IQ file timestamp accuracy

I have failed to find a forum post which I think reported those timestamps to be better than 20 usec accurate.
Was that true and if so has the accuracy been improved by the Kalman filter?


  • Hi Rob,

    Yes this was true and was based on measuring DCF77 time signals where due to the width of the time pulse no better than 20usec accuracy could be verified. Now with the Kalman filter GNSS position solver this should be redone:
    * possibly using the digital modulation of DCF77
    * measuring 1pps signals from a GPS mouse
    * measuring the phase of VLF MSK modulated signals which are known to be very stable.

    In principle GNSS time accuracy should be better than 100ns (10m ~ 33.3 ns). Note also that one sample at 12kHz lasts for 83.3 usec, so one has to take averages over many samples to verify time accuracy to better than 1 usec.

  • Hi Christoph,

    Let me know if you want me to connect the 1PPS RF marker to my spare KiWi.

    Also does John's new 20KHz Bandwidth modification upset the IQ stream when used for TDoA purposes ?


    Martin - G8JNJ
  • edited January 2019
    If there are signals in the WWV transmissions, Glen N6GN's Bodgar trained Kiwi at is line of sight to WWV.
    See this discussion about its oscillator accuracy:

    What could I record on WWV which would show the accuracy of the record time stamp?
  • Be careful with N6GN/k2's Bodnar clock in the near term. Last trip up there the GPS antenna got moved to a location that turns out not to be great. It wasn't even locked initially but was the last time I checked. I'm not sure what will happen as the constellation of satellites shifts around, the antenna has a pretty solid wall/shadow to the East until we move it next time we visit. It turns out that the glass window it sits next to has a coating which appears to be a pretty good reflector at 1575 MHz.
    Glenn n6gn
Sign In or Register to comment.