I'm currently using a Maximum Integrated MAX7375 MEMS 8MHz oscillator (http://www.digikey.ca/product-detail/en/maxim-integrated/MAX7375AXR805-T/MAX7375AXR805-TCT-ND/4967873) as a clock source for a microcontroller project.

While doing some tests on the microcontroller, I noticed that the internal time of the microcontroller was going a little faster than the real world time. For example, when I would set an alarm on the microcontroller for 30 minutes, it would be go off after 29:44. When I would set it for 1 hour, it would go off at 59:28.

I then attempted another timing method within the microcontroller and had the same results. I decided that my 8MHz clock must be going a little faster than 8MHz. I then took the oscillator to the oscilloscope and was able to confirm that my clock was operating at 8.09MHz.

So my question: is this performance typical, or in range for this oscillator? I notice that the data sheet specs the oscillator as 8.00MHz, which seems to suggest that my oscillator is operating at the worst possible rate for its tolerance.

Furthermore, why exactly would this oscillator be performing so poorly? I thought MEMS oscillators were often considered superior than cap/crystal resonators? Thoughts?

  • \$\begingroup\$ Well, have you like, looked up exactly that information in their datasheet? \$\endgroup\$ – PlasmaHH Oct 26 '16 at 13:46
  • 3
    \$\begingroup\$ The datasheet shows an accuracy spec of 2%. So anything between about 7.84MHz and 8.16MHz is within spec. This device isn't particularly suitable for precision timekeeping. \$\endgroup\$ – brhans Oct 26 '16 at 13:49
  • \$\begingroup\$ @PlasmaHH All it states is 8.00MHz, but does that mean the last significant digit can vary? So 8.09MHz would still considered meet tolerance? Furthermore, the datasheet lists other chips in the family at 4.19MHz. I'm just trying to figure out if the last digit is free to vary. \$\endgroup\$ – Izzo Oct 26 '16 at 13:50
  • 1
    \$\begingroup\$ @Teague: intial accuracy is the accuracy that it is delivered with. It can get even worse over time. \$\endgroup\$ – PlasmaHH Oct 26 '16 at 13:52
  • 1
    \$\begingroup\$ When reading a datasheet, always remember that it was drawn up by the marketing dept. There will be some "specmanship" involved - particularly for any number related to accuracy/performance/etc. The most obvious specs shown will be the best case and you're going to have to dig and add up to get a realistic number. \$\endgroup\$ – brhans Oct 26 '16 at 13:59

The answer is behind the characteristics:

enter image description here

So, this part has a minimum frequency of 8 x 0.96 = 7.68MHz and a maximum frequency of 8 x 1.04 = 8.32MHz. You measured 8.09MHz which is in guaranteed range.

| improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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