I'm encountering a problem with a ceramic resonator as soon as our device is subjected to ultrasonic vibrations (ultrasonic cleaning bath).
Our device uses a semi-custom IC which integrates the oscillator circuit, it's a simple Pierce oscillator with the matching resistors built into the ASIC. It is optimized for the Murata CSTCC_G_A 2 MHz ceramic resonator.
If our device is put into an ultrasonic cleaning bath (electronic is sealed) the resonator starts to act up. What I can measure with the oscilloscope is a spurious frequency of around 66 kHz which is close to double the frequency of the ultrasonic bath (35 kHz). I measured this at the oscillator input of our ASIC.
It looks like this (in run mode):
Now I understand that a ceramic resonator contains some sort of piezoelectric ceramic element which has the resonance frequency you like. It being piezoelectric of course also means that external vibration will be transformed into a voltage. But the amplitude just seems so massive, that it seems like I hit another low frequency resonance of the resonator.
Is there a way to reduce this effect so no 67 kHz oscillations will occur?
My first try was to half the feedback resistance Rf of the Pierce oscillator (externally) but that didn't change anything.
My second try was to use a resonator vom AVX/Kyocera, the housing is made differently, so I hoped it would behave differently. But my hope was smashed - it's the same there as well.
I was thinking about a high pass filter structure in the oscillator circuit, but I'm not quite sure on how I'd do that and if that would even help.
I find it a bit hard to find the right keywords to do an effective internet search on this topic. Most stuff I found is concerned with higher order vibrations of the crystal itself (like 2 MHz -> 8 MHz vibrations) and how to suppress them.
Additional info you may skip:
We need a stable frequency for our measurement. The device is required to withstand -40 to 125 °C which limits options significantly. The device is low power running with just around 1 mA @ 3 V, the oscillator takes around 60 µA and shouldn't take much more than that. I haven't contacted Murata or AVX/Kyocera so far - communication with Japanese manufacturers were a bit difficult last time, so I'm asking here first.
Part number of the currently used resonator: Murata CSTCC2M00G53A-R0
Part number of the resonator used in the previous device: Murata CSTCC4M00G53A-R0
Part number of the AVX resonator tested: PBRV2.00HR50Y000