Captures of PLL LNB drifts over time. WIth & without mods.

  • Hopefully I will do some measurements of my own Octagon LNB's in the next week or so, although I intend to measure over a 24 hour period as well as short periods.

    I also have a gnuradio flowgraph to measure a carrier's frequency. It will not be much extra code to correct the receive frequency, thus hopefully keeping the carrier 'straight' within the pass-band.

  • Here i made a screenshot of the running SW correction.

    Upper window is the signal from the LNB.

    Lower right window is the AF output after SW correction.

    Unfortunately the time span of the lower right AF window can't be set higher than 1m in the used version of HDSDR and with the newest HDSDR version the control via mouse scrollwheel doesn't work good.

    But believe me the demodulated signal sits on 1Hz for hours.

    vy73 DB8TF

  • The SW correction has nothing to do with my transmission signal!

    The EBs or any other stable signal from the sat can be used.

    Clearly the solution with 2PCs is not very usefull for portable use but as i mentioned already months ago on my hp i hope someday someone will write a sdr software which can run this correction with only one RTL SDR and one PC (portable RSPI).

    But for now it is nice to see some measurements of various solutions to the drift problem.

    For demonstration purposes i have uploaded a small video in which the running SW correction can be seen.

    See it here.

    vy73 DB8TF

  • That is great... is that something you added to that software or is it just a special setup of the existing software?

    I think the ultimate solution would allow to do this within a single program, where you select one carrier as a reference and then are free to tune around the passband but remain locked relative to that other carrier.

    (entirely in software, no hardware mod required to LNB or SDR)

  • The HDSDR is a "stock" unchanged Version 2.76.

    But would be work with any other SW which is capable of changing the frequency from an external source. I use the mouse scrollwheel in my setup.

    The whole correction is done with a small microcontroller which measures the drift of the signal on the master PC and the slave PC is following and demodulating the audio.

    No expensive GPSDO or fiddling (mod) needed around a standart PLL LNB.

    vy73 DB8TF

  • The difference is that your controller must send the Up and Down (or direct frequency ) commands via TCP instead of my mousewheel solution.

    Direct freq control would be awesome fast. My solution can (only) tune about 5kHz per min.

    But for sure this is working also very well ;)

    vy73 DB8TF

  • They key point is that you should tune the "offset" value in gqrx, not the "hardware freq". Because the "offset" can be tuned without any transient effects, while changing the "hardware freq" will cause the PLL in the SDR device to unlock/lock which is clearly audible and would cause errors when receiving digital modes.

    And of course preferably you want to change the tuning without changing the "dial frequency" as well. The "LNB LO" frequency would serve that purpose if any changes to it would modify the offset rather than changing the hardware frequency, or affecting the dial frequency.

    Unfortunately the gqrx author has expressed that he will make no more changes to this program, but of course it is open-source so someone else could do that instead.

  • I repeated the same experiment with my implementation of locking the LNB to a 10 MHz GPSDO.

    Couldn't find an easy way to reduce the waterfall speed sufficiently, so increased the FFT resolution to slow things down.

    I am not disappointed. In the 100 Hz resolution screenshot some 'linear' jitter is visible, although also 'binary' drift (ca. 2 Hz) is present. I suspect the (warm!) RTL-SDR dongle since it's not disciplined and susceptible to environmental aspects.

    E.g. when I approach the laptop where the dongle is inserted, some drift appears.

    The 27 MHz LNB signal comes from an Abracon ASVV VCXO and is locked with an ADF4001 with PFD 1 MHz. As far as I could ascertain these VCXO's are not numerically controlled.

    Anyway, it's fine for now. When I want to do narrow band digital modes (WSPR, JT9, FT8) further investigation is desirable.

  • I did some short test with Gqrx remote interface: Setting LNB LO take no effects to rx frequency. Only the displayed rx frequency changes. Setting new rx frequency via remote interface works fine.

    If there is a drift so we can set the new rx frequency at first and then set the new LNB LO frequency.

    I could not hear any (disturbing) clicks during changing the rx frequency. For SSB/CW i think this will be ok.