Switching On To New Technologies
Which is the single most important technology currently in development at RFMD?
Of course, I would say they re all important, so that s a hard question. I think our MEMS technology has the best potential to revolutionize several aspects of our next-generation systems, not just the RF switch.
There are several ways that MEMS could change the way front-ends work. It could reduce all of the cellular power amplifiers (PAs) that might be in a multimode phone down to one PA with a load switch. The MEMS switches could allow one PA to be frequency-switched between bands, then switch between a linear mode and a saturated mode, allow for load changes and retain the optimum performance from the PA.
They can also allow other types of function. We ve talked about the potential for adding resonators or sensors, such as accelerometers and gyroscopes, in the same technology that we re developing for RF switching. These functions can be implemented in a very low cost, incremental fashion.
When will we see the MEMS switches on the market and in what format?
The first implementation is going to be on CMOS because our initial concept is to put it directly on the driver circuitry for the switch. It will be a low-cost driver that supports the MEMS switch, which can then be applied in a module with our PAs or with other circuitry. Longer term this process can be integrated directly on GaAs as well, so it s really fairly flexible. It s still in development but we expect to release the MEMS technology for production next year.
So, will MEMS be replacing PHEMT switches?
We believe it would be possible, yes. Initially we are focused on mode switching [between bands], which is currently done with PHEMT and is a slower process with fewer switching cycles than transmit-receive (TR) switching. We believe the MEMS technology that we re developing will have the lifetime to support TR switching as well, so in the longer term we could replace the PHEMT switch for both.
What does this mean for the long-term future of GaAs at the company?
We see opportunities in the mobile phone area and other areas that we re addressing for compound semiconductors and silicon CMOS-based circuits for a long time. We don t see a transition from one to the other. We re beginning to pick up more CMOS, doing development and applying more of our efforts in silicon for various applications. That s not transitioning from our core competency in GaAs. You can see that obviously we re committed to our GaAs because of the capacity that s coming with the Filtronic deal.
Why has RFMD been slower than other companies to use BiFET technologies?
We re working on BiFETs at the moment and we re well aware that our competitors have these on the market. We are really, really focused on what our customers needs are. We felt that the technologies we had with HBT and PHEMT as separate die met the needs of our customers very well. We can see some applications as we move forward that will need BiFET technology, so we re developing it, but we didn t feel that it was something we had to have years ago, when they first appeared.
There are some applications that can benefit from on-board control logic and bypass switching, so these types of applications are interesting. But full integration of a PA and a TR switch with decoding logic onto a single HBT or BiFET die? We re still not convinced about that direction for the transmit modules. We re going to have BiFET technology in some products, but as far as I m aware we don t have a definitive plan to offer products that are fully integrated in that way, at least in the very near term.
Is RFMD really moving into photovoltaics?
We re actually quite excited about our opportunity in concentrator photovoltaics. From my perspective, this technology will have the biggest benefit to the use of photovoltaics in the end. We re actively pursuing that business area and I believe we have a good opportunity to have a significant impact on the availability of these high-efficiency concentrator photovoltaic cells. The way that we can address that market is by providing very large volumes at the lowest cost.
Would RFMD s focus on MBE be a problem?
We agree that MBE is not widely used. We re evaluating it to make photovoltaic cells, but we re not depending on that. Whether MBE is the technology we use, or MOCVD, it s not going to affect our entry into the business.