(iTers News) - Quartz crystal oscillators have been around for almost one century, sitting in every complex and sophisticated device from PCs to mobile phones to digital cameras to tablet PCs to control the timing of the operations of their built-in key parts and components  

Yet, the 100-year old technology has hit a dead-end in reducing manufacturing costs and foot prints, as other components and chip technologies are increasingly getting smaller and cheaper using a low-cost and high volume chip manufacturing technologies.

For example, traditional quartz crystal oscillators are manufactured out of a blank quartz ingot undergoing several steps of labor-intensive manufacturing processes down the road from initial crystal-growing process to expensive ceramic packaging and assembly to the finish, resulting in long, unpredictable lead time as well as high manufacturing costs. The traditional crystal oscillators also are very much susceptible to vibration, shock, abrupt changes in temperature, and aging. 

To break through the technological barrier, the chip makers have muscled in to fabricate MEMS-based oscillators using a CMOS silicon die technology.

“The big step forward was to replace a quartz resonator with a MEMS resonator. The first generation of MEMS resonator was one where MEMS resonator was fabricated with one specialized semiconductor fab and the end-line die is produced on CMOS fab. The total solution was the combination of those two dies in a multichip module format - the resonator die wire-boned to the end-line CMOS die-making an assembly process more complicated. That was a big step forward in changing the oscillator layer. Another big step forward is the full integration of MEMS on top of a CMOS die, so that a monolithic solution can be provided, “said Mike Petrowski, vice-president, general manger, Timing with Silicon Labs.

Resonator on top of CMOS silicon

(Photo caption: Mike Petrowski, vice-president, general manger, Timing with Silicon Labs)

Silicon Labs is one of the chip makers that are leading the industry’s technology developments for MEMS-based oscillator chip solution, seeing huge potentials in MEMS oscillator market.

As recently as June 27, Silicon Labs has debuted a highly integrated one chip MEMS solution – Si50x family -in what the company said is the world’s first monolithic single MEMS oscillator chip solution.

A short for micro-electro mechanical system, MEMS solution is an integration of physical mechanical structure on a single CMOS die. As it is fabricated with a high volume CMOS chip-making process, chip makers can take advantage the semiconductor industry’s economics to produce it in higher volume at lower costs than traditional crystal oscillator makers.

Called after the way that N-type transistors and P-type transistors are complementarily integrated on a single silicon die, the CMOS, or complementary metal oxide semiconductor technology is a very low-power chip processing technology. So, it can allow chip makers to integrate more of components and even firmware in the MEMS chip solution without compromising power consumption. As it is built on a silicon die, it also carries all the beauties of the chip-making technology, including programmability, reliability, and smaller footprint.

Exploiting the technological attributes combined of the MEMS and CMOS processes to their advantage, chip makers are fabricating a wide array of MEMS solutions on a CMOS die, including gyroscopes, accelerometers, proximity sensors, RF tuners, microphones.

Economics of chip industry

Crystal oscillators are no exception. There are already an array of MEMS-based oscillator solutions available in the market, but, according to Silicon Labs, they are not a single die solution, but is a sort of multichip package solution that combines a physical MEMS dies and CMOS chip die on a package-level.

Silicon Labs took one step forward, successfully fabricating one single die monolithic MEMS chip solution.

By combining everything in a single die using its patented CMEMS, or CMOS MEMS, Silicon Labs has achieved many improvements in the performances, power consumption, usability, reliability, and lead time.

“The CMOS MEMS gets all better. As it is the semiconductor manufacturing process just like making chips. You don’t need rock anymore. Rock goes away, quartz goes away. You just manufacture it on standard and high volume CMOS fab and assemble it using an injection molding plastic packaging like every other IC companies and store programmable devices for in very short lead time of less than 2 weeks,’ added vice president Mike Petrowski.

According to Silicon Labs, it usually takes 12 to 16 weeks of lead time to produce first generation of traditional MEMS quartz oscillator, because manufacturers have to grow crystal as a base material for a resonator die and wire-bonded together it with a silicon CMOS die using very expensive ceramic package going all the way through for final testing.

Multi-frequencies    

The technology advantage does stop there. The 4-member Si50x family are all so programmable that they allow users to configure multiple operating frequencies depending on their applications. For example, according Silicon Labs, the three of the 4-member Si50x family –Si501/2/3-can be pin-controlled to operate either on 1 to 2 or 4 user configurable frequencies. In addition, the  Si504 device is a good fit for customers that want more flexibility, because it is user-programmable through one pin interface.

The Si50x family also sports an ability to allow customers to trade off power versus clock jitters. For example, when power is more important, users can select low power option that consume as low as 1.7mA. When the clock jitter is one important specification for the system, they can select slight more power to achieve lower jitter which is as low as 1.1 ps RMS. 

Another easy-to-use feature is a customizable drive strength output that allows customers to control the rise and fall time of clock edge to reduce EMI. It comes into three industry standard packaging options -3.2x4mm 4-pin DFN, 2.5x3.2m 4-pin DFN, and 2x.2.5 mm 4-pin DFN. 

What really sets Silicon Lab’ Si50x family apart from other first generation of MEMS oscillator and quartz crystal technology is its patented CMEMS resonator design technology that utilizes two different materials –silicon dioxide and silicon germanium-to prevent frequency drifts.

‘Traditional MEMS oscillators drift or move frequency as temperature changes by quite a bit. We solved this problem with our patented CMEMS technology,” said vice president Mike Petrowski.

According to him, the two different materials –silicon dioxide and silicon germanium behave differently as temperature changes. For example, silicon germanium gets hard as temperature gets cold, while silicon dioxide gets as temperature falls. As a result, the combination of the two materials can balance the effect of temperature over resonator structure, so that it can minimize the frequency drift. 

Immune to aging and temperature changes 

Take for example traditional MEMS resonators. They show frequency drift of -30 to 40 ppm/ ℃. Meanwhile, Silicon Labs CMEMS Si50x family’s frequency drift is somewhere near 0 ppm/℃, boasting 5 to 10 times tighter temperature stability. The ppm/℃ (parts per million ) is a measure of frequency drift when temperature changes by one degree.  

That means that the Si50x family can operate more reliably and consistently across a wide range of temperature.

“The reason for this is that the compensation circuitry needs not to be complicated, and  as a result, requires less die area, which means lower costs and requires less power,” explained he.

The immunity to shock, vibration and aging is another outstanding attribute of the CMEMS Si50x family, because it has 5 anchor points. Meanwhile, traditional MEMS oscillators have 2 anchor points.

The CMEMS Si50x family is also very immune to abrupt transience in the room temperature, leading to less frequency shifts, which means little changes in frequency, which is another measure of reliability and sustainability.

The aging test shows that its frequency less shift as time goes by, compared with first generation of MEMS oscillators and traditional quartz oscillators.

 

Photos & Videos by JH Bae