The handset’s RF front end has never been regarded as a strategic component like the hotly discussed baseband or apps processor. But in the world of LTE, where spectrum bands are fragmented and every device sports MIMO antenna arrays, the radio element has become both hugely complex, and very strategic. A product that can handle many frequency bands elegantly while retaining battery life and overall performance will be a great competitive asset. This should create new opportunities for the RF/power amplifier specialists like Skyworks, Avago and RFMD, and for providers of enabling technologies, such as Nujira. But it will also attract new players, and Qualcomm exploded into the sector just before Mobile World Congress with a radio chip promising to support no fewer than 40 LTE bands.
This, of course, raised the prospect of a true LTE ‘worldphone’, and of a shake-up in the RF chip segment ‘ shares in the three main suppliers all fell on the launch, though they recovered somewhat later as analysts questioned how far Qualcomm would push into the merchant space. But even if its main goal is to drive further integration of the handset platform, and it sticks primarily to devices running on its own processor/modem combinations, that is still a huge part of the LTE market which it will be harder for third parties to penetrate (Qualcomm has about 50% of the mobile apps processor space and an estimated 85% share of FD-LTE modems).
Qualcomm has always done its own RF design, but has not offered a full RF front end and has relied on third parties even for some transceivers. Any move by a firm of this scale both unsettles a market segment and accelerates change, especially as it is leaping in ahead of some specialists, which will be keen to get returns on their earlier generation platforms first. Qualcomm has been halfhearted about RF in the past, but it does nothing by halves, as its RF360 chip shows. Even if its inexperience and lack of track record weaken the product’s chances, it will change the agenda, putting the fully integrated handset platform and the truly multiband 2G/3G/LTE device firmly on the roadmap, thus forcing established players to respond.
The RF front end is the group of components which transmits and receives signals between the radio transceiver and the antennas, and includes signal filters, power amplifiers and antenna switches. This faces a challenging task in LTE, where there are seven band/mode (TDD/FDD) combinations being deployed on a large scale basis, and many niche ones. Indeed, it has been the main obstacle to devices capable of supporting at least a large subset of the bands specified in the LTE standards – 23 FD-LTE and 11 TD-LTE frequencies in 3GPP Release 10, and more to come in subsequent releases. The importance of addressing the frequency issue was highlighted when Apple chose a somewhat quirky selection of bands for its iPhone 5 ‘ for instance, European carriers with 1.8GHz LTE networks could run the 4G device, but not those with the equally common 2.6GHz/800MHz combination.
Supporting multiple bands and modes was then, unsurprisingly, a key feature of component suppliers’ discussions at MWC, but Qualcomm stole the thunder from its rivals with the RF360.
The RF360 supports LTE in TDD and FDD modes as well as all the 3G standards plus EDGE, and is enabled for upgrade to LTE-Advanced features, notably carrier aggregation. It works seamlessly with Qualcomm Snapdragon and Gobi hardware but could also be combined with third party mobile platforms. The company says it reduces chip footprint by up to 50% and also reduces power compared to current offerings.
“The wide range of radio frequencies used to implement 2G, 3G and 4G LTE networks globally presents an ongoing challenge for mobile device designers. Where 2G and 3G technologies each have been implemented on four to five different RF bands globally, the inclusion of LTE brings the total number of cellular bands to approximately 40,” said Alex Katouzian, SVP of product management at Qualcomm Technologies. Handsets featuring the Qualcomm RF360 chip should hit shelves during the second half of this year.
The company actually announced five products in the radio area. One is a follow-on ‘ its latest RF transceiver, the WTR1625L, which doubles the number of supported RF bands in a smaller footprint compared to its predecessor, and adds support for carrier aggregation and TDD. The other products were firsts for Qualcomm, filling in the other elements of the RF front end – the antenna matching tuner; multiband power amplifier (PA); and an integrated antenna switch with envelope tracking support. All these are based on CMOS technology and Qualcomm also announced a new manufacturing process for them, called RF POP. This allows frequency specific components like third party filters to be layered like a sandwich on top of the PA and switch module.
The barriers to a worldphone in LTE have been seen less in basebands and transceivers than in the RF front end (although, as Michael Thelander of Signals Research points out, no transceiver, including Qualcomm’s latest, comes close to simultaneous support for all 34 Release 10 bands). In the front end, each new band could require additional switches and filters, increasing cost and footprint and degrading overall performance. With RF360, Qualcomm addresses many of the issues ‘ Signals Research believes it has only omitted two key elements of a worldphone solution, full support for a standards-based interface like MIPI or Open ET, and a tunable duplexer (though RFPOP does allow duplexers to be switched in and out easily).
The new platform does support other essential new enablers including envelope tracking (ET) and a CMOS power amplifier, allowing for power efficiency. Indeed, if Qualcomm does deliver a CMOS PA with ET support for LTE this year, it will have brought forward the timelines for the industry ‘ most analysts thought this vital technology would be ready for HSPA+ in 2013, but not for LTE until 2015 or later. Tim Haynes, CEO of Nujira, the UK-based pioneer of ET, gave his own timeline of 2014 when the firm recently announced its Woodstock reference design for smartphone RF front ends. This is based on Nujira’s recently launched NCT-L1200 ET chip, and covers up to 16 bands from 700MHz to 2.7GHz in a printed circuit measuring under 200 square millimetres, supporting GSM, 3G and LTE.
Qualcomm’s other important technical move is to incorporate an antenna matching tuner to allow a single antenna to span a far wider range of frequencies. This is not new technology, but it has not been deployed by mainstream baseband providers, so Qualcomm’s decision is important to the future of integrated devices. RF front end makers will shudder, given what happened in the RF transceiver market ‘ as most baseband vendors adopted their own transceiver, the merchant market contracted, and the danger for RFMD and the rest is that the same may happen in front ends. Qualcomm is probably thinking most urgently about its baseband rivals, particularly MediaTek and Intel, and seeking to put itself several steps ahead of them while owning as much of the handset component count as possible to increase profitability and control. But if it scores some success with RF360, it will start to eye the RF vendors, and will have to make some difficult choices between stealing their lunch, and continuing to cultivate them as allies.
For smaller firms, this could be good news. Qualcomm’s weight means its endorsement of the ET/CMOS PA combination, and its belief this is ready for commercial real time, will be significant for companies like Nujira and Quantance (ET pioneers). It is not clear which ET specialist is working with Qualcomm, but the technology gains as a whole from the giant’s confidence in it. Indeed, Nujira was moved to put out a response to the RF360 announcement, though the fairly critical tone suggested Qualcomm had looked elsewhere for its ET.
The UK firm said: ‘First and foremost, it’s a great validation of Envelope Tracking, and confirms that our technology is one of the key enablers for global, multiband RF front end solutions. For the last 18 months, we have been evangelizing the benefits of the ET/CMOS PA combination … We have also been promoting the benefits of a single, global RF front end for as long as we can remember. Qualcomm has executed on a strategy that is a combination of both these threads, and taken it a stage further with the 3D POP packaging.’
Meanwhile, if Qualcomm gets serious about the RF front end, it may be shopping around for expertise and patents, which could raise market excitement about small firms like CMOS PA makers Amalfi, Javelin and Black Sand. Similarly, other baseband suppliers may feel obliged to exert greater control over their RF offering, whether launching/acquiring their own or pushing their partners to redouble their efforts. This will only become more pressing as LTE deployments reach mass market levels, and the bottleneck in the RF front end threatens to cancel the benefits of advanced baseband technologies like Nvidia’s Icera software defined modem.
However, there are drawbacks to the integrated approach Qualcomm loves, and is now pursuing in the RF chain. In Nujira’s view, the ‘one size fits all’ approach could compromise RF performance, which is a key differentiator in LTE. It said in its statement: ‘Offering the whole RF chain as a module has some significant drawbacks – OEMs will be unable to choose their own band combinations, instead being even more reliant on Qualcomm’s timescales, product offerings and features. The number of antennas, their type and performance, will be dictated by the module. If the Qualcomm module provides a single broadband antenna output, and the OEM wants to have three separate antennas for better performance, then that’s tough luck.’ There is always suspicion of single-source solutions, and in addition, there are still some very specific RF requirements to address in certain regions, such as Verizon’s B13 700MHz band being adjacent to public safety spectrum, which requires specific tuning.
Of course, another opportunity for Nujira may lie in patents. With most of the major chip vendors working on ET support, Haynes told Electronics Weekly: ‘We know there are two ways to do envelope tracking, our patented way and one other less efficient way. We will be looking at these other vendors’ chips very closely with an oscilloscope and a microscope when they come on the market.’
And while shares in Skyworks, Avago and RFMD fell on Qualcomm’s announcement, they were recovering somewhat by the end of last week as analysts gave their investors reason to hope that the RF360 would expand the market for everyone, and that Qualcomm would not be an aggressive competitor. For instance, Lazard Capital Markets said in a research note that the US giant would not risk damaging integrated circuit partnerships for a market opportunity it estimates to be between $400m and $600m in total, but that the RF360 would be ‘largely a ‘proof of concept’ to pressure the industry to solve band fragmentation, vs a dominant solution that could threaten RFMD.’ Qualcomm holding back from seizing another slot in the smartphone and a chance to carve out new territory in LTE? Lazard sounds hopeless over-optimistic. It’s too early to tell whether the RF360 will gain significant traction, or whether Qualcomm will be effective in this market, but there is no doubt it will throw its full weight at the challenge.