Selecting Filters for Challenging Mobile Applications Worldwide
The global transition to 4G networks is driving a dramatic increase in the need for advanced filtering technologies in smartphones and other mobile devices. As a result, filter selection has become a key consideration for designers of these devices. As 4G LTE networks proliferate and WiFi becomes ubiquitous, smartphones must support an increasing number of RF bands. These bands must be isolated within each device to avoid interference, using filters.
While 3G networks used only about five bands, there are already over 20 4G LTE bands and this number could rise to more than 40 in the future. Though it’s not practical to support all worldwide bands in a single smartphone, a feature-rich model for international use might need to filter transmit and receive paths for 2G, 3G and 4G in up to 15 bands, as well as WiFi, Bluetooth and GPS. Such a phone might require as many as 30 to 40 filters. The situation is likely to become even more complex in the future: next-generation high-end smartphones could include 50 or more filters.
The filtering requirements differ in each region or country, due to local differences in spectrum allocation that are creating a worldwide patchwork of LTE bands. Adding to the challenges, many of these bands are very close together and therefore require highly selective filters. One reason is that the global demand for spectrum is leading governments around the world to re-farm existing spectrum; this often results in the allocation of new bandwidth that is adjacent to existing bands, often with minimal or even non-existent guard bands.
Some LTE bands are located next to the unlicensed international, industrial, scientific and medical (ISM) bands between 2.4 and 2.5 GHz, which are used to support existing standards including WiFi, Bluetooth and ZigBee, as shown in Figure 1. These bands are crowded with signals from many devices, including PCs and cordless phones. Smartphones may therefore require advanced filters that support LTE-WiFi coexistence.
The complexity of filtering requirements will further increase due to carrier aggregation, a new 4G LTE-Advanced capability that enables carriers to aggregate multiple fragmented slivers of spectrum into a single wider channel to enable higher data rates.
The combined effect of these trends is that mobile devices must isolate signals with bandpass filters that provide extremely high rejection of adjacent bands combined with low insertion loss. Advanced acoustic filters have these attributes, and are essential for addressing these challenges.
In this article, we’ll discuss 4G filtering challenges with a focus on the regional differences that are unfolding in North America, Europe and Asia. We’ll identify which filter technologies designers should consider as they seek to build devices for each market. Acoustic filter technologies continue to evolve to meet the growing challenges: advanced bulk-acoustic-wave (BAW) and temperature-compensated surface acoustic wave (TC-SAW) filters can solve some of the toughest mobile device filtering problems and are key enablers of the global 4G transition.
BAW Benefits: The Relationship between Low Insertion Loss, High Q and Sharper Corners
BAW filters offer lower loss and higher Q than SAW filters. These characteristics mean that for challenging applications, BAW filters are often a better choice – even for some frequency bands that are within the range of either filter technology. The diagram here compares the effectiveness of a higher-Q BAW filter (Q=2500) and a lower-Q SAW filter (Q=1000) for the same frequency band (Band 25 TX).
Clearly, the BAW filter offers lower insertion loss. But there is another important difference: the BAW filter’s response curve has sharper corners, while the lower-Q SAW filter’s curve has more rounded corners. Because of this, the difference in loss between the two filters increases at the corners. At 1.88 GHz, the center of the band, the difference in loss is about 0.5 dB. But at 1.85 GHz, the lower edge of the band, the difference is 0.8 dB, while at 1.915 GHz the difference increases to 1 dB. The rounder corners of the SAW filter’s response curve impose the greatest penalty on the channels at the edge of the frequency band. Effectively, the passband of the lower-Q filter becomes narrower.
Narrow modulations such as GSM (200 kHz) and CDMA (1.25 MHz) will suffer the most sensitivity loss at the band edge due to this effect, while WCDMA will suffer less. LTE results depend on the system bandwidth, with narrower bandwidths more affected.
Drift due to temperature variation will make the problem worse, unless temperature-compensated processes are used. At higher temperatures, the SAW response drifts downward by as much as 4 MHz – nearly twice as much as BAW – resulting at an additional 1 dB difference in loss at the high end of the frequency band.
It might seem tempting to widen the lower-Q filter’s response curve to improve the insertion loss problem. However, this will result in reduced selectivity to nearby interference. This selectivity degradation is not acceptable for many demanding filter applications.
Overview of Filter Technologies
Several acoustic filter technologies exist, differing in their capabilities, cost and packaging technology. Today, the primary choices are surface acoustic wave (SAW) and BAW technologies. TC-SAW filters, which are much less sensitive to temperature changes, have also recently been developed. Each of these technologies can offer low insertion loss with good rejection when used for suitable applications. Each technology is a good match for a specific range of applications, as shown in Figure 2 and described in this article.
SAW is a mature technology widely used in 2G and 3G receiver front ends, duplexers and filters. SAW filters are well suited for frequencies up to about 1.9 GHz, including several standard GSM, CDMA and 3G bands.
A key advantage of a SAW filter is its low cost. In addition, techniques such as wafer-level packaging are being used to shrink SAW filters, allowing the integration of filters and duplexers for multiple bands onto a single chip. This is becoming increasingly important as smartphones incorporate more functions.
Limitations of SAW filters include frequency range and temperature sensitivity. Above about 1 GHz, selectivity declines; at about 2.5 GHz, the use of SAW is limited to applications with modest performance requirements. SAW is also very temperature-sensitive; a SAW filter’s response may shift downward by as much as 4 MHz as temperature increases. This limitation has become more significant as guard bands become narrower and consumer devices are specified to operate across a wide temperature range (-35° to +85°C).
Compared to SAW, BAW filters generally offer superior performance (higher Q) with lower insertion loss. With BAW technology, it is possible to create narrowband filters with exceptionally steep filter skirts and excellent rejection. This makes BAW the technology of choice for many challenging interference problems.
BAW delivers these benefits at frequencies above 1.5 GHz, making it a complementary technology to SAW (which is most effective at lower frequencies). BAW can address frequencies up to 6 GHz and is used for many of the new LTE bands above 1.9 GHz. BAW is also highly effective for LTE-WiFi coexistence filters.
Because BAW filters offer low insertion loss, they help compensate for the higher losses associated with the need to support many bands in a single smartphone. Besides improving signal reception, lower loss also contributes to longer battery life. BAW excels in applications where the uplink and downlink separation is minimal and when attenuation is required in tightly packed adjacent bands.
For some challenging applications, reducing sensitivity to temperature changes is critical. This includes situations in which bands are located extremely close to one another. New TC-SAW technology enables the manufacture of highly selective filters that address these challenges. These filters have minimal or zero drift with changing temperatures, which is essential to ensure good rejection of adjacent bands and minimize insertion loss.
TC-SAW filters are effective within the same frequency range as SAW2up to about 1.9 GHz. The reduced sensitivity to temperature change makes TC-SAW a good choice for challenging specifications including some new 3G and 4G WCDMA duplexers and filters. For example, TriQuint uses TC-SAW to support Band 13, which is close to the U.S. public safety band, as well as Band 20 and Band 26 duplexers. The enhanced SAW process adds slightly to filter cost.
The Importance of Temperature-Compensated Filters
Temperature drift has become an important issue as spectrum becomes more crowded and high selectivity is required to minimize insertion loss and ensure rejection of adjacent bands.
The diagram illustrates a difficult Band 20 duplexer filtering problem that is solved by using a TC-SAW TX filter. The TC-SAW filter used in this design has a temperature drift that is roughly half that of a typical SAW filter. This duplexer must meet a critical specification for attenuation of the signal in the transmit path at the Band 20 receive frequencies. As shown in the diagram, the specification requires attenuation of at least –45 dB for the 791 to 821 MHz Band 20 RX frequencies.
At +25°C, the filter easily meets this requirement, achieving –45 dB attenuation across a frequency range that includes the RX frequencies and extends to an upper limit of 822.3 MHz. At +90°C, this attenuation point shifts down by 1.07 MHz, but the filter still achieves –45 dB attenuation at 821.23 MHz and below, including the entire Band 20 RX range, and therefore remains compliant with the specification. However, a SAW filter that is not temperature-compensated would drift by approximately twice as much — by –2.14 MHz at +90°C. With this drift, the –45 dB attenuation point would shift to 820.36 MHz and the filter would therefore violate the specification.
Local differences in spectrum allocation, particularly the bands used for LTE, are creating a complex worldwide patchwork of frequency band assignments. Not too long ago, the picture was relatively simple. The primary four 3G bands were 1, 2, 5 and 8. Within the U.S., Bands 2 and 5 were the most commonly used; adding support for 1 and 8 allowed roaming in other countries.
With 4G networks, the situation is much more complicated; there are already significant differences in band allocations between regions and even between countries, and the situation will become more challenging as more LTE bands are allocated. Smartphones must of course also continue to support the primary 2G and 3G bands, as well as WiFi and Bluetooth. In some regions these bands are being refarmed for LTE, which may also change the filtering requirements; narrow 5 and 10 MHz 4G LTE bands require filters with sharper corners than 3G WCDMA, for example.
Adding to the filtering requirements, smartphones need multiple filters for each FDD-LTE and TDD-LTE band. For each FDD-LTE band, most smartphones require three filters: a duplexer for the primary TX and RX paths plus an additional filter for the secondary RX path. For each TDD-LTE band, smartphones will typically require at least two filters.
Because it is impractical to build a single phone that will work with all bands, manufacturers typically design smartphones for regional or country-specific use, or for a specific regional carrier’s network. Phones also may include additional filters and other components to support roaming in other regions. Manufacturers may choose to include support for additional bands, where it is practical to do so, in order to be able to certify a single device for use in multiple countries. Because of the differences in local spectrum allocations, the challenges facing designers vary from region to region.
The United States, along with Japan and Korea, is leading the worldwide transition to 4G networks; it is estimated that about half of the world’s LTE devices have been sold in the U.S. The filters required in any specific model destined for the U.S. market will vary depending on factors such as whether the phone is designed for use with a specific carrier’s network or for use on multiple networks.
In addition to Bands 2 and 5, major carriers are using a number of FDD-LTE bands across a broad range of frequencies, including Bands 4, 13, 17, 25 and 26. Of these, Bands 2 and 25 will require BAW filters.
WiFi coexistence will likely remain less challenging in the Americas than in other parts of the world. Most LTE bands used within the region are not immediately adjacent to the 2.4 GHz WiFi spectrum, and therefore most devices will not require high-performance filters to separate LTE and WiFi signals. An exception is the TDD-LTE Band 41, which will require a high-performance BAW filter. Other bands requiring highly selective filtering include Band 13, which is close to the U.S. public safety band and will require a TC-SAW filter.
There are significant differences between the European market and North America. LTE adoption is proceeding more slowly, and there are differences in band allocation. Band 3 was originally used for GSM, the dominant 2G technology used in Europe, but is now being refarmed for LTE. In addition, two new “greenfield” frequency bands, 7 and 20, are being assigned for LTE use. It is likely that manufacturers building LTE phones designed for Europe will support all three bands to allow roaming.
The use of Band 7 introduces a challenging requirement for a high-performance BAW WiFi coexistence filter, since the Band 7 transmit path is only about 15 MHz from the 2.4 GHz WiFi band. Band 3 also requires BAW, while Band 20 may be best supported with TC-SAW.
Looking to the future, Europe is also auctioning Band 38 spectrum, which lies in the gap between the Band 7 transmit and receive bands. If a device needs to support both Band 7 and Band 38, a high-performance BAW filter may be required to allow coexistence.
The LTE picture in Asia is a complex map of regional band assignments with several distinct local markets. China is a huge potential market with unique requirements; other countries also have distinct needs, notably Japan and Korea, which, like the U.S., are the two nations that are most rapidly moving to LTE.
In China, the predominant LTE technology is TDD-LTE, as opposed to the FDD-LTE used in North America. Many of the LTE bands are at higher frequencies, including several that are adjacent to the WiFi band. This situation creates a strong requirement for BAW WiFi coexistence filters. For example, two of the TDD-LTE bands allocated are Bands 40 and 41. The WiFi frequencies are sandwiched between these two bands. As shown in Figure 3, there is no gap between Band 40 and the lower end of the WiFi band, and only a minimal gap between Band 41 and the upper end of the WiFi band.
High-performance BAW coexistence filters will be needed; in addition, tradeoffs may be necessary depending on customer priorities. Supporting the full width of Band 40 may require giving up some of the lower WiFi channels. Alternatively, manufacturers may choose to give up part of Band 40 if supporting the full WiFi bandwidth is their top priority. The coexistence situation with Band 41 is slightly less challenging because of the minimal guard band between Band 41 and the WiFi spectrum. Within China, there is some local variation; for example Bands 7 and 38 replace Band 41 in Hong Kong.
Korea is particularly interesting because of its high smartphone use and rapid LTE adoption. About 26 million people 2more than half the population 2already have smartphones, and about 15 million of those are expected to be using LTE by the end of 2013.
Korea is refarming Bands 3 and 5 for LTE, and starting this year, every phone is expected to support Band 7. Band 26 has also been allocated and will require TC-SAW; though it overlaps with Band 5, Band 26 includes some frequencies that a Band 5 filter cannot cover. Bands 3 and 7 will require BAW filters; there is also a substantial need for WiFi coexistence filters.
The situation in Japan further adds to the regional complexity. Japan is unusual in its use of Bands 26, 11 and 21; Band 41 is also used, requiring a BAW WiFi coexistence filter.
Other Asia-Pacific Regional Challenges
Another interesting filtering challenge throughout the region is being created by the so-called digital dividend – the reallocation of 700 MHz spectrum freed up by the switch from analog to digital TV. Band 28, within this range, is uniquely designed so that it will require a pair of duplexers with overlapping frequency ranges.
Strategies for Asia
The question for manufacturers is whether to build phones for individual Asian countries or attempt to make phones that can be used throughout the entire region. China clearly is a large enough market to drive its own requirements; Korea, with its high percentage of LTE users, is another example of a market with its own identity. India is another huge potential market, although the high cost of smartphones relative to average personal income tends to restrict adoption; today most sales are of relatively low-end phones.
Within each local market, fragmented spectrum presents a challenge for service providers trying to address the insatiable appetite for high-speed wireless data networks. Carrier Aggregation (CA) is a feature of 4G LTE-Advanced designed to allow service providers to offer higher data rates by bonding non-contiguous fragments of spectrum into a single wider channel of up to 100 MHz. CA is expected to become a major trend beginning in 2014; not surprisingly, markets leading in LTE deployment, such as the U.S., are expected to be the first to utilize CA. Initially, service providers are likely to focus on bonding widely separated bands, which may require only a diplexer; however, later implementations may present more challenging filtering problems that require high-performance filters.
Packaging and Integration
As smartphones support an increasing number of RF bands, combining multiple filters into a single package can free up valuable real estate. Some manufacturers organize filters in various combinations, from duplexer banks that consolidate several filters into a single module to two-in-one duplexers that permit the use of two-receiver operation simultaneously and independently. Wafer-level packaging (WLP) is a valuable technique for reducing size; it will increasingly be applied to BAW as well as SAW. Different filter technologies may be combined for specific applications. For some newer LTE bands, filters are more likely to be discrete simply because the bands are relatively new. This enables designers to simply add discrete LTE “satellite” components to existing layouts in order to offer regional 4G variants.
The worldwide transition to 4G networks is creating complex new challenges for designers of smartphones and other mobile devices. Fortunately, advanced acoustic filter technologies are becoming available to address these challenges. Each filter technology 2 SAW, BAW and TC-SAW 2 is suited to specific applications. With techniques such as WLP, it’s also possible to combine multiple technologies into duplexers targeting specific applications. By careful selection of duplexers and individual filters, designers can meet the exacting requirements of each region worldwide.