TCXO Performance issues

  • Hallo ...

    after some time ....the CAT is out of the BAG ...

    the D75F dont work anymore like before ...

    the old ones from 2017 are GREAT ...

    but i have a new source for 27MHz TCXOs with good quality...they are in test now...

    please mail me if you have interest (but they are more expensive but ok)


    73 de michael

  • @DL1GNM: Please have a look to the earlier document from DD1US here in this thread.

    He uses it in an Octagon and instead of 27MHz he uses 26 MHz and with the resulting LO frequency he ends up around 1110 MHz IF (with a few slightly odd kHz, but that can be corrected via calibration in case of receiving with an SDR)

    This shifted LO has the advantage that it can be directly used for DATV also as that resulting IF can be directly received with a SAT-TV receiver.

    73 de Oscar DJ0MY

  • I am very sorry, I need to correct my earlier too optimistic results (70Hz slow drift / 30 minutes) with the TXC 7N type.

    When I did my initial tests there was no wind and overcast sky with constant outside air temperature.

    Today I did another longer term test and the LNB was in plain sunshine and it was a bit windy with varying wind speeds. I had the LNB without the outer plastic case (just metalic part) in the plain wind.

    After initial warmup the TCXO oscillated mostly up/down (this time audible) within 1-2 minutes in a range of around +/- 50Hz with sometimes some slower drift of the average center frequency of up to +/- 150Hz over longer period of time, e.g. 10-20 minutes.

    Since this setup is intended primarily for my portable station I am still OK with it.

    But my initial impression was definitively uch bettter.

    Not sure if this would improve with the outer plastic case mounted. Maybe some additional thermal isolation can't hurt. I will keep monitoring the performance.

    73 Oscar DJ0MY

  • Hallo ...

    i got some measurement result for the alternative 27MHz TCXOs

    DRIFT (ppm) attached.... phase noise is good

    if anyone need .... as replacement for the "bad D75F" delivered after mid of 2017...let me know...


    PLEASE ASK THE RESELLERS ... about the BAD D75F not me (i have some old ones for personal use which working very nice)

    have the resellers verified their products they sell with the given information on my homepage in april 2017 ??

    please ask again ... to me directly if you need an good working and stable LNB for QO-100

    73 de michael dg0opk

  • @ PA3FYM

    stability can maybe corrected (if needed) via software tools ....

    phase noise is build inside the oscillator....AND

    the right LNB is to use withthe GOOD CHIPs for low NOISE ...

    maybe the RDA chip in the old octagons or other LNBs

    so the RX results will maybe very good...

    73 de dg0opk

  • Don't trust devices that have no specifications. Particularly programmable TCXO's.

    As a reference to what can happen with an advertised "TCXO" that is actually some sort of NCO, I made some measurements with two models available on ebay.

    My friend first tried a couple of 25 MHz versions with a "201" LNB and all was fine and dandy with both the NB transponder and a homebrew signal source. These "Ultra Precision Golden Oscillator Clock" devices were advertised as programmable to the customer specification and were programmed for the 25 MHz needed here (for the 9750 MHz LO).

    As most LNB's perform extremely well also with a direct 70cm IF, it was decided to experiment with TCXO's programmed for this IF too, i.e. LO = 10057 MHz, so 25.787179 MHz. This is where the fun started: the signal from the TCXO looks fine on an oscilloscope and the frequency can be verified with a frequency counter, but absolutely nothing can be detected at 10 GHz, nor with the transponder nor with the homebrew signal source.

    So the TCXO's had to go into the lab. The SSB phase noise at 25.000000 MHz is pretty high, but only just low enough to end up with an SNR of about 20 dB (in a 3 kHz bandwidth) with a strong CW signal at 10 GHz (cavity VCO driven Step Recovery Diode 10369 MHz PLL Brick Oscillator with +13 dBm output). But the 25.787179 MHz SSB Phase Noise is about 20 dB higher and after "multiplication" by 390, the 10 GHz LO is nothing but a pure noise hump, i.e. even a very strong signal cannot be detected in SSB mode!

    Self-deception is very easy when using the 25.000000 MHz version as the signal to noise ratio is fairly good. But it takes some effort to realize that the SNR-limiting factor at 10 GHz can be the PLL LNB/TCXO combination and not the transponder per se.

    Actually, if the SNR on the transponder downlink were 20 dB for a given signal, along with the ~20 dB SNR offered and limited by the TCXO * 390 Local oscillator, the total SNR would be only 17 dB. Tolerable, but quite a bit more would be available with a good old ovenized crystal oscillator reference or even the Bodnar GPS Mini.

    Even my quite average R&S SMX signal generator offers better performance than the 25.000000 MHz version of this "TCXO". Yes, the TCXO frequency accuracy and and long term stability are pretty good, but the phase noise is horrendous at the highly multiplied microwave frequencies we desire. It is not even particularly good at the primary frequency. But yes, it was cheap.

    And the 25.787179 MHz version is just a ~1 MHz wide noise hump at 10 GHz.

    Some plots:

    When it comes to affordably priced SSB Phase Noise performance, there still is no competition for a well designed, ovenized crystal oscillator, in use since the early twentieth century ;-)

    Michael, OH2AUE


    "If you have data, you have something, if you do not, you have nothing." (Bengt Hultqvist, SK 24.02.2019)

  • Actually, I recall we (AMSAT-DL, specifically DJ4ZC) were looking at an LD cut crystal affair from the woods of Neuchatel for P5A. It takes some stability to demodulate 3bit/s BPSK from a Mars-orbiting spacecraft transmitting on 10GHz into an omnidirectional antennna in "emergency mode"... Those were the days...


    "If you have data, you have something, if you do not, you have nothing." (Bengt Hultqvist, SK 24.02.2019)

  • Regarding the Connor Winfield date codes: I have just taken delivery, from Digikey, of a D75F 27MHz which has code 1915 and a D75F 25Mhz which has code 1827.

    So, the first one could be either year/week or week/year. The second one can only be year/week. When you compare these codes with those on David's devices, there is no consistency (David's replacement was 26/19). Very confusing!



  • I have given up on D75F as they are not reliable. I have respun the PCB to be able to use an Abracon LLC / AST3TQ-V-25.000MHZ-28 and other oscillators in the same package. Actually the packages are practically the same... so the old board is probably usable with the new TCXO. I added provision for a VCTCXO for fun.


  • Hallo everyone....

    in the picture attached you can see the difference in performance of the old Conner Winfield D75F (2017 and before - upper part of picture) and the D75F oscillators are delivered after around mid/end of 2017 and later (lower part) ....

    Measurement was done in MAY2019 on a 35cm DISH with the QO-100 lower BEACON as reference...

    So i think that are no questions about anymore and i can confirm all the result from David G0MRF...

    I am happy that i got some old ones in 2017 for my OTSLO LNBs and found a replacement source for future (but much more expensive)

    73 de Michael dg0opk

  • dg0opk That's very helpful Michael - thanks. My experience (August 2019) is that Digikey are supplying the D75F (25MHz and 27MHz) with date codes 2018 and onwards. So, that supplier seems to be ok now. 73, Russ.

    Too late now - Connor Winfield have lost my trust and I do not recommend them. As soon as the company knew bad chips had made it into the supply chain they should have immediately withdrawn them and initiated a recall. Instead it was left to users to find out and complain and seek redress individually.

    Meanwhile, there are other possibilities but I think in the long run it is better to have a stable low phase noise source in the shack than out on the LNB.

    Phase noise is not a single number and it's impact depends on the bandwidth of the signal. You can have oscillators that sound bad in CW but work well with DATV and ones that sound excellent in CW but are near useless for DATV. An approximation to the performance is an integration over the signal bandwidth, so close in phase noise is very important at low bandwidths and far out phase noise increasingly important at wide bandwidths. A good test is being able to receive both 33ks and 27500ks with an MER that is realistic with respect to the SNR.

    I was using a modified Octagon LNB on my 1.2m dish and it's performance on Badir BBC was several dB worse than an unmodified identical LNB. That was driven by 25MHz from a Leo Bodnar GPS locked source. It turned out the phase nose was dependent on the precise reference frequency, presumably due to the combined reference phase noise and PLL phase noise. With a 25 MHz reference, this LNB could only manage an MER of 9dB on the beacon with my 2.4m dish. Changing the reference to give a 9.5 GHz LO improved the MER to 10.6 dB. I am sure each case if different but it's worth spending some time tuning. It might be best with a good quality 100MHz Ovanised reference divided by 4. Such references are often available in otherwise obsolete test equipment and because they are often 30 years+ old, their stability is excellent.

    If using the single cable solution with the reference injector board mentioned above, adding a low pass filter to the reference is also a good idea, especially if you are using a low IF.