For years, crystal and ceramic resonators have been the dominant components used for frequency generation in mobile handsets and other electronics equipment. However, new silicon-based products and technologies are challenging the hegemony of the crystal and ceramic parts in the resonator market.
There are two silicon-based products in particular that are giving crystals and ceramic resonators a run for their money. These include Discera's microelectromechanical systems (MEMS) resonators that are designed as alternatives to crystal resonators, and Micro Oscillator Inc.'s all-silicon, CMOS-based clock oscillators, which can be used to replace ceramic resonators.
MEMS resonators move to replace crystal counterparts
A MEMS resonator is a tiny vibrating beam that is fabricated on a 1 mm x 1 mm silicon die, and are 10 micrometers thick. These devices can achieve frequencies of up to 100 MHz with stabilities and other technical parameters that are comparable to or exceed those of crystals.
Like crystals, MEMS resonators must be hermetically sealed. Packaging advances now allow MEMS resonators to be vacuum-sealed in wafer-level packaging.
The key advantage of MEMS resonators is the integration possibilities they offer with silicon ICs. In wafer-level packaging these devices can be attached to other silicon die. As advances in MEMS manufacturing continue, MEMS resonators potentially could be manufactured directly on active ICs.
MEMS technology also allows multiple frequency resonators to be manufactured on one silicon die. This enables the production of a single device that supports multiple frequencies.
In contrast, crystals always have been single-frequency devices. If multiple frequencies are required, many discretely-packaged crystals must be used.
In June 2003, Discera released the MRO-100, a device that incorporates a MEMS resonator and an oscillation IC to form a fully functioning MEMS-based oscillator. The MRO-100 is housed in a 3 mm x 3 mm ceramic package.
Discera's initial target application is mobile handsets. The MRO-100 can serve as a replacement for voltage-controlled temperature-compensated crystal oscillators (VCTCXOs) used in mobile handsets. The initial frequency for the MRO-100 is 19.2 MHz for CMDA handsets.
The MRO-100 can achieve the same performance as a VCTCXO, including frequency stability and phase noise. The advantage of the MRO-100 as a standalone product is its potentially lower price compared to crystals. Because the MRO-100 is made using silicon, it's cheaper to produce than a crystal oscillator.
The biggest threat for crystals and oscillators will come when MEMS resonators are integrated into other silicon ICs.
The trend in mobile handsets is toward more module-based solutions. Crystal and oscillators have always inhibited the development and use of more integrated ICs or module-based solutions. Crystal resonators cannot be integrated into silicon. Furthermore, the size of crystal resonators, along with the space required for hermetic sealing, have prohibited their use in modules.
Companies including Discera are moving quickly to develop more integrated MEMS resonator devices. The capability of the MEMS technology to eliminate the external crystal reference is a benefit that designers will embrace, iSuppli believes. Furthermore, the capability to support multiple frequencies on a single MEMS die is attractive because it eliminates many discrete crystals or oscillators from a design.
Even more compelling is the capability to combine multiple MEMS devices. For example, Discera's product roadmap includes MEMS filters, which would compete with discrete RF SAW filters.
Mobile handsets typically include three or four discrete RF SAW filters. With MEMS resonators, all of those RF filters could be combined onto a single silicon die. The megahertz reference also could be combined, forming a highly integrated device that could be incorporated into a RF transceiver.
The ability to integrate multiple components into one IC or module package will not only reduce the overall cost, but will also save on board real estate, allowing for the addition of more functionality. In addition, with integrated devices, designers of mobile handsets can achieve a quicker time to market, as well as focus more of their time on adding more features into the handsets.
All-silicon clock oscillators target ceramic resonators
On another front, all-silicon clock oscillators are challenging ceramic resonators. These devices produce an output clock without the need for an external frequency reference, such as a crystal. All-silicon clock oscillators have been in the market for a few years.
Companies producing all-silicon clock oscillators include Dallas Semiconductor Corp. (now part of Maxim Integrated Products Inc.), Linear Technology Corp., and Semtech Corp. These clock oscillators typically are low frequency, i.e. 30 MHz or less. However, Dallas Semiconductor sells a silicon clock oscillator that can operate up to 133 MHz.
An all-silicon clock oscillator offers the obvious benefit of being able to eliminate the need for an external frequency reference, such as a crystal. The key advantage is integration; since the clock oscillator is an all-silicon IC, it can be integrated directly with other silicon ICs.
At this time, the primary limitation of the technology is tolerance. Typical frequency tolerance is ±1.5% or more, which translates into ±10,000 parts per million (ppm). A typical crystal oscillator has a frequency stability of ±100 ppm.
The tolerance of all-silicon clock oscillators prevents them from challenging the crystal and oscillator market. However, the all-silicon clock oscillators could pose a threat to ceramic resonators. The tolerance of ceramic resonators ranges from ±0.5% to ±1.5%.
The all-silicon clock oscillators that have been in the market have never challenged ceramic resonators due to several drawbacks including ease of use and price.
The clock oscillators have not been plug-and-play. For example, the Dallas Semiconductor clock oscillator must be programmed on the board, and the Linear Technology clock oscillator uses an expensive resistor to set the frequency.
All of the silicon clock oscillators have been priced at $1 or more, making them too expensive to compete with ceramic resonators, which typically are priced from 10 cents to 40 cents.
However, Micro Oscillator Inc. (MOI) has developed a product that can compete directly with ceramic resonators. MOI's MOI-2000 can operate up to 20 MHz with a total tolerance of ±0.5%. The MOI-2000 is plug-and-play and only requires power and ground. The device is priced at less than 50 cents, which is a cost-competitive level.
The MOI-2000 offers the reality of integration with other silicon ICs. At one level, the device can and has been offered as a bare die that can be die attached to other silicon ICs. The smallest ceramic resonators today are still 0.5 mm in height or greater, which is prohibitive in certain applications where component size and height clearance is critical, such as smart cards.
The MOI-2000 is an all-silicon, CMOS-based IC that can be integrated into other silicon, CMOS ICs such as microcontrollers. A microcontroller with an integrated clock oscillator offers design and performance advantages over a microcontroller design that uses a discrete ceramic resonator for clock generation.
iSuppli predicts that MEMS resonators will have a significant impact on the crystal and oscillator market, while all-silicon clock oscillators will eat into the ceramic resonator market.
Scott Smyser is iSuppli Corp.'s senior analyst for frequency control, RF, and wireless products. Contact him at firstname.lastname@example.org