We often get this question from customers and it’s worth looking into. For almost 20 years, MEMS resonator-based oscillators have existed amidst promises of being a viable, disruptive replacement for quartz-based oscillators. More than a dozen companies have started the development of MEMS resonators over the years and only two companies have successfully survived as a major supplier during this period. 

There are many claims about improved vibration sensitivity, lower manufacturing costs, and improved reliability when using a MEMS resonator. Once you begin to look into these areas, however, the data sometimes fails to present real world conditions. As an example, a meantime between failure (MTBF) rate of 130,000 years for MEMS-based devices versus 30,000 years for quartz-based solutions should not be a major concern for any designer. But, when presented as a 4 to 1 improvement in meantime between failure, this creates a different impression in a designer’s mind without adding any real advantage to the long-term reliability of the part. Designers would consider 30,000 years as an acceptable MTBF level making this claim moot.

One of the key features of MEMS resonators is vibration sensitivity (sometimes called gravitational sensitivity or vibration induced phase excursion). This was recognized around 2001 and has been marketed as a key disruptive feature of the resonator. Once we begin to look at the data, though, it is fair to begin to question whether there are any real advantages. Consider the following:

• Typical vibration range is specified from the sub-Hz levels up to 2 kHz 

• Vibration density levels stop or greatly drop off above 2 kHz in most applications

• Phase jitter typically has an integration range of 12 kHz and 20 MHz based on the phase noise measurement

• This range is six times higher than the maximum vibration level that any customer would specify

• Vibration sensitivity units of measure are parts per billion per g of vibration (ppb/g)

• MEMS vibration sensitivity ranges from 0.01 ppg/g to 1 ppb/g based on resonator orientation

• Quartz vibration sensitivity ranges from 0.1 ppb/g to 1 ppb/g based on resonator orientation

• Quartz manufacturers have greatly improved vibration sensitivity over the last 20 years to those of MEMS levels

• Changes in blank size, higher frequency applications, and increased precision in blank alignment being the main improvements 

• These changes have greatly improved vibration sensitivity and  shock sensitivity in quartz-based oscillators

• The Renesas ProXO series of XF and XP field programmable clock oscillators uses these quartz process improvements to obtain low shock and vibration sensitivity levels

• Vibration-induced phase excursion (increased phase noise levels) is mutually exclusive from temperature and network-induced phase noise

• Vibration-induced noise is rms additive to noise induced by temperature and network

• Until the level of the vibration-induced noise is equal to or higher than that induced by temperature and network noise, vibration-induced noise can be ignored

If we look at the typical phase noise levels of both a MEMS-based oscillator and a quartz-based oscillator at 156.25 MHz at a vibration level of 10 gs we see:

• Under static conditions, the quartz-based oscillator typically has 40 db lower noise than its MEMS counterpart

• With a high continuous vibration level, the vibration sensitivity of the quartz-based oscillator is equal to the static vibration levels of a MEMS-based oscillator

• Since the static and dynamic levels are almost equal for the MEMs based oscillator, there is very little increase in its phase noise level in this range

When a continuous vibration at 10 gs occurs over the vibration range of 10 Hz to 2 kHz, the phase noise levels of quartz-based and MEMS-based oscillators are equal. But, a continuous 10 g vibration level over this frequency range will result in decreased long-term reliability and breakage of the product due to the increased weight of the printed circuit board and other larger and heavier components within the system. Assuming that 10 gs will only occur a ***all percent of the time, the quartz-based oscillator will have lower phase noise performance, making it the better solution for real world conditions.

In John Vig’s tutorial on Quartz Crystal Resonators and Oscillators (July 2016), he outlines the effect of vibration sensitivity in detail. From his tutorial, we can use the following to look at the effect of vibration sensitivity (Slide 167, section 4-73)

With the data reviewed here, we see there are no real advantages to using a MEMS resonator oscillator versus the more common and stable quartz-based oscillator.  As such, Renesas dropped the product from our portfolio several years ago and has continued to focus on quartz resonators and improving the devices we offer that incorporate internal resonators, including the new ProXO family of programmable clock oscillators.