1
\$\begingroup\$

I'm planning on making an audio system that will be used in vibration-inducing environments (roller coaster rides that can reach up to 5/6G peaks and public transportation systems). One of its components is an oversampling DAC IC (the TI PCM1781), that requires an external 11.2896 Mhz clock signal to operate.

I was originally planning on using a crystal oscillator as my clock source, but then I learned that they're quite sensitive to G-forces and vibrations, which can even drastically and irreversibly change their output frequency in some extreme cases.

So I then learned about MEMS oscillators, which are less sensitive to vibrations. The main problem with them here is that they output quite a lot of phase noise, which I've been strongly warned about, since my system is an audio oriented one.

I'm therefore stuck with this question : should I feed my DAC with a crystal oscillator or a MEMS oscillator, given the harsh vibrating environment it will function in and the drawbacks of each oscillator ?

\$\endgroup\$
5
  • \$\begingroup\$ Is the desired audio quality hi-fi or more like AM radio? \$\endgroup\$
    – Andrew Morton
    Aug 9, 2021 at 16:48
  • \$\begingroup\$ It would be more towards Hi-fi, but the goal here is not to reach audiophile specs. Let's say, my system would need to output sound with a quality equivalent to a consumer grade micro-hifi system. Something standard (im only dealing with 16 bit PCM anyway). \$\endgroup\$
    – DashNode
    Aug 9, 2021 at 16:55
  • \$\begingroup\$ How would you connect the oscllator to PCM1781? Not only it needs the master clock, but the audio interface needs to be synchronous to the master clock. So simply giving a random oscillator signal that is unrelated to the audio bus will not work. \$\endgroup\$
    – Justme
    Aug 9, 2021 at 17:37
  • \$\begingroup\$ I would build a crystal oscillator and test it under vibration. 11.2896 crystals used to be standard items. Chances are it'll be fine. But if you're still worried, I'd consider rolling it in sponge foam and stuffing it in a 35mm film canister. (Assuming you can find one nowadays!) \$\endgroup\$
    – user16324
    Aug 9, 2021 at 18:25
  • 1
    \$\begingroup\$ You'll get so much improvement in phase noise as the signal is effectively divided down from 12 MHz to audio sample rates that any crystal will be adequate for audio, especially if the audio is being listened to on a roller-coaster or in a subway. \$\endgroup\$
    – Neil_UK
    Aug 9, 2021 at 19:06

1 Answer 1

Reset to default
1
\$\begingroup\$

The Q factor at resonance of the resonator , PCB unit and excitation has more to do with it than comparing MEMs and AT-cut Xtals in SMD packages. HC- leaded packages might another story depending on worst case axis.

The stress of shock levels depends on the duration such that it is the velocity of impact or g-s product that increases with the mass and inertia of the component. Thus heavy parts are more fragile to long duration low g shock levels and low mass parts can withstand much greater shock levels of short duration <1ms , but these short durations tend to excite mechanical resonances in the fixtures. If you are planning on testing fragility, I try to include some vibration damping material on the device interface and compare with a solid structural mount sense chassis vibration. 5g vibration is pretty uncomfortable at certain frequencies but easily survived by Xtals.

The g level can be calibrated as the drop height, d,mdue to gravity and the ratio to stop height,s with a damping material on a spring, g=d/s

  • e.g. a 0.1m drop to a hard surface with 1mm deflection is 100 g
  • A 10m drop with a 10m vertical stop (at any horizontal speed) is 1g.
  • They do make ruggedized mounts for SC-cut OXCO, but I only used those for rockets.
  • When in doubt, it is better to choose an XO chip that has been certified to …

Additional Reliability and Qualification

PARAMETER Mechanical Shock Vibration CONDITION / TEST METHOD MIL-STD-202, Method 213 MIL-STD-202

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.