If you could start out with both 10MHz and PPS using GPS signal, set all clocks to the same time (e.g., zero) at the common PPS, and then switch to using external ref, it might be possible. My understanding is that the only thing the PPS is needed for is setting the common time. In other words, it's basically a trigger.
I appreciate the suggestion. If the FPGA only uses the PPS when certain UHD API calls are invoked, and doesn’t use it for internal correction or skewing of the timebase, that might work. However, I was also planning to occasionally use get_time_last_pps() to monitor clock drift with respect to GPS PPS over long durations, which I wouldn’t be able to do with this approach. Giving it more thought, I’m perplexed as to why “clock_source=external,time_source=gpsdo” isn’t supported. The FPGA would be willing to accept an external timing source that has no correlation to the ref clock, so conceptually, it seems like it should be able to use the GPSDO PPS the same way here. Sure, over time there will be frequency drift by not using the GPS-disciplined oscillator, but that's an expected consequence of using an independent clock ref. I’d love to hear why this constraint exists. David
_______________________________________________ USRP-users mailing list -- usrp-users@lists.ettus.com To unsubscribe send an email to usrp-users-le...@lists.ettus.com
