Exploding RFFE Complexity


Nearly every day, I am surprised how senior digital communications engineers, systems architects, and senior technical marketers at broadband wireless operators/carriers, 4G LTE application processor SoC suppliers and even 4G cellular digital baseband processor suppliers seemingly fail to fully appreciate the Radio Frequency (RF) front-end circuits that make all wireless communications possible.

Even among 4G LTE-A PHY layer modem transceiver engineers, the RF front-end is a black magic analog circuit that sits beyond the ADCs (receiver or RX signal chain) and DACs (transmitter or TX signal chain).  Most wireless communication technologists understand that the RF Front-End (RFFE) modulates the TX OFDM signal from baseband to RF and demodulates the RX OFDM signal from RF to baseband.  The sobering reality is that RF circuit design is not routinely taught to most Electrical Engineering students (in undergraduate or graduate programs), largely, I believe, as a consequence of the unquenchable demand for software used in microcontrollers / microprocessors in embedded systems such as Internet of Things sensor devices, which are perceived as somehow more technically interesting or ia safer bet in terms of employment opportunities.

This state of affairs is great news for the four dominant suppliers of RFFEs for smartphones – Skyworks (SkyOne™ product family), Qorvo (RF Fusion™ product family), Qualcomm (RF360™ product family) and Peregrine/Murata (Global 1™ product family).  It would be very risky from a technical perspective, I would posit, for a Tier 1 smartphone supplier (such as Apple or Samsung) to develop their own RFFE modules (note that three or four RFFE “SKUs” are required to support the regional cellular bands worldwide), precisely because the available pool of talented/experienced RF engineers is so small (and definitely focused outside of Silicon Valley and South Korea).

So the good time roll at these four suppliers, and will roll for some time. For example, industry sources have reported that SkyWorks has significant RFFE content (not likely a SkyOne design win as yet) in the iPhone 6 and 6 Plus:

“Skyworks enjoyed a stellar year in which its analog sales increased 42% mostly due to strong worldwide smartphone sales. Skyworks Solutions makes analog and mixed signal semiconductors for Apple, Samsung, and other suppliers of mobile devices. Multiple power amplifier components from Skyworks are found in Apple’s iPhone 6 models. It has been estimated that Skyworks supplies $4 worth of content from every new iPhone 6 handset.” Source: IC Insights (May 2015)

Wow, $4 per unit with Apple selling about 50Mu of the iPhone 6 SKUs per quarter.  If my math is right, that is about $800M of revenue for Skyworks for the iPhone 6 / 6 Plus design wins alone.

One of the smartphone industry dynamics very few people (including institutional investors and financial analysts) seem to understand is Qualcomm is alone in offering a complete smartphone chipset: application processor plus digital baseband processor (“modem”) plus radio transceiver plus RF front-end, as shown below:

Qualcomm’s commitment to providing a complete 4G LTE and legacy standards smartphone chipset puts them at a huge advantage in my view.  If I were a senior manager at a tier 1 smartphone OEM, I would be 100% focused on technical risk reduction and relentlessly hitting production schedules.  A pre-integrated complete cellular chipset with RFFE is clearly a big risk reduction.

Peregrine makes the complexity of contemporary 4G LTE-A Carrier Aggregation RFFEs (caused largely due to the number of FDD bands than must be supported worldwide) easy to understand with the following image:

Careful Peregrine/Murata, if I were a design manager, an increase of 5000X in complexity sounds like major schedule risk to me!

And hey, let’s do away with FDD mode in 5G, okay?  TDD mode is just so much easier!