhaps the critical component—of RF and microwave design. Filtering is what enables proper operation of everything from our mobile devices to smart homes/buildings/cities to autonomous and electric vehicles. Put simply, filters let desired frequency elements in and keep undesired elements out of a given device or system. Without filtering to isolate sections of our crowded RF spectrum from one another, interference from, say, the 5G bands to the closely adjacent Wi-Fi spectrum, would wreak havoc.
As our mobile devices take on more functionality and comprise multiple radios, the problem becomes insidiously complex. Consider the 1G phones of yore, which didn’t even yet have texting capability: These devices required only rudimentary filtering schemes that could be handled by just one or two filters. A 4G/LTE handset for worldwide use contains from 50 to 90 filters.
In contrast, today’s emerging 5G phones, which allow you to download a two-hour movie in 30 seconds, can have more than 100 distinct filters (Fig. 1). The more antennas and frequency bands a device covers, the more filters it will require to make everything play nice together.
In such a complex environment, there’s necessarily been some advances in how RF filters are designed and implemented. In a design and engineering environment that has seen a great shift toward approaches based on intellectual property (IP), it was inevitable that this approach would make its way into the RF-filtering realm.
This is where a company like Resonantenters the picture. Resonant is wholly concerned with RF filtering, but it doesn’t produce parts or hardware directly. Rather, the company’s foundation is a software platform called Infinite Synthesized Networks (ISN), which uses modeling to rapidly design and simulate filters while also targeting a given foundry’s capabilities. The goal is to design the optimal filter for the task at hand while also ensuring that it can be fabricated with reduced prototyping cycles, making the process more cost-efficient.
“What we bring to the table is our ISN software, which performs full finite-element modeling,” says Mike Eddy, senior marketing advisor at Resonant. Armed with the material parameters and physical dimensions for a given filter, the software predicts with high accuracy the filter’s performance as it would come out of the fab.
Large, vertically integrated players in the RF-filtering market typically approach filter design in an empirical fashion. They create a filter design and prototype it in their fab, evaluate its performance with respect to the performance requirements, and, if necessary, tweak the design and repeat the process. This can result in a lengthy procedure to build a device that meets the specs. “Instead of 12, 15, or 18 iterations, we try to have a spec-compliant part within two or three iterations,” says Eddy.
Initially, Resonant developed its ISN software for surface-acoustic-wave (SAW) filters in a process that would minimize variations in frequency response over temperature. Subsequently, the company sought to extend the platform for design of bulk-acoustic-wave (BAW) filters. “But as we looked at doing that, we were also seeing 5G coming along,” says Eddy.
5G filter requirements are very difficult from those for 4G, which needed filters with about 60 to 70 MHz in bandwidth at about 2 GHz. In contrast, 5G filtering calls for 600 to 900 MHz of bandwidth at 3 to 5 GHz for operation in the sub-6-GHz bands.
Resonant’s next move was to use its own software platform to develop a next-generation resonator as a building block for filters in 5G devices. That resonator, called XBAR, meets not only the requirements of 5G, but also those of Wi-Fi 6/6E.
XBAR resonators generate a BAW, but because of the nature of their structure, they can be fabricated on a simple process like those used for SAW filters. 5G filters must provide high bandwidth at high frequencies while handling higher power to maximize signal coverage. XBAR resonators meet those requirements, and because of the ISN software’s ability to factor in process technology, manufacturing a filter using XBAR resonators needs only three to five processing steps. The resonator’s simple structure consists of a metal comb line atop a piezoelectric surface with an air gap underneath (Fig. 2).
A prototype of an n79 5G filter was unveiled at the last Mobile World Congress and seized upon by Murata Manufacturing. Murata has since made a strategic investment in Resonant and committed to manufacturing four devices using the XBAR resonator for mobile applications. Today’s 5G phones are still being built with integrated passive device (IPD) or low-temperature cofired ceramic (LTCC) filters. Such filtering designs are adequate for now because there hasn’t yet been the anticipated explosion in 5G traffic.
But when the 5G bands from 3.3 to 5 GHz do experience that traffic boom, the proximity to the 5-GHz Wi-Fi bands and 6-GHz Wi-Fi 6E bands will result in interference. As a result, IPD or LTCC filters may not provide sufficient rejection of out-of-band signals. That’s when higher-performance filters, such as those possible with XBAR resonators, will become essential.
“We see, over the next two to three years, a transition to acoustic-wave filters to manage the looming interference problem,” says Eddy. Resonant believes its ISN software will provide value to both vertically integrated filter manufacturers, such as Murata and others, as well as to other companies in the RF/microwave space who don’t have access to filters except through foundry partners.