0
\$\begingroup\$

Hey I'm looking for suggestions for implementing a very high precision timekeeping device.

I need to record independent signals in multiple devices and then time match them to an accuracy of about 60uS (3 samples at 48khz).

I was thinking of using GPS for time syncing (good to about 10-50nS with cheap commercial receivers), but some of the recordings could be done in areas without good gps signal.

My current idea is to sync all the devices initially (either with gps, or plugging them to a single pulse generator) and them keeping time locally with a high precision RTC of some sort and assume that the RTC can keep the needed accuracy (±30uS) for a reasonable amount of time (6-12hours) before needing to time sync them again.

Unfortunately the best RTCs I have seen have an accuracy of ±2.0ppm which is a time drift of about 41mS per 6hours, which is just too much for my application.

I'm looking for recommendations of how to keep an accurate clock signal across multiple non connected devices.

\$\endgroup\$
  • 2
    \$\begingroup\$ An atomic clock... \$\endgroup\$ – Solar Mike Feb 24 at 15:49
  • \$\begingroup\$ @SolarMike - actually available. microsemi.com/campaigns/atomic-clocks-mac \$\endgroup\$ – Peter Smith Feb 24 at 15:53
  • \$\begingroup\$ Irig b is worth a thought and can be resynced to satellite time when back in range. \$\endgroup\$ – Andy aka Feb 24 at 15:54
  • 2
    \$\begingroup\$ You can buy an ovenized crystal oscillator at digikey with 50 ppb stability (0.05 ppm) starting at $32 (qty 1), or 10 ppb starting at $55. \$\endgroup\$ – The Photon Feb 24 at 15:55
  • \$\begingroup\$ Do you have access to any common reference signal, even if it’s not that precise? For example 60Hz line? This should allow you to model individual oscillator drift at a much narrower time scale than 6hrs. \$\endgroup\$ – Edgar Brown Feb 24 at 17:26

Your Answer

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

Browse other questions tagged or ask your own question.