Since Freescale Semiconductor was divested from Motorola in 2004, the RF Division has made significant changes in both the markets it serves and the technologies it is developing. The RF Division is making continuing advances in LDMOS, its core technology, while also pursuing GaAs and GaN technologies to satisfy the needs of emerging applications. The division has also entered the general-purpose amplifier market and is reentering the industrial market. They are also applying their core technologies toward the WiMAX market and are gaining significant market acceptance. Microwaves & RF spoke with Gavin Woods, vice president and general manager of the RF Division, about technologies and markets and how the company plans to pursue them.
MRF: What is the middle and long-term road map for LDMOS, from the perspective of technological achievement, such as power level, efficiency, and increasing frequency? Compare it to its compound semiconductor alternatives?
Woods: Our roadmap for LDMOS is as always to improve intrinsic die performance in key areas such as efficiency and gain. We also are continuing to focus on cost-effective device packaging, which can be as important as performance. As our customers use new design approaches to ensure savings in operating costs and capital expenditures, Freescale will focus on optimizing our devices to meet their requirements. As for the long-term outlook for LDMOS, we're convinced that it will remain the technology of choice in many RF markets for years to come. In the high-volume RF markets, LDMOS has a significant lead compared to mature compound semiconductor technologies such as GaAs and emerging technologies like GaN.
MRF: Is maximum operating frequency the primary limiting factor of LDMOS?
Woods: LDMOS currently has limitations when it comes to performance at high frequency, and by this I mean higher than 4 GHz. However, If you look back about 10 years, the prevailing opinion was that LDMOS would be fine at 1 GHz and below, but that was about it. However, continuous device engineering improvements have pushed LDMOS operating frequencies up to over four times that frequency, while continuing to improve performance. As an example, in January we announced LDMOS power devices that can meet WiMAX requirements at 3.8 GHz. WiMAX places very stringent demands on the linearity and efficiency of power devices, and our new devices meet them as well as any device on the market. They're also significantly less expensive than, for example, a comparable GaAs device. So we have proven we can continue to extend LDMOS to higher frequencies, and our device designers tell me there is no reason why we cannot continue to extend LDMOS upward in frequency while still maintaining good performance.
MRF: There has been much talk and increasing research into the potential of gallium nitride (GaN) for use in RF power devices. Is Freescale developing this technology?
Woods: Yes, we are investing in GaN for potential future applications. Our position is that we will be the leader in RF power regardless of the die technology required to keep us there. The type of device that makes the most sense in a given application, that delivers the best cost/performance tradeoff for our customers, will be the one we will offer in that market. Right now LDMOS is a primary focus because it offers the best cost/performance tradeoff. If GaN or other technologies help us increase our lead in RF power markets and help our customers design and develop more advanced products, then we will merchandise those technologies in products.
MRF: In what applications do you see GaN becoming competitive?
Woods: The major advantages of GaN for generating RF power come from its high saturation current density, high breakdown voltage, and very high power density per unit gate width compared to competing technologies. These characteristics make GaN capable of efficiently delivering very high power levels over very broad bandwidths. It has the potential to be useful through about 50 GHz, so communications, radar, satcom, EW, and digital microwave radio are likely applications. The biggest challenge has been in growing a high-quality GaN epitaxial layer on a silicon or siliconcarbide substrate. The lattice constant mismatch between the epitaxial layer and the substrate leads to defects that could impact quality and reliability. Obviously, this is the subject of intense development at Freescale and elsewhere. Once the quality and reliability unknowns are sorted out, the volume of the market for GaN will determine its cost. If the big applications are there, the costs will come down.
MRF: WiMAX has been touted as a revolutionary development, complementing or even displacing existing wirelesscommunications technologies, replacing microwave links, T1 lines, and free-space optics in cellular backhaul, and providing an alternative to DSL, cable, and fiber for delivery of residential (and possibly commercial) broadband Internet access. What is Freescale's position on WiMAX, and how is the company addressing this opportunity?
Woods: We are developing devices for WiMAX at 2.3, 2.5, and 3.5 GHz, with the assumption that these markets will develop. We'll expand our offerings to 5 GHz if this market develops. We're moving forward like it's the next big thing and we will only stop when our customers tell us to. Six to nine months ago when we visited customers, they weren't very interested in WiMAX. Now, we can't leave without a comprehensive review of what we are doing, where we see the market going, and what we're developing. So obviously momentum is building. We certainly believe WiMAX will be a significant player in the fixed-access application, where we already see activity. Whether it will become a truly high-volume market depends almost exclusively on whether it can offer true mobility as defined in the IEEE 802.16e standard. If WiMAX can do this, the second hurdle is whether there is a business case for deploying the WiMAX infrastructure, and how it will fare against the next generation of CDMA2000 with EV-DO, and DSDPA and HSUPA for WCDMA. There will be some nomadic deployments, which let you walk through a park with your WiMAX-enabled device, but the key to its true potential is full mobility. If it passes this test, the next few years could be very interesting for everyone in the wireless industry.
MRF: In terms of devices produced, the "gain block" seems to be one of the larger markets for RF devices. Since Freescale entered this crowded arena, how have its products been received? What are the challenges ahead?
Woods: We have had a very good response to our general-purpose amplifiers. It's obviously not a trivial task to beat out well-entrenched competitors who have been in this business for years. However, we got into GPAs because our customers asked us to. They wanted Freescale's reliability, quality, and highvolume manufacturing ability. We also have designed some improvements in the devices that we feel give us a competitive advantage.
MRF: Little media attention is paid to industrial applications of RF power devices, such as MRI, scientific systems, heating, and others operating in the ISM bands. This is no doubt because the devices are deeply embedded in products that don't immediately bring RF to mind. The applications are also scattered over a wide array of disciplines. The industrial market nevertheless appears to be a substantial one. Can you address Freescale's activities in developing the industrial market for RF power devices?
Woods: We're getting back into the industrial market in a big way. Six years ago we had a significant presence in the industrial RF power market, but we sold the business to focus on cellular infrastructure. Now we are reentering the industrial market and bringing to it all the technology we have developed for cellular infrastructure over the yearsLDMOS and overmolded plastic packagesthat will allow us to deliver products with significantly better performance and lower cost than current products from other vendors.
MRF: How do you view the current industrial market for RF power devices?
Woods: There hasn't been a lot of innovation in the industrial RF power market for 10 or 15 years except for a few minor thermal improvements. Other than year-on-year incremental decreases, there have been no major cost reductions either. We are going to introduce step-function improvements in performance and decreases in pricing. LDMOS technology and overmolded plastic packaging give us that ability.
MRF: Can you give some examples?
Woods: Well, on our flagship industrial device, the MRF6V2150N, we are getting significantly higher gain and efficiency than what is currently available. So, for example, the size of the drivers you need will be much smaller because of the amount of gain you get with the MRF6V2150N. This has a very positive impact on cost. The unit cost will also be significantly lower than what is currently being offered.