Frequency drift

Question – I am also using the Kiwi to decode QRSS signals. In the attachment, you can see the steady decode of the CQ beacons on 40 meters...rock-steady. Then suddenly at 0200Z the signals start to drift down by about 20 Hz and restablize. Then at 0300 they start to return to original.

WSPR spots show the same behavior.

I am using the GPS with default cable and default antenna. I have the antenna hanging vertically, resting outside the window of my office.

GPS position in wikiamap.org is accurate. Any suggestions on how to investigate further?

Thanks, Mark WD4ELG





«1

Comments

  • jksjks
    edited October 2020
    Please post a screenshot of the admin page GPS tab during a time when the drift offset is occurring. Set the "select graph" menu to "az/el". I want to see what the sky distribution is of the sats. Black out your lat/lon location on the screenshot if you wish.

    Also, please do this after the Kiwi has been running for a long time (like at least 1 day) so the az/el graph has a good cyan-colored history of the sat tracks.

    Is the GPS really oriented like that? With its base on the side of the structure and the "top" of the antenna not facing toward the sky?
  • What direction is the wall facing, that the GPS antenna is mounted on?
    More accurately, what compass bearing is the flat top of the GPS antenna pointed towards?

    Are there other buildings in the vicinity?
    Where I have my antenna, I can see a shadowing effect of the two houses closest to the antenna, towards the E in this image:

    image

    Also - the local environment where the KiwiSDR itself is located, are there strong temperature fluctuations through the night? If the GPS signals are flaky and intermittent, and there are gross temperature chances of the KiwiSDR itself, then the drift offset previously calculated will be invalid for the different temperature times, and that would easily explain the frequency drift you're seeing.

    Attachments:
    https://forum.kiwisdr.com/uploads/Uploader/0a/3f7c20d19b8308a66ea416dd661d32.jpg
  • Thanks for the replies.

    John, the GPS antenna is oriented like that...since I can't get satellites to the north, I figured I would try orthogonal to the south. Attached GPS screenshot. This is with the unit running for 3 weeks.

    Cathal, no temperature fluctuations, no sunlight on the unit and plenty of AC air to keep the room cool.

    I am including another screenshot showing the long-term capture of the 40 meter slow-speed CW beacons...0730 UTC shows the behavior with correction at 0830 UTC, and it starts again at 1550 UTC.





    Attachments:
    https://forum.kiwisdr.com/uploads/Uploader/cd/6668f2d4a693881195ec36a144ae7d.jpg
    https://forum.kiwisdr.com/uploads/Uploader/79/250b4fbd276ee57e50e8965f27fb5b.jpg
  • jksjks
    edited October 2020
    Okay. So that's truly terrible GPS reception. It might be the case that the periods of short-term deviation are when there are enough sats being received that the GPS is actually correcting. And the longer-term stable periods are when the xtal oscillator is free-running in your temperature stable environment and behaving itself nicely. BTW, try checking the "acquire if Kiwi busy" checkbox and see if the situation improves.

    The gradual transition shape might be the averaging that is applied to the GPS corrections so the frequency doesn't jump around. Although I thought I had set the time constant a bit shorter than that. It is a compromise between smoothing noisy GPS time/position results and dampening sudden temperature changes (e.g. Kiwi in garage when the sun goes down, door is opened letting hot air into room Kiwi is located in etc.) However, the noisy GPS issue is now essentially eliminated with Christoph's Kalman filter.

    I have thought about having a long-term graph on the admin page showing GPS reception and clock correction details to help answer some of these questions.

    Specifically in your case: The GPS tab shows 142 hours of runtime. That's 5.9 days. 6.2k fixes over that interval is very small given fixes occur every two seconds in the best case. That would have been 255k corrections in 142 hours, so your Kiwi was only correcting 2.4% of the time. This also lends support to the argument that the short-term excursions are when GPS correction is occurring.

    The az/el also shows very limited sky coverage which means the GDoP is low which in turn means the quality of the corrections is low (although, again, I think Christoph's Kalman filter / Single point positioning (SPP) makes this issue moot).

    So, unless you can improve your antenna placement I think you might be better off to disable clock corrections (setting on admin "config" tab) and do a one-time calibration to WWV (see here: http://kiwisdr.com/quickstart/index.html#id-cal). And then check the calibration periodically. [Anyone remember the good old days of checking ham rig calibration using the harmonics from a 100 Kc xtal calibrator? Am I really that old? Jeeze.. :smile: ]
  • OK there is room for improvement in reception there...
    I'm sure I've seen that level with a broken coax (the shield corroded to white powder).
    I really would check that antenna and feed, also move it away (shield) from WiFi routers, motors or DECT phones.

    There are other threads on GPS antenna placement but first off I'd do my best to put it out horizontal on a metal plate (about the size of a CD/DVD), ideally more than about 20cm from the wall. If it has to be on a window try a plate behind it.

    Here I use a car GPS repeater to get the antenna up at gutter level, Kiwi GPS antenna and repeater TX in a can close to the Kiwi.

    73 Stu
  • jksjks
    edited October 2020
    Now here is excellent GPS reception. From one of Stu's Kiwis actually. Almost 100% fixes over 40 hours of runtime. Note the very good sky coverage and the characteristic "polar hole" at the top that we've discussed elsewhere on the forum (so we know this Kiwi is in the northern hemisphere at least).

    image

    Attachments:
    https://forum.kiwisdr.com/uploads/Uploader/fe/4908a7dce3202fa54bd6daa17fc144.png
  • Mark.... your giving up the overhead stuff by that "sideways" aim. Put an old tin pie plate on a sidearm, as long as the antenna coax permits, and face it up.
  • Great feedback and suggestions, everyone.

    Here is the new implementation of my GPS antenna. If this does not get the better reception, then I will go "old school" using the method John described and I remember using in my early ham days (back about the time the lightbulb was invented LOL).

  • You should know right away if it helps by seeing what sats you get compared to the (few) you have in the posted graphic above.
  • No real change, Jim. I am concerned about extending the coax for the GPS antenna and ending up with a degraded signal with longer feedline.

    What are other folks doing for their KiwiSDR antenna setups?

    I do have a workshed with power, out in the back yard. I wonder if I should just shift the unit to there....
  • On the chance that it IS the antenna/cable (easy way to eliminate that as an issue), I will order a replacement.

    Is this an active GPS antenna? Any suggestions on a replacement unit?

    Thanks
  • Extending the feed line is a topic with many posts on the forum. Also, please see the Kiwi FAQ for more suggestions: http://kiwisdr.com/quickstart/index.html#id-faq-gps-cable
  • And I noticed something interesting about the antenna cable...it looks like a pinch or kink. Wonder if that may be a factor...
  • jksjks
    edited October 2020
    Possibly. But I do that to mine here all the time without much effect. The cable runs out a louvered window with a foam seal that compresses the cable when the window is closed and latched (to keep air from leaking into the room). The cable has all kinds of kinks in it as a result.
  • jksjks
    edited October 2020
    Can you rig up a long Ethernet and power cable and temporarily move the whole thing closer to the outdoors so the antenna can be placed in the clear in the yard on the pie pan facing skyward for testing?

    Remember that a sat is not actually being received until the "subframe 54321" blocks light up with colors. The stuff to the left may look like a sat is being received but often it's just the acquisition software desperately trying to pull the PRN-coded signal out of the noise and do a tracking lock. For example, there are no sats being received in the 4th post of this thread.
  • I figured out a way to get the antenna outside the window....there is a planter for flowers that I can rest it on.

    Instantly I am now picking up more satellites.
    WA2ZKD
  • Thanks for mentioning the "polar hole", John. I mistakenly thought it was due to the line of evergreen trees directly to the north of my outside GPS antenna :)

  • The so-called polar hole is not a hole concentric with the pole. The coverage is more scalloped shape and its impact depends on your longitude.
  • Consequence of orbit inclination I think? I don't know exactly. Best I could find with a quick search: https://www.e-education.psu.edu/geog862/node/1871
  • It's non-concentric as a result of perspective, as the satellites visible at the highest points between the pole and the northern horizon are significantly farther away than the satellites between the pole and overhead.
  • At least I know now that I can shelve any plans to further raise the height of the outside GPS antenna to improve coverage to the north.

    Speaking of GPS... I have the one outside high gain antenna feeding my Thunderbolt 10 MHz reference as well as three KiwiSDR receivers. This is probably going to make some RF purists cringe... but here's how I do it:



    The Thunderbolt directly feeds the antenna so it can power it. I have four capacitors (0.047 uF?) tapped off that, each going to an F connector. These feed the KiwiSDRs and provide DC blocking. Ugly, but as you can see above, it works. The 4th output is in case I get another KiwiSDR :D
  • JKS,All

    This thread is super informative and helpful. I hope you will organize the graphics and supportive text here and add this to the official documentation stack for the Kiwi. While I have a fairly good update rate on my Kiwi, I now understand why the legs on my sat tracks are so short. This is good material for someone like me, who is not knowledgeable in GPS.

    Mike N8OOU 73
  • The only thing I'm worse at than coding is documenting :/

    Also, obligatory: https://xkcd.com/1356
  • Here's a graphic which shows the scalloped pattern I mentioned
  • That brings up an interesting point. The Kiwi GPS software lacks most of the features of commercial equivalents, which were developed over many decades now. One of them is masking out sats (i.e. not using them for position/time solutions) if their elevation angle in the sky is too low. There are various reasons this is a good idea as you might imagine.
    WA2ZKDrz3dvp
  • "masking out sats if their elevation angle in the sky is too low"

    The mountains do that for me ;)
    Good for GPS(?), but bad for DX.

    However, I own a GPSDO, (Navstar only, no Galileo) which never archives specified accuracy. Maybe due to lack of visible satellites, or reflections, I don't know.
  • I am puzzled by the frequency offset of this KiwiSDR. Despite that fact that GNSS reception seems to be ok there is a large offset.

    The ADC clock frequency shown in the IQ display window is the default value:
    ```
    ./init/clk.h:#define ADC_CLOCK_TYP (66.665900*MHz) // typical 20 degC value
    ```

  • jksjks
    edited October 2020
    I didn't think this Kiwi (Mark's) was public. What URL are you using to get to it?

    My guess is that on the admin "config" tab the option "Enable GPS correction of ADC clock?" has been switched to "No".

    I say this because on the IQ extension, as you show, the manual "Fcal" adjustments are shown as opposed to the usual message saying "GPS is correcting". This in combination with the fact that the "Stats" tab is showing plenty of sats with good reception (i.e. the Fcal adjustments will be shown when a Kiwi is first started up until GPS reception is established and GPS correction starts occurring).

    And lastly, due to the large offset of (presumably) 15 MHz WWV it looks like a manual calibration has not been performed, or performed improperly.
  • Thx. Most likely the "enable gps correction for clock" option is not active on this kiwisdr. (http://irelandnorthwest.proxy.kiwisdr.com:8073/) which besides that is excellent for transatlantic MW DX.
Sign In or Register to comment.