There are a few common ways of generating a periodic signal. I have another article on choosing between different types of oscillators in the context of microcontroller designs, and it provides a good general overview of the standard options. One type of oscillator that is not mentioned in that article is the MEMS oscillator.
As you probably know, MEMS technology can be used to make a variety of interesting devices, including accelerometers, gyroscopes, microphones, and switches. It turns out that it can also be used for oscillation.
Oscillation à la MEMS
MEMS devices combine a teeny-tiny mechanical element with a signal-processing circuit. In the case of an oscillator, we have a MEMS resonator and a circuit that does various complicated things, as follows:
Diagram courtesy of SiTime.
One of these complicated things involves the excitation circuitry and sustaining circuitry. The MEMS resonator vaguely resembles a tuning fork, so let’s run with that analogy:
The excitation circuit (“charge pump” in the diagram) is the equivalent of knocking the tuning fork on your music stand. The physical impact causes the fork to sing its monotonic song and, likewise, the electrostatic excitation in the oscillator causes the resonator to start vibrating. So far so good, but we all know what a tuning fork does after you knock it—the volume gradually decreases. This is fine if all you need is a starting note for whichever Mozart aria you’re about to perform, but it’s a problem in the context of oscillators that are expected to function consistently for years on end.
Thus, we need a sustaining circuit, which (as the name implies) sustains the oscillatory behavior. It’s as though you devised a mechanical device that would carefully strike the tuning fork so as to perfectly maintain its vibratory action. I don’t pretend to know the details of how this MEMS oscillator thing works, and I realize that the tuning fork analogy is limited. The bottom line here is that the charge-pump and sustaining circuitry work together to turn the MEMS resonator into a stable, enduring source of electrical oscillation.
Temperature affects everything, including all the components in an electronic circuit. Fortunately, we can often ignore temperature, simply because its influence is not significant when a device is always operating at or near room temperature.
However, temperature variations can be particularly problematic for oscillators. It is not uncommon for an application to require fairly high precision and/or stability with respect to a certain clock or signal frequency, and this is not easily obtained when the circuit will be exposed to nontrivial temperature variations.
A standard solution to this challenge is frequency compensation. RF systems often rely on a temperature-compensated oscillator (TCXO), and this same technique can be applied to MEMS technology:
Diagram courtesy of SiTime.
This is the architecture used in SiTime’s very-high-precision oscillators. A temperature sensor is integrated into the device, and then additional processing circuitry automatically adjusts the output frequency to compensate for temperature-induced variations in oscillatory behavior.
The SiT2024B oscillator is described as “ultra-robust,” offering “widest frequency range, tightest stability . . . , and best reliability.” It’s AEC-Q100 compliant and is available with frequencies from 1 MHz to 110 MHz.
Deviation from nominal frequency vs. temperature for twenty SiT2024B devices. Plot taken from the datasheet.
And I don’t think there’s anything complicated about the implementation: give it power and ground and, after a 10 ms (max) startup delay, you have a typical logic-level clock signal at the output pin.
Diagram taken from the datasheet.
The fact is, SiTime claims that MEMS oscillators (or at least their MEMS oscillators) outperform quartz devices to an extent that is almost comical. I’m honestly not casting doubt on their assertions or their specs; my expertise in this subject matter is extremely limited, whereas they are actually designing and manufacturing these components. Reliability, aging, susceptibility to electromagnetic interference, mechanical robustness . . . the “MEMS superiority” list is rather long. My personal favorite is the EMI susceptibility—SiTime’s MEMS oscillators are “up to 54 times more immune” to EMI than quartz devices. Fifty-four!
If you’re interested in learning more about the performance improvements offered by MEMS oscillators, a good place to start is the Supplemental Information section of the SiT2024B datasheet.
Do you have any experience with MEMS oscillators? Leave us a comment and let us know.
Featured image from SiTime.