Contributing Editor, MEMS Investor Journal
RF MEMS devices are tiny replacements for the dozens of discrete components used in conventional radios and hold the possibility of enabling an entire cell phone to be cast onto a single CMOS chip. Instead of duplicating the circuitry for each band that a cell phone serves, as is done today, RF MEMS chips use arrays of switchable devices to dynamically reconfigure a tunable radio – saving on size, cost and power consumption simultaneously. The goal has been the subject of multiple initiatives in the U.S., Europe and Asia.
Now the first components necessary for realizing a cell phone on-a-chip are rolling off IBM's fab lines in the U.S. and will soon be rolling off Infineon's lines in Europe, according to WiSpry president and founder, Jeff Hilbert. In recent deals struck with IBM in the U.S. and with Infineon in Europe (as a part of the $8 million Smart Antenna Front End, SAFE, project) WiSpry will supply the necessary MEMS technology to enable IBM, Infineon and any other licensees to start serving up tunable RF MEMS components for cell phones.
"On the heels of our IBM foundry partnership, we are already fabricating our first prototypes for SAFE at IBM. We had products in process there that were related to making SAFE's tunable antennas, so rather than take our own products to market, we now have a partner in Infineon for a smoother path to integration and a shorter time to commercialization," said Hilbert.
SAFE aims to make the antenna itself tunable, by installing multiple antenna elements and matching networks that can be dynamically reconfigured using WiSpry's arrays of RF switches and capacitors. Those same switchable capacitor arrays will also be used in the tunable RF front end.
"Our vision has always been to ultimately figure out how to make the entire RF front-end tunable," said Hilbert. "The first parts that we are working on are the ones that are closest to the antenna. That turned out to be a very convenient entry point for us, both because we could add a significant amount of value there, and also because our MEMS components do not require that handset manufacturers make any changes in their existing designs – making it relatively low-risk and high-return, which is something we worked very hard to identify."
WiSpry patented its basic invention for tuning RF circuits – an array of MEMS capacitors and switches which allows any range of capacitance to be “dialed-in” to tune an RF circuit. Today, tunable RF circuits are adjusted and set at the factory, but dynamically tunable circuits could be adjusted on-the-fly, resulting in better, more consistent reception with fewer dropped calls – in addition to the size, cost and power consumption savings.
"Our MEMS sits right on top of the interconnect stack, and then we do a wafer level encapsulation," said Hilbert." Underneath our MEMS capacitor array is a whole CMOS circuit. The functionality of the CMOS really serves two purposes, first to drive and control the MEMS and secondly to interface our chip with the applications processor or the baseband processor. But our intention going forward is to do single die integration where the economics of doing that makes sense. From a roadmap perspective we are talking about a single footprint in the front end where we have groups of frequencies that are binned up together, and each one of those groupings in itself is tunable."
The first MEMS chips, just rolling off IBM's fab lines, will take on the task of tuning the RF front-end circuitry as well as the antenna itself, but eventually will evolve into a multi-chip module that realizes the holy grail – a cell phone on-a-chip.
"We are starting with this first product which basically matches the PA [power amplifier] and the antenna, then we are expanding to look at making the antenna element itself tunable and then from there were are going to move back toward the transceiver by making the filters tunable," said Hilbert. "We will be adding more RF functionality both on the same die and in the same package with several MEMS die. At this point it is not clear that everything will eventually be on a single chip. That may not be the best answer. What we know we would like for sure is to have is a single module on the front-end right next to the transceiver, basically a multi-mode multi-band front-end with a very significantly reduced number of components, because of its tune-ability."
Copyright 2010 MEMS Investor Journal, Inc.