Interview with TriQuint's Ralph Quinsey
Ralph Quinsey joined TriQuint in July 2002 as President and Chief Executive Officer. From September 1999 to January 2002, he was with ON Semiconductor, a manufacturer of semiconductors for a wide array of applications, as Vice President and General Manager of the Analog Division. Prior to that, Quinsey was with Motorola, a manufacturer of semiconductors and communications equipment, from 1979 to September 1999, holding various positions including Vice President and General Manager of the RF/IF Circuits Division, which developed both silicon and gallium arsenide technologies for wireless phone applications. He received his BS degree in electrical engineering from Marquette University.
MWJ: I thought we could start talking about how TriQuint sees the general mobile device market, how you break it down and perhaps see it differently than other vendors might?
Quinsey: TriQuint is very excited about the rising tide of the mobile devices industry, primarily led by the advent of 3G/4G smartphones.
Typically with high-end smartphones, there are multiple data bands plus WiFi. Supporting the data and WiFi bands significantly increases the RF content, in the 4 to 6 times range, in a high-end smartphone. The software development of applications for these smartphones and the expanding RF content needed to support the increase in data, has created good unit growth for smartphones and their vendors. Various industry analysts suggest smarthphone growth of 30 to 40 percent this coming year. I refer to this growth as wave #1.
I think of wave #2 as the growth of low-end smartphones. There will likely be 500 to 600 million low-end, voice-only 2G dual band GSM phones sold this year. I believe over the next few years, virtually all of those are going to transition to low-end smartphones. They will be different than the high-end smartphones and contain less RF content but still have a minimum of 2x the RF content of voice-only phones. I believe this because I think mobile access to the internet is THE killer application. We use to think that the emerging world would access the internet through low cost computers. We no longer think that way. We think most people’s first experience with the internet will be with a smartphone, because smarphones offer a significantly lower cost of entry than developing the alternative infrastructure. This killer application [the internet] creates another wave of lift for the RF suppliers.
The 3rd Wave is machine to machine communication, which includes devices like e-readers, remote health sensors, power grid management, and security surveillance. These applications will utilize connected devices for many of the same reasons that mobile communications are so successful today, that is, low cost, ease of installation, and the ability to be mobile. I think much of the m2m communications will fit under the 3G/4G and beyond grid. Over the next decade, I see billions of nodes out there, eventually outstripping the demand for person to person communication devices.
To summarize, wave #1 refers to high-end smart phones, wave #2 refers to low-end smart phones, and wave #3 refers to machine to machine communications. These waves will be very strong markets for as far as I can see into the future.
MWJ: Do you see these waves evolving concurrently or maybe staggered a little bit and how does that impact engineering and your approach to product development?
Quinsey: I see these waves overlapping. TriQuint focuses on advancing the core technologies and then customizing the building blocks into RF solutions for each application, be it a smartphone or surviellance system. We are strategically focused on the complete RF solution. In the simplest form that means we sell highly specialized filters called duplexers, and we sell power amplifiers, which are GaAs devices. We also sell switches. All three elements – filtering, amplification, and switching – comprise the major components to any RF front-end. We can sell standalone components, but our real value comes when we integrate the components into one tiny module - simplifying the RF portion for a mobile device vendor. We take the signal from the transceiver, amplify it up and send it over great distances, mostly on the transmit side. That is the nature of our business. Innovations from the core technologies needed to produce these three major components and integrated modules influence product design across the company.
As you transition from 2G to 3G - from voice to data, duplexers become very important because it becomes a frequency domain duplex market as opposed to a time domain based on the switches. The market for 3G and 4G duplexers is going to be billions of dollars in the future. It’s a new market that we participate in that our primary competitors don’t participate in first-hand. They buy duplexers and integrate them into modules. I think TriQuint has a competitive advantage because we custom design our building blocks for each integrated module utilizing in-house technologies, such as SAW, TC-SAW [temperature compensated] and BAW. We have great switching and power technology that allows us to create long-term roadmaps, focused on integration that addresses performance [measured in battery life], size and cost. We think owning the underlying technology gives us an advantage at squeezing out a little extra electrical performance while cutting down the size and cost.
MWJ: The other vendors in our article take an approach of buying their filter technology from leading sources, down playing any advantage to having this technology in-house. Would you care to comment on that?
Quinsey: The other companies in your article are great competitors; they’ve taken a different approach and have a different opinion. We feel that it is a definite strategic advantage, particularly going forward, to have the technology in-house as opposed to buying filter and duplexer technology from another source. Our approach enables us to remove a lot of the back-end packaging out of the components with wafer-level packaging. By utilizing flip chip, we eliminate die attach and wire bonding. This gives us a competitive advantage not only in size but also in production cost. The result is really elegant small modules and back-end assembly line procedures that eliminate steps. We’re not completely there yet but we’ve got our roadmap. That’s our approach.
MWJ: In my conversations with Eric Creviston (RFMD) and Greg Waters (Skyworks) about developing products for the handset and mobile device markets, comments were made regarding past market growth, which had been choppy and R&D was niche focused with low to moderate design complexity, and that there were many competitors. Moving forward they both expect market growth will be high and R&D scale will shift from niche focus to high and broad based. Has TriQuint re-organized its approach to R&D and engineering to align itself with these trends and accommodate these three large waves as you put it and if so how?
Quinsey: What is compelling about our product development strategy is that we are keenly focused on a roadmap that improves the performance or lowers the cost of our customers’ applications. That is the goal for every generation of products we develop. We are an innovator; it is how we got into this business. RFMD and Skyworks used to look a lot different. They look more like us now. In these markets, we are 70ish % their size- we used to be a fraction. They both have strategies that have taken them into the domain of Si that didn’t turn out successfully.
Back in the 2G GSM/GPRS heyday (around 2005), we went to our customers and said ‘you need transmit modules’. Transmit modules offer a higher level of integration over a PA module. And our customers looked at us and said ‘are you crazy? We’ve got lots of suppliers selling us GSM PA modules, we don’t need any more’. And our sell was that they could get on a different performance and cost slope if they integrated in this fashion. Sometimes it was hard to talk them into it because the volume and full benefits weren’t there in the first generation. But long story short, we did convince one of our major customers to convert and virtually the whole world now has converted to transmit modules for 2G. You can’t build a solution on discrete devices as inexpensively and you can’t get the performance. That all happened.
I believe the same thing is going to happen for 3G with PAs and duplexers and then other forms of duplexers, which are multimode and then beyond. We could get into a lengthy discussion about the “then beyond” technology, but as far as your question goes that has been our approach to product development from day one. We don’t just throw devices at the market – we have compelling solutions that improve the performance or lower the cost of our customers’ applications. That’s what drives us.
MWJ: With regard to architecture, you mentioned discretes and moving your customers toward integrated solutions, does that mean hybridized modules and different architectures that try an alternative approach to stacking PAs with a transmit or duplexer module on the board?
Quinsey: I look at architectures in generations as well. There is a 3 x 3mm PA module generation for WCDMA or stand-alone power amplifier modules, which aren’t really used anymore for GSM. The next generation is the PA duplexers combination. This module is actually a complete line up of bands. Our first generation of PA duplexers were 8 mils by 5 mils, and had detectors, couplers, inter-stage filters, duplexer, amplifiers - a fairly complex FEM. The current generation, 7 mils by 4 mils, has more integration, and a complete transmit line-up. Our roadmap going forward has some very compelling products using technologies like flip-chip and wafer-level packaging to offer higher performance with smaller size and improved cost structure. That is what I call the PA duplexer generation for 3G.
The next generation goes by different names -- sometimes referred to as multimode PAs or ‘two in ones’, and puts multiple line-ups in the same package. But fundamentally the guts are similar inside. And then after that, we will transition into a converged solution that will handle a large amount of bands, possibly 5 bands and up. This architecture will be used in the very highest end world phones that cover multiple standards and will likely represent a very small percentage of the [overall] market.
The challenge for these phones will be providing an integrated solution that doesn’t sacrifice performance, i.e. battery life which is affected by efficiency and losses. A lot of the converged solutions, such as the one from RFMD, have extra switches in the line-up, which adds insertion loss, impacting the PA efficiency and degrading the battery life. I think the market will go to converged solutions for the very high end, but that will be a small part of the market. The bulk of the market is adopting multimode and PA duplexers because it offers a lot of flexibility. The number of bands for the bulk of the market is probably going to average around three. Below that, the low-end may have one data band.
MWJ: Regarding switches: Are you considering alternative technologies other than pHEMT, such SoI as price and size pressures abound?
Quinsey: We are focused on having the best technology and being the best integrators. At the highest level, we are technology agnostic. We just want to have the best technology and we think we do. The switch story line has evolved. It started with Pin diode, but lost popularity in early Quad band phones, mostly due to battery life. That is, in quad band configurations, pin diodes were such a current hog that the pHEMT switch became more popular. Then there is Si on Sapphire, SoI switches and CMOS switching available for some applications. The characteristics of the linearity of the switch, whether it’s a mode switch or a transmit/receive switch and other such characteristics makes it too complex of a situation to give you a generic answer. We are evaluating these other technologies and if we find a better solution, we will certainly go with it. However, there are just too many characteristics to consider. In short, I think pHEMT switches have a good life ahead of them and will carry the bulk of the market.
MWJ: I have a similar question regarding CMOS PAs.
Quinsey: Similar situation. The CMOS PAs do and will service some portions of the market, particularly the very low end of the market. They have made a play, primarily in the grey market in China. It’s a small component relative to the complete solution. It doesn’t have the same leverage over the PA market as it did back in 2004. Many good engineers and companies have worked on those solutions and they have still not broken into a high volume position, primarily because the CMOS PAs lack the capability for performance.
It is my belief that the PA, filter, and switch, from a handset, chip manufacturers position, are “un-integratable”. The PA is a heat generator, different technologies are not commercially viable to integrate a PA into your transceiver because it slows down your progression towards critical dimensions, etc. Then there are a number of technical issues and other major obstacles such as cost. CMOS PAs don’t have the leverage to make a significant impact. In addition, you need to make the CMOS die a little bigger to get acceptable performance. Instead, we’re learning how to make the PA die in GaAs smaller and smaller and less expensive everyday. So CMOS, except for low end, will not be a dominate technology.
MWJ: Yes, there seems to be a consensus that CMOS may co-exist, but that is about as far as it is likely to get in replacing the PA at least. I’d like to just finish with your thoughts about these three waves of market opportunities and how they affect the way TriQuint approaches its engineering. For example, are you developing solutions for the 3G entry phone business with the same R&D group that is going after the e-readers or m2m business? Are there similar considerations between user products in that second and third wave that greatly impacts the RF content such that lessons learned from one aren’t applicable to the other or, are your engineering groups able to share knowledge easily. This is of course assuming different teams work on different targeted application, in which case perhaps we should start with how your R&D is structured around these opportunities.
Quinsey: Last year we did re-structure our company to align better with the customer market. For example, we moved our wireless LAN development from our networks organization into our handsets organization and also redefined the charter of the handsets organization more broadly to mobile devices. Anything requiring mobile connectivity, whether low-end phones, smartphones or m2m communications is going to come out of the Mobile Devices Business Unit. The mobile devices division is targeting a rich release of new devices. Our networks business is now more structured around radio access points, transport, and emerging markets.
We’re thankful to companies like Apple and RIM for making handsets exciting again, for us in the RF community, by adding applications and mobile internet. The dialog has changed from the color and thinness of the phone back to a rich conversation about expanding RF content and RF appliances [m2m]. There are something like 21 defined bands in 3G and another 10 or so in LTE. Not all will be widely used, but those bands represent a lot of RF content and this is a good thing for our industry.
During the GSM period, people used to characterize our industry as a zero sum game. I started out working for Motorola and the transmit board in the brick phone contained north of $50 worth of RF content. By the early 2000’s we worked our way down to $1 of RF content. Now the curve has turned again. With the addition of 3G and the story we’ve been talking about with RF appliances, the RF content goes back up and to the right for some time. It is a good time to be an RF provider.