Octagon Oslo PLL LNB with external Reference

  • Heiner, it is not wise to let it generate 10 MHz because these Ublox modules have NCO's (numerical controlled oscillators) inside. The master clock inside the unit runs at 48 MHz. 48/10 = not an integer, so inherently there will be very much jitter.

    It is better to use an integer fractional value (google on it and see Ublox-7 Receiver Description document or previous posts in this thread ; -).

    Additionally, the (10 MHz or whatever) signal quality is useless as reference.

    However, I can be used in a 'clock recovery' circuit (like also mentioned in the Ublox-7 receiver description document). And that is exactly we did: take another (short term stable) oscillator and make this long term stable with a PLL, using a 'dirty' reference. Google on it, there are several publications on this topic.
    So nothing new here ; -)

    Last but not least: my recommendation would be to remove the LED from the PPS output. It generates harmonics at higher frequencies.

  • PA3FYM Hi Remco, i calculated the loop filter by using this pdf file: http://www.k9ivb.net/GPSDO/fil…0GPSDO%20PLL_11092014.pdf .

    You are right the jitter will disappeare by using a 22µF capacitor. Perhaps it would be better to recalculate the loopfilter for a smaller loop bandwith. But what really is bad is the SI5351. You can not use this device for producing a signal which can be fed into a lnb. The signal of the si5351 is awful. In the moment i take the 10MHz gpdso signal and use it as reference for a 27MHz pll described by df9np with an adf4001 and a VCTCXO with 27MHz

  • DL5MGD Yep, it's awful but I read people use it (with some filtering?). My first idea was to generate the 27 MHz reference with the 'ProgRock' but the signal gave me goosebumps (Gaensehaut) ; -)

    Concerning the loopfilter. I didn't knew/found that K9IVB publication. Thanks for the hint!

    His 2nd order loopfilter seems a little different than the 'normal' 2nd order filters.

    I just did some simulations with the 2nd order loopfilter from ADISimPLL. Needs some experimenting though. First build the 'junkbox GPSDO' in a casing and then refine it.

  • My technical knowledge is shallow compared to most of you so I make this suggestion with care! I use a PIC micro to control my si5351a over the I2C connection. I have found that the output at 27MHz - as seen on my SDR - is very noisy. However, putting the PIC to sleep after a command has been sent seems to result in a clean output. I hope this helps. 73 Russ

  • Certainly a good point, when I program PLL's I put my PICs (or Atmels) asleep.

    Only tested it with ADF-PLL's, so perhaps DL5MGD can try it with his contraption?

    set_sleep_mode(SLEEP_MODE_PWR_DOWN);  // select PWR DOWN mode
    sleep_enable();                       // set sleep bit in MCUCR
    sleep_mode();                         // put the CPU asleep
  • Hello,

    I have a "powebank" with LiFePo4 accus with 12V (20Ah). It has a 12V output (as emergency backup for starting small cars) - but allso good for protable HAM-radio operations. There is really "nothing" as pure 12V DC ;-) - not even a fuse ;-). This is my "benchmark" for "clean" power supply.

    73 de Johannes

    P.s. it's amazing to see how the forum awoke from its "hybernation" since the anouncement from DB2OS of the opening ceremony of the transponder on 14. februray....

  • DL5MGD Unfortunately there is no (hardware) lock detect (like e.g. in the 4046) and/or in the more sophisticated PLL's like ADF's.

    As an idea I tried to determine if the stability of the VCO-voltage could be an alternative indicator of a lock. I don't use an Arduino (ATMega328p) but a nice and small 'tuned' opamp, aka ATTiny13. So I had to code the RS232 routines etc (because flash is only 1024 bytes).

    Below my first trail. My phase detector is max 4V out, so ADC resolution is 3.9 mV.

    Measuring the VCO-voltage with my multimeter, it remains stable within +/- 10 mV
    (I can't measure it more accurately) when a lock is achieved. So, my reasoning was that 3*3.9 mV (= ~12 mV) could be a nice figure of merit whether the VTCXO is in lock.

    The first trial seems promising. While starting up the LED doesn't burn. When a lock is achieved the LED burns, and goes on and off for a while. After a few minutes it keeps burning. Maybe an idea for you to try or refine it?

    Code below:

    Edit: for those interested, compete code is here (rename to .ino) gpsdo-ublox-1MHz-ref.ino.txt

  • PA3FYM I think that at the beginning the vco voltage oscillates a few times around the setpoint until it reaches it. The oscillation is probably caused by the dimensioning and the resulting damping factor of the loopfilter. Seems really fine and good working; i will impelement it in my code. I will try the idea to sitch off the atmega 328p tonight - after work - to see if this reduces the hum and hiss in the si5351.

  • DL5MGD I connected the junkbox GPSDO to the transverter. Tuning to Es'hail-1 11199.8H beacon the tone sounds a bit 'wobbly' (the jitter is around 4 ticks per second) but I am not unsatisfied with the result. I think that redimensioning of the loopfilter may give hope. After all .. it's a cheap GPS module, not made for generating a stable reference (like the M8-LEA (iirc). Overall feeling: I am satisfied with this nice combination of junkbox parts!