Everybody seems to be using their mobile phones for their business and social lives these days, so each new model brings increasingly more features to our portable devices. The problem is that all these new uses bring an explosion of frequency bands that must be packed into one portable device. This can lead to dropped calls and slower web browsing, frustrating users. The answer, according to San Jose, Calif.-based Cavendish Kinetics, is the company's tunable RF MEMS technology that provides 3G and 4G networks with improved RF communication. The company has raised a $10 million financing round with Qualcomm and other investors. We recently caught up with Dennis Yost, Cavendish Kinetics' CEO, to learn more.
MEMS Journal: We have all experienced annoying dropped calls and snail-paced Web browsing on our mobile phones, and there are many different explanations for them. Which of these areas does Cavendish Kinetics address?
Dennis Yost: The quality of the RF signal has been largely ignored in favor of larger screens, thinner form factors, expanded frequency bands, and cool software apps. The fact is that since about 2001, the ratio of "theoretical to realized" RF signal quality has been declining by about 1 dB/yr. That may not sound like a lot, but remember 3 dB is a factor of two which is huge.
What Cavendish can do is to recover a lot of that lost signal by making the RF chain a lot more efficient over the existing and new frequency bands being used. Our technology allows one element of the RF chain (an antenna) to be about twice as efficient when compared to current designs when using the new LTE bands (B17, B38, B40) for example. This results in twice the signal to noise ratio for the receiver which enables more of the potential of 4G LTE to be seen by the end user. The user sees much faster data rates and longer battery life.
MEMS Journal: Can you explain what the wireless performance, size, and cost tradeoffs are and how you are addressing the issue?
Yost: That’s an excellent question. Let’s take the example of antennas in a modern smart phone. The trend is to make the smart phone a large screen with a few buttons on the side. This, unfortunately, leaves little room for an antenna. Remember those telescoping antennas that you pulled out when you wanted to make a call? There was a very good reason for that: RF needs a radiator and a significant area to capture the incoming signal. When you shrink that area to less than 5% of what it used to be, something’s got to give.
So what gives is antenna bandwidth – which means that a single antenna can only be tuned to a very narrow set of frequencies or cover a limited number of bands. But LTE requires even wider frequencies with new bands being allocated.
So the solution is to allow the narrow-band antenna to change its fundamental resonance by tuning it. Cavendish is the first company to produce parts that are small enough with high enough efficiency to allow antennas to be tuned to the new frequencies without incurring additional losses. Our parts are delivered as a bare, bumped die for flip chip mounting and are small enough (<2 mm2) to be easily mounted onto the antenna itself or within a multi-chip module.
MEMS Journal: What are some of the main design challenges for tunable RF components and how has Cavendish Kinetics handled them?
Yost: When Cavendish embarked on serving this market, we understood that several specifications were extremely important to RF designers: Quality Factor (Q), capacitance resolution, minimum capacitance value, capacitance tuning range, linearity, and size. The Cavendish solution optimizes the MEMS process, the capacitor design and the overall architecture to come up with optimal performance metrics for all these critical requirements.
In the process of developing our technology and products, we developed many key innovations that enable superior RF performance, extremely small sized device, while still integrating the technology into a standard CMOS process flow using standard wafer fab equipment. Cavendish now has over 40 patents issued and many dozens more in process – all focused on RF MEMS.
MEMS Journal: How does Cavendish Kinetics address the issue of multiple standards in smart phones and how do they impact antenna tuning problems?
Yost: Actually, the more standards proliferate, the more designers need tunable RF. Not just for antenna tuning, but tuning broadband power amps, agile, adjustable filters and dynamic matching circuits. Every major cellular device supplier has a tunable component development program underway right now. And several are using MEMS because of the high efficiency and small size.
Cavendish is partnering with visionaries in the marketplace who understand that tunable RF is not a passing fad, but the next step in cost-conscious, high performance consumer RF products. The overriding standard in the next number of years will be fully tunable RF front-ends in cell phones. That is the standard and vision that we are driving to enable.
MEMS Journal: What is your NanoMech MEMS technology and how does it address RF tuning issues in your products?
Yost: NanoMech is our proprietary processing technology that allows us to manufacture our variable capacitor fully integrated in a CMOS process flow and in a standard CMOS foundry. As I mentioned above, our NanoMech MEMS technology has been developed to deliver superior performing RF solution and give our customers a wide degree of freedom in their design space.
Our NanoMech technology is 100% CMOS compatible for simplified foundry sourcing. Because our MEMS device uses CMOS interconnect processes (as opposed to silicon as a construction material), we can deliver products with a very low ESR (Equivalent Series Resistance) which results in extremely high RF Q (Quality Factor). Also, our devices are specifically designed to be able to handle the power levels used in cell phone applications (> 4 Watts).
MEMS Journal: What are some of the initial tunable RF applications?
Dennis Yost: We see two initial aplications -- antenna tuning in the aperture and impedance matching. Antenna tuning in the aperture is a method of changing the fundamental resonant frequency of an antenna by altering its electrical load or physical configuration. This technique preserves the efficiency of the antenna to radiate more of the power that is presented to it.
Impedance matching is a method of matching the load and source impedances so that the power presented gets passed on instead of reflected back to the source. The example with an antenna is to match the impedance of the output of the power amplifier to the input of the antenna. Impedance of the antenna changes with the antenna environment (such as holding a cell phone up to your head) and can increase the mismatch and cause less power to be transmitted.
MEMS Journal: Who are your main competitors and what is the key competitive advantage for Cavendish Kinetics?
Yost: Our real competition is the perception of the reliability and manufacturability of RF MEMS. There have been some false starts that have called MEMS reliability into question. All RF designers that have heard of RF MEMS clearly know the superior performance and capability that it could offer. RF MEMS has been around for nearly 30 years in universities, government labs, mostly targeted at defense and aerospace applications. The results have been very high costs, or poor quality, or poor reliability.
The key breakthrough and advantage that Cavendish is delivering is a technology specifically designed for the high performance, high reliability, and high volume consumer market for cell phones. We designed our technology and products with the end in mind: make sure we can support the market needs at the performance, size, reliability and costs that are needed.
Our product qualification is nearly completed. We have developed long-term reliability models and clearly understand and can predict our wearout mechanisms – just like the semiconductor industry has relied on for decades. So Cavendish will be shipping high-reliability devices with predictable, guaranteed lifetime performance.
MEMS Journal: Are there competing MEMS and non-MEMS technologies in the space and how are they different?
Yost: There are various approaches that are available to improve the RF performance of a cell phone, but many of them lack in either performance or are too costly. MEMS in general has the ability to deliver the needed performance, but until recently was too big, too expensive and not reliable enough for the cell phone market. We have solved those issues.
Other technologies are being evaluated, but are not able to meet the needed performance. Those are based upon multi throw switches connected to fixed capacitor banks. These can either be discrete solutions or integrated. The integrated approach does require advanced substrates – either SOI or sapphire. The problem is the equivalent series resistance of the switch banks that significantly impacts the RF performance. The Cavendish solution is like having a 1P32T lossless switch connected to 32 different, high-precision, high-Q capacitors.
MEMS Journal: Are you sampling to customers now? When do you plan to ship?
Yost: We have been sampling to selected strategic customers since the beginning of 2012. A broad product announcement will happen before the end of this year and we expect to be ramping mass production in early 2013.
There is a lot of pent-up demand for our products, so we will have proven, high yielding, large-scale manufacturing in place when we make the announcement. Too many customers over the past 30+ years have already been disappointed in premature announcements of product availability (both MEMS and other technologies) and Cavendish is committed to keeping promises, and focused on not disappointing customers.
MEMS Journal: Where is the cost of your product and how does it compare to other companies?
Yost: The Cavendish technology produces the smallest, highest performance tuning solution on the market and is very competitively priced. We believe the value we are delivering to our customers and their customers are clearly recognized. Agreements with major manufacturers preclude disclosing actual prices, but our customers are very happy with our technology price/performance.
MEMS Journal: You recently closed a $10 million financing round with Qualcomm and other investors. How is this new funding being used?
Yost: Cavendish is in the product and process qualification phase right now. So most of our efforts are getting our supply chain fully lined up with backup plans to prepare for the fairly steep production ramp our customers are forecasting.
MEMS Journal: What are your future plans and expected revenue and market share?
Yost: Cavendish has a full line of tuning products in the pipeline that go beyond cell phone RF tuning. Within five years you can expect to see the 60GHz market to become mainstream with beam-steered, adaptable antennas and smart RF networks that adapt to the local environment to select the best band/channel combination for the task at hand. WiFi, Bluetooth, LTE, LTE-advanced, 60GHz and other standards and technologies will be used in heterogeneous networks – all enabled by smart, adaptable RF front-ends.
Cavendish sees this as multibillion-dollar opportunity with our technology poised to take a lion’s share of the market. Wireless communications, meaning in our case RF components, is only at the very early stages of the mass adoption that it has the potential to deliver. This is very exciting.
MEMS Journal: When we interviewed your then-CEO Mike Beunder back in 2005, Cavendish Kinetics was billed as primarily a MEMS memory company. How has the company evolved to where you are now a supplier of tunable RF components?
Yost: Seven years is a couple of lifetimes in this business! We recognized that the technology had a lot more potential than simply non-volatile memory, and that the differentiation in other markets (such as RF MEMS) was a lot greater. The fundamental NanoMech process technology developed for memory turned out to be ideally suited for RF MEMS. Our core NanoMech technology was developed to enable superior performing electrical components, as opposed to more traditional MEMS technologies used to sense the outside world.
We ended up modifying our fundamental memory cells to give the characteristics and performance required in RF instead of trying to shrink them as you would normally do in memory. So, in many ways, we had a more straightforward task of adapting the technology into the RF MEMS space than others.
This article is a part of MEMS Journal's ongoing market research project in the area tunable RF MEMS and non-MEMS technologies. If you would like to receive our comprehensive market research report on this topic, please contact Dr. Mike Pinelis at firstname.lastname@example.org for more information about rates and report contents.
Copyright 2012 MEMS Journal, Inc.