Technical Insight
CMOS amplifiers get ready to kick GaAs
Samsung's Galaxy smartphones are amongst the first mass-produced 3G handsets to feature high-performance CMOS power amplifiers. Will silicon displace GaAs in this sector, asks Compound Semiconductor.
Samsung Galaxy Ace Plus: Does Samsung's choice of CMOS PA in its smartphones signal an industry shift from GaAs to silicon?>
2012 has been a good year for Texas-based Javelin Semiconductor. Following a string of announcements that South Korean IT giant, Samsung Electronics, had chosen its CMOS power amplifier for Galaxy smartphones, the wireless IC start-up has at last proven that CMOS can rival GaAs in the 3G handset market.
As Brad Fluke, Javelin chief executive says: “We are now shipping in numerous handsets... manufacturers will be able to [take advantage] of the benefits of our PAs across all 3G designs worldwide.”
But will they? And if they do, what does that mean for GaAs-based power amplifier heavyweights such as Avago, TriQuint, RFMD, Skyworks and Anadigics?
As part of his “Samsung Galaxy Appeal Teardown”, Jim Mielke, vice-president of engineering at ABI Research, US, showed that the Javelin CMOS power amplifiers matched, and even outperformed, GaAs counterparts in some tests.
Rival semiconductors: The Javelin J5501 CMOS B1 PA alongside a BII GaAs PA (J2C) Credit: Business Wire
For example, the CMOS power amplifiers equalled GaAs efficiency across all power levels up to 22.5dBm, and actually beat GaAs counterparts at lower power levels. Cost-wise, both power amplifiers can be manufactured for less than $0.40.
However, GaAs power amplifiers are significantly smaller than CMOS equivalents, and as Mielke adds:“Achieving efficiency at higher power outputs is still an issue for CMOS power amplifiers, although [the technology] is getting better and that's why Javelin has managed to develop the 3G power amplifiers.”
Efficiency versus integration
The Javelin CMOS 3G power amplifier is based on a mixed-signal architecture, including an input matching network, tunable bandpass filtering - to reduce interference between wireless interfaces such as WiFi and Bluetooth - and an output matching network. The device also integrates the necessary circuitry for power regulation, bias and power control.
And this could be the killer for GaAs. While CMOS struggles to maintain efficiency at higher power outputs, the potential to, say, integrate the power amplifier and transceiver in a single chip may yet be the biggest draw for designers who could even choose to trade GaAs efficiency for CMOS integration.
“You can do much higher levels of integration in CMOS and maybe someday all components will be on one device,” comments Mielke. “That will never happen with GaAs.”
But Mielke will not commit to a silicon take-over yet. First he's closely watching how the industry responds to the imminent launch of a CMOS power amplifier from US-based wireless chip-set leader, Qualcomm.
He believes that if the company's amplifier is a success, alongside the existing wins from Javelin and Black Sand Technologies, CMOS power amplifiers can expect single digit market share numbers next year, with GaAs maintaining at least 90% of the market.
“But if this [CMOS] market is as successful as it looks like it could be, these figures could almost flip flop, so in the next five years, CMOS could have an 80% market share with GaAs down to 20%,” he adds. “And I would expect this trend to carry onto 4G”
Past experience indicates that once a CMOS solution emerges to replace a rival compound semiconductor technology, industry has tended to move en mass to the silicon implementation. Will history repeat itself?
As Mielke points out, CMOS has been trying to make its mark in power amplifiers for a long time and still hasn't quite made it.
“But then I look at the growth of components in the RF area as well as antenna switches and almost everyone has transitioned to CMOS,” he adds. “I can see it happening in the power amplifier sector too. I think it's for real this time.”
