External GPS Clock frequency suggestion?

I've recently used both the original 2-output Bodnar GPSDO and now the Bodnar mini-GPSDO successfully with the KiwiSDR. Per a suggestion from John, I've set the external clock frequency to 66.660000 MHz. In order to get moderately but not completely square output waveform, I've turned the current up to 32 mA. The result is that everything works fairly much as expected with no surprises. My one issue is that I think there may be a small frequency error that from my experience with Bodnar GPSDOs operating at 'strange' frequencies, I don't believe to be due to the Bodnar. The symptom is a small frequency error. For example, viewing the 25 MHz signal from WWV, located about 20 km from me and not via the ionosphere, I see a few hundred milliHertz error. This shows up when WSPR spots are compared with a separately referenced and believed accurate Apache SDR with a very good quality Wenzel OCXO and GPS discipline. It also shows up when I use the IQ display to look at the carrier. Rather than sitting still as it really should for a 1 ppb-class reference, it rotates.

Perhaps I've misinterpreted John's suggestion but I'm wondering if there might be a better reference frequency to use.

Glenn n6gn
Fort Collins, CO.

Comments

  • Hold on, there might be a problem with how the Kiwi mixer NCO phase value is computed. I am investigating..
  • Okay, v1.233 is out with a possible fix for this problem. Please check and see if the rotation in the IQ display is less for you now.
  • Yes! That fixed it. It's holding still with just a couple of degrees of jitter which may be either phase noise or simply SNR on the signal. I'm only 13 miles from WWV but there is actually quite a lot of absorption by this frequency ( see my QEX article for details http://www.sonic.net/~n6gn/Elmore4.pdf)
    I haven't yet checked WSPR spots or similar but I'm almost certain they will be perfect again.

    Glenn n6gn
  • jksjks
    edited September 2018
    Great. Now if you happen to try this with WWVB at 60 kHz (given your excellent proximity) you will probably still notice a small residual rotation. That turns out to be another problem whose significance I hadn't appreciated until now (it only applies to signals at LF/VLF). There is a possible fix for this too, but it will require an FPGA change. I'm looking at this now.
  • Well, I did look at WWVB but unless you had commented, I probably wouldn't have stuck around to notice the issue. I see the expected 180 degree flips but there *does* also seem to be a very slow residual rotation. I estimate that it is on the order of 4 minutes so I suppose that's around 70 ppb and clearly worse than either WWVB or the ext clock on the kiwiSDR, so I guess it's real. Down that low in frequency it starts to not matter as much. Getting HF fixed is significant though since with WSPR, for example, with good timebases you get what is effectively a bi-static radar and one can see Doppler due to path length changes from ionosphere height change and such, which can give very useful insight into propagation.

    This SDR is getting more and more useful all the time as a measurement tool.

    Let me know if I can measure anything else for you.

    Glenn n6gn
  • jksjks
    edited September 2018
    Wow, that worked out better than I thought it would. v1.234 is out with the receiver (audio channel) DDC mixer NCOs changed from a phase word size of 32-bits to 48. The images below say it all. The signal is North Dakota MSK on 25.2 kHz received by a Kiwi in Kansas. The IQ extension density plot has been captured over a 5 minute period and shows no rotation. The second image is after 30 minutes and only a slight rotation (or perhaps wandering). Note the PLL that Christoph added to the IQ extension is switched off. So this is solely the ADC clock as corrected by the GPS.

    There will be some improvement for signals at HF. It would be interesting now to do some long term frequency measurements of the Kiwi's GPS-corrected XO versus an external frequency standard applied as the ADC clock.

    Thanks very much Glenn for prompting me to take a second look at all this.

    image

    image

    Attachments:
    https://forum.kiwisdr.com/uploads/Uploader/f5/49bce8106a7efe9dc3191e55753900.png
    https://forum.kiwisdr.com/uploads/Uploader/14/890ef9c838dd76b828e146c28e0c8c.png
  • jksjks
    edited September 2018
    This is fun! Already I can guess that the French are probably not using primary standards on their MSK signals. HWU 20.9, FUG 62.6 and FUE 65.8 all rotate to varying degrees. Their competitors are all rock solid. But TDF-162 Allouis, a time/frequency standard, is of course dead on.
  • FWIW, NAA Cutler Maine is < 100 miles from WA2ZKD/1 and is always rock-steady strong+ there
  • This is great work!

    I remember that besides the MSK signals you have mentioned also DHO on 23.4 kHz is/was not locked to a frequency standard.
    WA2ZKD
Sign In or Register to comment.