Evolution of the Smartphone

MWJPerspective

Todd Gillenwater, Mobile Products CTO and VP of Engineering Qorvo, Greensboro, N.C.

n less than 10 years, mobile Ride-sharing companies like Uber smartphone designed for global phones have evolved from rely on the ubiquity of mobile de- use may support more than 30 tools used mostly for talking vices that let users find and pay for bands today, compared with fewer and texting to highly complex, rides that are automatically direct- than 10 in 2010. The trend is con- Isophisticated devices that are cen- ed to their locations. tinuing, as regulators allocate new tral to almost every aspect of our All of these applications mean that higher frequency bands at 3.5 GHz lives. They have changed the way high speed, highly reliable data con- and above and refarm existing we communicate, letting us share nections are essential to everyone’s spectrum, such as the 600 MHz experiences in real time by upload- daily life. Demand for mobile data band previously used for TV broad- ing and downloading live video. continues to grow rapidly: the No- casting. Channel bandwidth has in- They guide us while driving, mak- vember 2016 Ericsson Mobility Re- creased, from 200 kHz with 2G to ing paper maps almost port predicts that global smartphone 20 MHz with 4G, even more using obsolete. They have subscriptions will rise from 3.3 bil- carrier aggregation (CA) in LTE Ad- become univer- lion in 2015 to 6.8 billion in 2022, vanced (see Figure 1). This is be- sal controllers with the data used per smart- ing combined with techniques for for Internet of phone expected to rise nearly 8x producing higher data rates and Things (IoT) to 11 GB/month over the same capacity from a given bandwidth, devices rang- period. Trends such as virtual re- such as MIMO, more complex ing from home ality and real-time video require modulation and network densifi- thermostats high performance networks cation, including the use of small to hotel room that provide low latency as well cells. door locks. as higher data rates. Today, key trends include LTE And they have To handle the growth in mo- Advanced, which network opera- enabled com- bile data, successive network tors are now deploying, and its panies to create generations—2G, 3G, 4G (LTE) successor LTE Advanced Pro (de- entire new cloud- and, in the future, 5G—have used fined in 3GPP Release 13 and based business a variety of methods to progres- later releases). These LTE network models. sively increase network enhancements ensure that 4G will capacity and data continue to play an important role rates. One way even after 5G begins to be de- is allocating ployed. LTE will operate in paral- many more lel with 5G and provide adequate frequency performance for many applica- bands. tions. LTE Advanced introduced A flag- CA, which aggregates the band- ship LTE width of up to five RF carriers,

TABLE 1 COMMON CELLULAR BANDS 1 Gbps 600 Mbps Band Uplink (MHz) Downlink (MHz) Duplex Type 4 CA 450 Mbps 2 × 2 MIMO 1 1920 to 1980 2110 to 2170 FDD 3 CA Cat 11/12 3 1710 to 1785 1805 to 1880 FDD 300 Mbps 2 × 2 MIMO 4 1710 to 1755 2110 to 2155 FDD 20 + 20 MHz Cat 9 150 Mbps 2 × 2 MIMO 7 2500 to 2570 2620 to 2690 FDD D ownlink ata R ate 10 + 10 MHz Cat 6 2 × 2 MIMO 17 704 to 716 734 to 746 FDD 25 1850 to 1915 1930 to 1995 FDD 2013 2014 2015 2016 40 2300 to 2400 2300 to 2400 TDD 41 2496 to 2690 2496 to 2690 TDD Fig. 1 Increasing downlink data rate enabled by carrier s 66 1710 to 1780 2110 to 2200 FDD aggregation. called component carriers (CC). between different bands, between LTE Advanced allows maximum the transmit and receive frequen- network data rates up to 1 Gbps, cies of each frequency division du- Third and LTE Advanced Pro will increase plex (FDD) LTE band and between Harmonic the possible number of CCs to 32, LTE and other wireless services, allowing up to 3 Gbps. Wi-Fi is be- such as Wi-Fi and public safety coming more important with the communications. High perfor- introduction in LTE Advanced Pro mance acoustic filters are needed Uplink Downlink Downlink of licensed assisted access (LAA), to achieve the required isolation. Band 17 Band 17 Band 4 which uses CA to combine licensed Bulk acoustic wave (BAW) filters LTE spectrum with unlicensed Wi- provide better performance (high- Fi spectrum at 5 GHz to achieve er Q factors) than surface acoustic s Fig. 2 With carrier aggregation, the greater data rates. wave (SAW) filters, especially at third harmonic of the Band 17 uplink higher frequencies, and are typi- signal can interfere with the receive SMARTPHONE TRENDS AND cally used for the most demanding signal in Band 4. CHALLENGES applications. As smartphone data perfor- The isolation challenges increase interference, the filters in the RF mance requirements continue to with CA, because the RF front-end front-end must provide very high grow, so does the complexity of must communicate simultaneously rejection of the problem harmon- the RF front-end, creating a series on multiple bands. This leads to ics without adding unacceptable of new challenges for the engineers new requirements for cross-isola- insertion loss. High linearity is re- working on smartphone designs. tion between bands, to avoid situ- quired in all the components— Today’s requirements are creat- ations where the transmit signal of PAs, switches, filters—to minimize ing design challenges that include one aggregated band interferes the generation of harmonics. maximizing linearity and isolation, with receive signals on another ag- A different cross-isolation issue managing power consumption and gregated band, which will degrade occurs when aggregating closely antenna tuning. Greater integration the sensitivity of the receiver. There spaced bands, which typically is required, since an increasing num- are various problem scenarios. share the same RF pathway within ber of bands must squeeze into the When aggregating widely sepa- the RF front-end. Examples include limited space allocated to the RF rated frequency bands, issues can bands 1 and 3 and bands 25 and front-end. Another important trend arise with harmonic frequencies 66. In these cases, the problem is is the emergence of smartphone generated at multiples of the trans- that the transmit frequency of one market tiers, each with differing RF mit frequency by non-linear com- band is close to the receive fre- requirements. This article discusses ponents in the RF chain, including quency of the other aggregated each of these and how the smart- power amplifiers (PA), switches and band. Multiplexers, which combine phone will evolve to handle 5G. even filters. With some band com- all the transmit and receive filters binations, such as bands 17 and 4 for multiple aggregated bands into Isolation (see Table 1), the third harmonic of a single device, provide an effi- The requirement to accommo- the lower frequency band (17) falls cient solution, allowing simultane- date more LTE bands within the into the receive frequency range ous use of the aggregated bands RF front-end creates challenges of the higher frequency band (4), while providing isolation between achieving the necessary isolation as shown in Figure 2. To prevent them. Multiplexers will become MWJPerspective increasingly important as network operators aggregate three or more Hexaplexer bands (see Figure 3), due to limits Band 1 on the available space and number Quadplexer Tx of antennas within the smartphone. Band 1 Band 1 Tx Rx Power Management Duplexer Single Filter Band 3 Band 1 Band 3 Several trends require better Rx Rx Tx Common Band 3 Common Common Common power management to maximize Port Rx Port Port Port smartphone battery life and avoid Band 3 Band 3 Band 3 Tx Tx Rx overheating. Greater transmission bandwidth, achieved with tech- Band 3 Band 7 niques such as uplink CA, will re- Rx Tx quire more power. Also, intra-band Band 7 uplink CA, which combines compo- Rx nent carriers within a single band, involves higher peak-to-average s Fig. 3 Multiplexers become increasingly important as network operators aggregate power ratios than standard LTE sig- bands to increase data rates. nals, increasing the demand on PA linearity. For example, doubling the a Network Problem” in this issue, space allocated to the RF front- uplink bandwidth with CA to 20 + page 102.) end has not increased. In today’s flagship phones, typically only 10 20 MHz (200 resource blocks) more Antennas than doubles the probability that to 15 percent of the internal area the peak-to-average power ratio of The number of antennas in is dedicated to cellular, Wi-Fi and the modulated signal envelope will smartphones has grown with the Bluetooth RF functionality. Smart- exceed 4.5 dB. requirement to support faster data phones are getting thinner, reduc- Another emerging requirement services as well as the growing ing the internal volume, and manu- is Power Class 2, a new standard range of RF frequencies and tech- facturers need to use the available that doubles output power to 26 nologies. Today’s handsets may space for new functionality and to dBm to overcome the greater prop- include as many as six or seven an- maximize battery size to respond agation losses of high frequency tennas, including primary cellular to user demands for longer op- bands (e.g., Band 41). The greater and diversity receive (DRx), Wi-Fi, erating time. As a result, the RF output power enables operators to near field communications (NFC) front-end must accommodate the maintain cell coverage as higher and other standards. It becomes an growing complexity, including sup- frequency bands are used, to add increasingly difficult engineering port for a growing number of fre- spectrum and support higher data challenge to fit more antennas into quency bands, within roughly the rates. Thermal performance be- the limited physical space available same space. Higher levels of RF comes critical at this higher power; within a typical handset. MIMO front-end integration are key to reliability depends on maintaining adds to the challenge, because it accommodating this growing RF acceptable device operating tem- requires simultaneous transmission complexity, particularly in flagship peratures by efficiently dissipating on multiple antennas. The need for phones that support a large num- the additional heat. additional antennas has been mini- ber of bands and CA combinations. Envelope tracking (ET), which mized by sharing the same antenna Besides saving space, the integrat- continuously adjusts the PA sup- among multiple services that use ed RF front-end modules simplify ply voltage to track the RF enve- similar frequencies, such as Blue- smartphone design, because many lope and maximize PA efficiency, tooth and Wi-Fi. Antenna sharing RF challenges are solved within the is spreading from flagship phones is also used to support faster Wi-Fi module rather than requiring the to mainstream use because it can performance, enabling 2x2 MIMO smartphone manufacturer to en- reduce power consumption and Wi-Fi by sharing the LTE DRx an- gineer solutions. This frees smart- heat dissipation. The increased ef- tenna between LTE and a second phone manufacturers to focus on ficiency provided by ET facilitates Wi-Fi RF chain. other aspects of smartphone de- the use of broadband PAs, which Antenna tuning, which adjusts sign, to attract consumers in an in- minimizes the number of PAs re- antenna impedance to optimize ef- creasingly competitive market. By quired in handsets. However, the ficiency at specific frequencies, has simplifying design, highly integrat- use of ET is limited to 20 to 40 MHz become increasingly prominent as ed modules also help smartphone channels and will need to evolve to a method to support the growing makers accelerate time to market. handle the wider bandwidth for ap- range of LTE frequencies. Also, smartphone manufacturing plications such as intra-band uplink RF Integration yields are improved because there CA. (Editor’s note: for further dis- Even though smartphones have are fewer components that can fail. cussion of ET, read the article “En- become larger, mostly due to us- An example of a highly inte- velope Tracking in Handsets Solves ers’ desire for larger screens, the grated RF front-end architecture MWJPerspective

GSM MB Tx B34/39Tx

B34/ SPDT B12 4/66 TRx1 Switch B1 GSM LB T x

B26 B25(2)

3 MIPI/Bias B8 (b)

B28A

S witch B7 B20

3 B28B B41 MIPI/Bias B29Rx SPDT

B40 SP5T Switch (a) 3 MIPI/Bias (c) s Fig. 4 To handle all major cellular frequencies, smartphones integrate the PAs, switches and filters into three front-end modules that cover the low (a) mid (b) and high (c) bands.

is Qorvo’s RF Fusion, which in- 5). Reducing losses makes it easier liferate worldwide, the market has cludes support for all major LTE to meet handset manufacturers’ stratified into distinct tiers: flagship frequency bands in three compact performance targets. This integrat- phones, which typically offer the placements: low, mid and high ed architecture also helps manage highest performance and global band (see Figure 4). Each module RF isolation, such as the interac- band support, and mid-tier phones includes PAs, switches and filters tion between bands aggregated that deliver good performance and and uses advanced packaging to for CA. For example, multiplexers are designed primarily for domestic achieve space savings of 30 to 35 supporting specific band combi- use, with only regional band sup- percent compared to the printed nations can be incorporated into port. Mid-tier phones have been circuit board (PCB) area required each module. In the future, it may popular in emerging markets such for discrete components. Besides be possible to integrate an entire as Asia and Latin America, where conserving valuable PCB real es- RF front-end into a single module, consumers are price-sensitive. tate, this approach has several ad- depending on whether the ben- According to Strategy Analytics, vantages: it improves performance efits to the manufacturer outweigh smartphone tiers are here to stay; by integrating components along the additional cost of including the company forecasts that mid- each pathway, which eliminates bands that may not be used by the tier phones will account for 37 per- the need for on-board matching, phone’s owner. cent of handset volume by 2022, reduces losses by as much as 0.5 up from 32 percent in 2016. How- dB, reduces current consumption SMARTPHONE TIERS ever, premium phones will contin- and the thermal load (see Figure As smartphones continue to pro- ue to generate the bulk of global smartphone revenue because they sell for a much higher price. This market stratification is re- M M flected in RF front-end trends. Highly integrated RF front-end (a) modules are typically seen only in premium handsets; mid-tier phones generally use a more dis- M M M M crete approach, such as Qorvo’s RF Flex, giving handset manufacturers

(b) the flexibility to reduce component costs by including only the filters required for local frequency bands. s Fig. 5 Integrated front-end modules (a) can reduce losses by up to 0.5 dB compared As the RF environment becomes to on-board matching (b). more complex, greater RF front- MWJPerspective end integration will be seen in mid- cluding enhanced mobile broad- techniques and sophisticated algo- tier phones. Market stratification band, and will focus on frequen- rithms to compensate for the great- has also spawned other trends, cies below 6 GHz. Currently, bands er path losses at these frequencies. such as mid-tier phones that sup- between 3.8 and 4.99 GHz are Coexistence between the millimeter port dual SIMs. These enable users under consideration. Since 4G LTE wave frequencies and 4G LTE will to minimize monthly costs, by us- will continue to be widely used, an pose other challenges. ing different services for voice and important goal is to maximize co- Tunable analog filters are being data or using local operators when existence by designing the specifi- explored to reduce the number of traveling to different countries. cations for the 5G new radio (NR) filters in the RF front-end. However, Dual SIMs are supported with dual to minimize interference with 4G it is extremely challenging to create a transceivers or techniques that use bands. In general, the RF front-end broadband tunable analog filter that a single transceiver, such as time technologies required for 5G Phase meets the power handling, insertion multiplexing. 1 may be similar to those used for loss, isolation and coexistence needs 4G LTE, with improvements to sup- of modern smartphones. The adop- THE FUTURE RF FRONT-END port the new requirements. tion of CA, which requires simulta- The smartphone RF front-end The main challenges for Phase neous operation on multiple bands, will continue to evolve and grow in 1 include PA performance and in- introduces complications that make complexity over the next few years creased linearity to support new broadband tunable filters even more to meet changing requirements. uplink requirements designed to difficult to achieve. In addition to supporting 5G, the improve performance for data- Silicon technologies such as smartphone will likely assume the intensive applications. These en- CMOS, SOI and SiGe will con- role of a hub for communications compass more complex modula- tinue to spread throughout the RF among rapidly proliferating IoT tion (e.g., 256-QAM) and optional front-end for control, switching and devices and sensors, which will re- cyclic prefix orthogonal frequency amplification. Nonetheless, GaAs quire the integration of new air in- division multiplexing (CP-OFDM). offers performance advantages terfaces into the RF front-end. The adoption of 4x4 MIMO to fur- for PAs and is likely to remain the 5G specifications development ther boost data throughput will dominant PA process technology. currently focuses on three high-lev- require increased integration and Digital signal processing is increas- el use cases, including IoT: more complex switching, as well ingly being employed to improve • Enhanced mobile broadband, to as increasing the number of anten- front-end performance through support consumers’ use of video nas. Enhancements to bulk acous- functions such as digital predistor- and other data-intensive mobile tic wave (BAW) filter technology, tion (DPD) and ET. applications, which will require which is widely used for higher- lower latency as well as faster frequency LTE bands, are being CONCLUSION data rates developed to meet the filtering The growing complexity of • Massive machine type commu- requirements for the sub-6 GHz the RF front-end and the ever-im- nications for IoT applications; frequencies for 5G. proving performance will require and 5G Phase 2 (3GPP Release 16), a mix of many different technolo- • Ultra-reliable and low latency expected by the end of 2019, will gies combined in highly integrated communication for applications include the remaining use cases and modules. Suppliers with the abil- where reliability and perfor- focus on much higher frequencies ity to design using such a wide mance are critical, such as au- (28 GHz and above). Though speci- range of technologies, including tonomous vehicles. fications are at an earlier stage than advanced filtering and millimeter 5G specifications are due in Phase 1, it is clear that communica- wave, will be best positioned to two main phases: Phase 1 (3GPP tions at millimeter wave frequen- succeed as the smartphone enters Release 15), due in late 2018, will cies will require radical new antenna its second decade.■ support a subset of use cases, in- array architectures, beamforming