ACCO Semiconductor closed a $35 million funding round led by investor Bpifrance Large Venture, with participation by Foundation Capital, Pond Ventures, Partech Ventures, Omnes Capital, Siparex Group and A Plus Finance. ACCO will use the funding to launch its RF front-end CMOS solutions for LTE smartphones. This is a sign of the potential of CMOS power amplifiers in cellphones and the confidence of the investors in this potential, a potential that is still largely unrealized.
ACCO has not publicly released technical details of its products, but the company says it has monolithically integrated the power amplifier and control functions into integrated circuits fabricated using standard bulk CMOS. In contrast, competitors offering CMOS power amplifiers for mobile phones generally use high-voltage CMOS or CMOS on SoI, with SoI preferred especially for switching. RF switching co-located with a PA can be used to select among external, band-specific duplexers, connect a chosen duplexer to the antenna, or both, reducing the number of component packages in the front end and simplifying the phone.
Note that some in the industry use the term TXM (transmit module) to describe a 2G or 3G/4G PA with switch such as developed by ACCO Semi. If the PA handles 2G, 3G and 4G, then suppliers often use the term “multimode, multiband PA.” ACCO has developed both using bulk CMOS.
The number of LTE bands supported in flagship smartphones has over the past 18 months increased to 26 or more depending on how you do the count, and flagship phones such as the iPhone 6s now require an extraordinary number of duplexers, filters, PAs, switches and switch throws. The GaAs-based approach to this challenge is to put as much of the RF front end as possible into modules such as one of more TRMs (Total Radio Modules such as the Skyworks SkyOne), S-PADs (switched PA-duplexers), antenna switch modules and antenna duplexer modules. CMOS has the potential to reduce the size and cost of the PA by combining the power amplification, switching and control functions monolithically, an attractive proposition to OEMs.
CMOS PAs have had great success in GPRS mobile devices, and some success in low-cost W-CDMA (3G) devices. In LTE devices, CMOS has not done quite as well. Qualcomm announced its CMOS PAs for LTE in 2013, but as of early 2016, the PAs had shipped in only a few lower-volume LTE smartphones. Peregrine Semi (Murata) announced a CMOS PA for LTE in 2014, but has not shipped in production volumes yet to our knowledge.
What’s the problem with CMOS for LTE? Suppliers of CMOS PAs have had difficulty matching the performance of GaAs-based PAs in leading-edge LTE smartphones, especially in bands above 950 MHz. LTE has kept evolving to higher data rates, and GaAs has improved, so CMOS suppliers such as Qualcomm have had catch up in terms of linearity and efficiency. ACCO’s claim is that they can not only match the performance of GaAs-based PAs, but they can provide smaller, lower-cost PAs by using bulk CMOS instead of more costly SoI substrates or HV CMOS processing for fabrication.
Is this another nail in the coffin of GaAs PAs? Not by a long shot. As lower-tier LTE smartphones and IoT devices proliferate, use of CMOS PAs for LTE should grow. At the same time, GaAs-based PAs will continue to meet the demands for higher performance, proven solutions in higher-end smartphones. Five years from now, perhaps, the two will meet somewhere in the middle.