LNB-LO: GPSDO or not GPSDO that is the Question

  • Hi,

    I asked myself, do I really need a GPSDO or would an x-tal oszillator stable enough? Whats about the phase noise?

    As I had no x-tall of ~25Mhz lying around, i tried it with a FA-SY2 from http://www.box73.de which I use for testing here...

    The test went much better than expected. After powerup for about half of an hour, it is stable enough for CW.

    And on top, the phase noise is very low and it can be tune to any frequency between 10 and 280 Mhz.

    73 achim

    see you soon on QO-100 ( waiting for the upconverter )

  • When it comes to phase noise, a good TCXO can be better than a badly designed (or set up) GPSDO. Its advantage is the absolute frequency accuracy, so if you can live without it, I would go with the TCXO solution. Much less of a hassle as there are many errors one can make with GPSDOs.

  • we worked a lot on this issue during the last two weeks:

    A megaset LNB was modified as decribed in cqDL 01/19. The crystal was removed and pin 14 of the mixer chip connected to one of the external connectors.

    First I ordered a TCXO from farnell, 2.5ppm.

    The TCXO was built on a PCB together with a voltage regulation, a 25 MHz filter and some decoupling caps.

    The mid frequency was quite stable, but it jumped up and down for a few 100 Hz every second. I read about digital-TCXOs, so this one obviously is digital.

    For the next try I purchased an analog-TCXO. This was also stable enough but now I got some random jitter for a few 100 Hz. The frequency moved up and down, like before with the other tcxo, but now it did'nt jump but was a continuous moving up and down.

    So I gave up using TCXOs and purchased a 1-port GPSDO from sdr-kits. It was delivered yesterday. Again I added a 25 MHz filter (which also adjusts the voltage level to about 6 dBm, about 1.2Vpp). And also a voltage regulation so I can supply it with 12 volts (it needs 200mA at 5V).

    The GPSDO was programmed to 25 MHz and connected to the LNB.

    The frequency was stable, very stable (!), but the signal had a low frequency humm of about 10 Hz and significant amplitude. Still not usable for listening to QO100.

    After some reading here in the forum I found an advise to alter the frequency by only a very little amount.

    So I tried 25.000.001 002 003 ... and also 24.999.999 and so on.

    With a setting of 24.999.999 MHz the signal was totally clean and stable like a rock.

    This is my setup now and I am very happy with it.

    best 73, Kurt

  • How about changing the LO-frequency to about 25.8 MHz. so that the LNB IF-frequency will be within the 70cm band, I 'am also interested if this will work with the Box73 FA-SY1 / SY2 - DB6QS ?

  • DB6QS

    Using not modified LNB (PLL-type) with a SDR-dongle 0,5ppm. After 10 minutes warm-up it is very usable for SSB.

    There notable drift during outside temparature change. Especially when clouds moves around the sun it can bee courious. But still workable.

    I have tried software compesation with SDR Console 3.07 https://www.sdr-radio.com/Console/EsHail2 and seems to be good idea. I am running Atom n270 with 1GB and that's not enough, so sometimes I hear cracks in the output sound. But with better machine You can probably obtain good results.

  • How about changing the LO-frequency to about 25.8 MHz. so that the LNB IF-frequency will be within the 70cm band...

    good idea, I will reprogram the gpsdo tomorrow and give it a try. According to the report in cqDL the megaset should be usable for 70cm output.

    In difference to the cqDL report my megaset had a different crystal. It is nearly impossible to desolder it. So I cut the trace and soldered the wire directly to the trace going to pin 14 of the mixer chip.

  • thanks for the link Patrick. I have already read this thread before doing my experiments with the megaset, it helped a lot !

    Today I tested the megaset on 70cm.
    The GPSDO was set to 25.769231 MHz. This moves the CW beacon to 439.550 MHz, so the kHz are the same as on 10 GHz.

    First I looked at the transponder noise level using gqrx. It is the same as on 739 MHz, no change. Then I compared the SNR of the beacons and other stations, no change. I could not see any drop in sensivity when using 70cm IF.

    Next I connected the IC9100 in parallel to the SDR (because the SDR has a built-in LNB phantom supply).

    The result was very nice, using the features of the Icom (i.e. noise reduction) the sound is crystal clear, significantly better then my PC-speakers. But I am missing the waterfall.

    The SDR must be running in parallel, just for the waterfall display. I plan to write a small piece of software which reads out the frequency via CAT and then shifts the SDR to this qrg.

    vy 73, Kurt