Microwave Journal

Start-up Introduces 3G CMOS PA

May 6, 2010

Bradley J. Fluke has been the President and Chief Executive Officer of Javelin Semiconductor Inc. since September 2008. He has 24 years of experience in marketing and general management in the mixed-signal semiconductor business. Fluke served as Vice President and General Manager of Silicon Laboratories Inc.'s Wireline Products Division since January 1999 and as its Vice President of Marketing from April 1997 to December 1998; he also served as Vice President of Business Development. Previously, he served as the Director of Marketing of the Computer Products Division of Crystal Semiconductor/Cirrus Logic from June 1990 to April 1997. From 1984 to 1990, Fluke served various marketing positions in the Data Converter Group for Analog Devices, a designer and manufacturer of integrated circuits. He holds a BS degree in electrical engineering from the Rochester Institute of Technology.

MWJ: A CMOS PA has been the industry’s Holy Grail for some time. What are some of myths (pros and cons) that you are hearing from potential customers?

BJF: History has shown that implementing a mobile handset PA in CMOS is extremely challenging: numerous companies have attempted it and failed while others fell short on performance. Therefore, there is an understandable perception in the industry that it simply isn’t possible to meet the stringent frequency and voltage swing requirements of a cellular PA that is implemented in CMOS.

However, our customers have ample reason to want a CMOS PA, which is why it has become the “Holy Grail” of the industry. Customers always demand longer talk times on handsets that contain multiple radios which translates into challenging requirements on the power amplifier. Furthermore, GaAs manufacturing limitations can create an allocation shortage for handset manufacturers, which is an unacceptable situation in today’s fast-paced market.

Our customers have enjoyed the benefits of CMOS products in the baseband and transceiver sections of their handsets and are pleased to see a CMOS PA that not only brings the well-known benefits of CMOS but actually exceeds the performance of the PAs they use today.

MWJ: What is your strategy to achieving a 3G mobile handset win? Go after small, medium or large handset manufacturers, partner with other discrete (2G GSM/EDGE) PA device manufacturers?

BJF: Early on, we identified customers who are instrumental in driving the industry and they partnered with us as we developed the technology. These customers are now progressing well with the JAV5001.

Our design process included extensive modeling that allowed us to enjoy first-pass success and avoid some of the trial-and-error cycles that other companies have experienced when trying to develop a CMOS PA. This illustrates the stability of our architecture and gives our customers confidence in our solution. We have tested and characterized the part extensively, and demonstrated superior performance results in mobile handsets with first-pass silicon.

Thanks to the well-known quality and reliability of CMOS, we have a quick path to production, plus guaranteed assurance of capacity to support the manufacturing volumes required by mobile handsets, further strengthening our position with our customer base.

MWJ: What are some of the biggest challenges for the adoption of a Javelin CMOS PA solution among handset and datacard manufacturers? Are handset manufacturers open to working with a relatively new untested company?

BJF: First of all, the members of the Javelin Semiconductor team have long histories within the wireless industry and track records of success with other innovative products in past roles. We are personally familiar with many of the key individuals throughout the industry. Our partners include well-respected, industry-leading sales representatives and semiconductor manufacturers shipping billions of CMOS products every year. So, although Javelin is a new company, the people involved and our partners are well-known and trusted by our customers.

Now that Javelin Semiconductor is sampling our PA, the company is transitioning from a promising technology company to a viable candidate for new designs. Our customers are seeing products from a team that delivers results and this leads them to qualifying us as a new supplier.

MWJ: What does your team feel are the most important factors for success of a 3G PA?

BJF: A 3G power amplifier (PA) has the challenging requirement of maintaining linearity while delivering over a watt of peak power at frequencies up to 2 GHz. There’s no question that performance is the key attribute of a successful 3G PA, regardless of its process technology. High efficiency to extend battery life and low noise to mitigate interference between multiple radios within a handset are two of the most important metrics to measure performance and are at the center of our philosophy when our designers tackled the original JAV5001 architecture design. As a result, we have produced a CMOS 3G PA that features the industry’s lowest average current and lowest noise, exceeding expectations.

Supply assurance is another important factor for handset manufacturers. Too often, handset time-to-market has been slowed when the custom process PAs are on allocation. The manufacturing lines associated with these PAs have not seen much investment or expansion in recent years and now the mobile handset market is back on a strong growth trajectory. The 3G PA volumes are picking up as end users migrate to 3G networks and also because an increasing percentage of 3G handsets ship with multiple PAs. It is only a matter of time before the demand outpaces the supply available from the custom, captive fabrication lines. Handset manufactures understand this.

MWJ: A large number of phones today are multi-mode, multi-band. Is this a solution you will be going after or will you be focused on a discrete 3G PA solution for the foreseeable future?

BJF: Today’s 3G handsets consist of at least one and up to five WCDMA frequency bands plus four frequency bands for legacy 2G communications. Javelin’s PA architecture scales to multiple bands and multiple standards, and we will be leveraging our core technology into a full family of products.

MWJ: How much headway have you made so far toward getting designed into a phone?

BJF: At Mobile World Congress in February, we demonstrated the JAV5001 in working handsets, and we have been sampling the product since December. We are in various stages of the design-in process with customers world-wide and anticipate volume shipments within the year.

MWJ: Patrick (Morgan, Javelin Semiconductor VP of Marketing) mentioned in his product feature that CMOS PAs have been tried before (Silicon Labs and Axiom) but had relatively small success, why?

BJF: Excellent performance is essential for a mobile handset PA. Silicon Labs and Axiom both successfully implemented cellular PAs in CMOS, but neither was able to meet the minimum performance requirements of the 2G handsets they were targeting, let alone exceed the performance of the GaAs PAs that the handset manufacturers were using. The benefits of CMOS alone are not enough to compel a handset designer to consider a new solution; there must be measureable performance benefits as well. However, those prior efforts proved that it is possible to develop a PA that works within the rules of the CMOS technology, for example such as those covering the breakdown voltage, which had previously been doubted.

Javelin started from the ground up and invented a new CMOS PA. By doing this, our designers have avoided some of the pitfalls that affected earlier attempts and in doing so we have created a product that exceeds the performance of the incumbents. Our customers are excited about the benefits of CMOS, but it is the improvement in performance, specifically in efficiency and noise, that compel them to adopt the JAV5001.

MWJ: Supply voltage reduction due to CMOS technology scaling has been a hindrance for CMOS based RF power amplifiers. Do you believe that a set of “standard” techniques are evolving which are helping to erode the barriers silicon has faced when competing against GaAs-based ICs in the handset power amplifier arena, in particular, ways to overcome the CMOS breakdown voltage limitations.

BJF: For those prior CMOS PA attempts which overcame the voltage limits, but fell short on efficiency, we do not expect the techniques used in those approaches to produce a market-ready product.

The approach taken by our team was to completely redesign the PA architecture to work well within the CMOS design rules and to provide superior performance, among other considerations. We have nearly 20 patents on our technology, several of them fundamental to CMOS PA design, and so our IP is well protected. However, I expect that over time, other clever engineers will learn from Javelin’s innovative design techniques and begin to apply similar approaches in CMOS for additional applications.

MWJ: How much of a challenge was it for your design team to address the on-chip passive losses stemming from the conductive substrate used in deep submicron CMOS processes? Did they have a special approach to addressing these losses and achieve the desired performance?

BJF: During the product design process, our engineers delved deeply into comprehensive modeling that allowed them to thoroughly understand the PA architecture and its performance. Certainly the Javelin PA architecture takes a unique approach to avoiding many of the potential performance issues, plus our technology is implemented in a standard CMOS process.

MWJ: A highly efficient, fully integrated power amplifier in CMOS technology has been a real challenge to others in the past and is critical to achieving commercial success with such a device. What can you tell us about Javelin’s approach?

BJF: As I mentioned, Javelin’s engineers truly took a ground-up approach when designing the JAV5001 architecture. They have the benefit of working in an environment where they are unencumbered by earlier design attempts, so they started with a clean sheet of paper.

In particular, the JAV5001 architecture has three key innovations. The first innovation is an integrated power combiner that enables signals to be amplified while the PA consumes very low current. Secondly, the architecture has a bandpass response that minimizes noise contribution outside the frequency band of interest thereby producing the lowest noise in the industry, over 10 dB better than the incumbents. Finally, the JAV5001 is a true single chip design that integrates all functions into the silicon, enabling the product to deliver consistently high performance even over extreme conditions and manufacturing variations.

MWJ: What process node was the JAV5001 designed and manufactured with?

BJF: The JAV5001 uses a standard CMOS process, but we don’t release any more detail than that.

MWJ: Is there an advantage of this process that is particularly well-suited for an RF power amplifier or is the design approach somewhat node-agonistic?

BJF: Our design approach could apply to any process node provided certain basic requirements are met.

MWJ: What are the future foreseeable developments in CMOS PAs?

BJF: The trend today is clearly toward increasing complexity and functionality in handsets. This means supporting multiple bands and multiple frequencies, and including multiple radios to support additional functionality like GPS. Javelin’s PA technology scales to support these evolving needs. The PA is a dominant contributor to noise, so the fact that Javelin’s PA has the lowest noise of any PA in the industry is very significant.

Numerous technologies have experienced wholesale shifts to CMOS once a viable solution is available. We expect that the JAV5001 will be at the forefront of such a shift in cellular PAs.

MWJ: Javelin Semiconductor, Inc was founded in 2007 in Austin, Texas, also the home of Silicon Labs where you served as VP and General Manager of their Wireline products division. How did you come to leave Silicon Labs and join Javelin?

BJF: After I retired from Silicon Labs, I took a little time off and pursued some other interests. David Bockelman and Vishnu Srinivasan, whom I knew well from past roles, approached me with the idea to start a company and tackle new technical challenges. Javelin gave me the opportunity to get back to the early roots of a promising startup, a chance I couldn’t pass up.

MWJ: You’ve been with the company pretty much from its beginning, what are some of the biggest challenges for a high-tech start-up? What are among your priorities for 2010/2011?

BJF: Certainly gaining funding in this tough economic climate was a challenge. High technology ventures, particularly semiconductors, are not viewed as the golden investment opportunity that they once were. However, once our potential investors learned about the caliber of the Javelin team and gained some insight into the technology we were developing and our capital-efficient business model, the funding has proceeded very well.

Since Javelin’s inception we have been very disciplined and been able to execute to our original plan. Now that we have successfully introduced a production-ready product, our priority is to support our customers and their design processes and to leverage our architecture into a full family of PAs.

MWJ: Do you see Javelin eventually going the IPO route or the acquisition route? Which companies in the RF handset PA space do you most admire and why?

BJF: We are focusing on our technology development and our early customers and are not driving toward a specific exit plan. There are several companies we respect for their innovation, integrity and commercial success. However, we believe that our culture, team and methodology are truly unique and very different from any other PA company. From a design perspective, we innovate in the chip; most PA companies innovate in their fab process. Because we have this unrelenting focus on innovation, we are able to develop ground-breaking technology, rather than making incremental improvements. This creates an uncommon energy and passion among our employees that is ultimately key to our success.

MWJ: You have left your role in corporate management for one as an entrepreneur. Is it a dramatic change? What is your favorite upside to this role?

BJF: It’s definitely a change from my most recent years in the corporate world, but I have had the great fortune to help drive the business growth at several world-class companies in their early years, including Crystal Semiconductor and Silicon Labs. In many ways, Javelin is a return to my favorite part of the business. It is exciting and very rewarding to be part of an early venture and see it grow into an international business. In a startup environment, you have the potential to really make a difference: there is immediate impact of each person’s efforts. Javelin presents an exciting opportunity with immeasurable potential, and I am honored to be part of this team.