logo
Most Viewed Last 7 Days
436MHz DATV Up-Converter (140)
DVB-S USB rx (98)
ATV Contest 2005 (61)
1.8M EME Dish Project (60)
23cm DG0VE Amplifier (55)
24GHz LNA box (55)

RTLSDR Si5351A reference



    Cluster Activity

    Spots per uW band today:
    UK spotter or spotted only.

    Solar data:
    SFI
    76
    A-Index
    9
    K-Index
    2
    Exp.K
    0
    Sunspots
    14
    Activity
    qui
    GMF
    qui
    Aurora?
    no

    My Weather

    Date:19/11/17 Time:00:03
    Temp:0C
    Wind Speed:0mph
    Wind Dir:157deg
    Pressure:1018.57mB
    Humidity:118%
    Dew Point:3.6C
    UKHASnet
    RTLSDR Si5351A reference

    This idea came about while trying to make a frequency and time locked receiver to use with my 50MHz chirp beacon.
    The receiver frequency needed to be stable along with the timing reference, 10MHz and 1Hz signals are used from my GPS referenced 10MHz ovened oscillator.

    I wanted to use the Realtek tv donge as the receiver because they are so cheap so needed a stable 28.8MHz reference for it instead of it's own wandering 28.8MHz crystal.

    I was going to use an AD9851 DDS but the output of these at 28.8MHz with a 10MHz reference signal and x6 multiplier is right near the nyquist frequency so hard to filter the alias, you could do it as seen below but did not follow that up..!

    I found some cheap SI5351A oscillator modules on ebay which use a 25MHz crystal reference, i wondered if they would work with a 10MHz reference signal instead (which is way outside the specified input range) and yes they do, but the output is not very clean with many spurious signals about 30dB down from the wanted signal.

    To overcome this i used two 28.8MHz crystals from Realtek dongles and some matching capacitors to filter the output of the SI5351A, I now have a clean 28.8MHz signal with enough amplitude to inject into the Realtek dongle and which is very stable.
    I have not actually measured the accuracy of the frequency at 28.8000xxxxxMHz output, but the main point of this was for a stable reference, a known offset is not a problem. There maybe some pll step size offset in the Si5351A and also probably the same applies for the R820 tuner, so don't expect a perfect readout just enjoy the better stability!

    The arduino code source used to drive the Si5351A was from hans summers: http://www.qrp-labs.com/synth/si5351ademo.html#arduino

    I modified the code to set 28.8MHz
    si5351aSetFrequency(28800000);

    and set the reference to 10MHz
    #define XTAL_FREQ 10000000

    ...and it just worked.


    Spectrum after filtering:



    And a wider span:



    Board photo:



    The not exactly neat filter consists of only 2x 28MHz crystal in series, 3x shunt 47pf capacitors, this was matched to 50ohm output but appears to be ok into the Realtek dongle.

    The 28.8MHz output is then fed into the rtlsdr dongle with the crystal removed (used in the filter!)

    Reference input and 1pps synchronisation











    Last page added:16/08/17 15:45
    M0DTS.co.uk