Other method to inject external REF into LNB!

  • Bought a second hand Triax110 dish with three LNB's for 20 Euro's total. Two LNB's were Inverto 'Red Extend' (never heard of this type). After opening it, it seemed that they contain a 25 MHz crystal, so a good candidate to submit to a novel external REF injection method.

    Now both xtal pins are fed with the REF signal with opposite phase.

    A friend of mine says this improves phase noise of the PLL.

    Just tested the modified LNB and it works excellent! Look also here.

  • No, that is not necessary. The internal oscillator has an 'input' and an 'output' and is sensitive enough to pick up the REF signal. Having modified a diversity of LNB's I remember that one was satisfied with -20 dBm and gave nominal results.

    First I wanted to throw away the LNB's supplied with the Triax110 dish, but it's like 'bush tucking' to see which (to 'us') 'unknown' LNB's can be modified.

    Beware, I also got a free Triax90 (which I am currently using), laid in a garden somewhere in my village for months -after asking I was allowed to pick it up- and that dish also had three LNB's . . . now waiting to be opened what's inside (I also thought to dump them ; -)

    So I forecast I can't say (the most spoken phrase on QO-100): "I am using a modified Octagon LNB" because I don't know the brand (these LNB's have been out for so long that all branding marks are gone).

  • I do not assume that this methode improves phase noise.

    But what may improve it, is the right LO level.

    One pin is an output pin and one pin is an input pin.

    Ususally between this pins is an c-mos amplifier as oscillator in the IC and further stages are internally connected to the output.

    I assume if you connect the "output" side of the coil to ground (via a C for DC-block) you will get the same result.

    But anyhow it is a real nice way for de-coupling the 25 MHz.

    Good Idea !!!

  • About the internal oscillator, I know ...

    This friend of mine, a VSAT developer and designer of LNB's, gave me the hint: to insert the reference with opposite phases.

    What is nice (or peculiar?) is that in my interpretation the in- and output of this -so called- C-MOS oscillator are electrically connected albeit that they are 'isolated' through XL (reactance of the 'secondary' windings).

    AL of my ferrite bead (FB) is 470 nH and I have 4 windings, so L = 0.47 * 4² = 7.52 μH --> XL (25 MHz) = 2π*25*7.52 = ca. 1200 Ω

    Apperantly it works ...

  • I expect that an improved phase noise would improve C/N of the DATV beacon. Interesting would have been to see the C/N before and after the modification.

    My modified HD-BP2 LNB did not show any changes in C/N before and after the ext. LO modification using the Leo Bodnar GPSDO as source.

    But anyway, Remco, it is an interesting idea.

    73 Mike

  • PA3FYM

    Changed the title of the thread from “New method to inject external REF into LNB!” to “Other method to inject external REF into LNB!”.
  • Hey this NXP application-note is high interesting, not only for external XO !

    Thanks to Mike !

    They say : "The TFF1024 LO needs differential signal as reference, therefore the X-tal driver should have two outputs with 180 deg phase shift."

    But I still wonder why ?

    In all LNBs MixerICs the Xtal-oscillator-circuit looks very similar to the external oscillator they describe in the app-note.

    In every LNB I modified up to now, at the osc. output pin automatically appeared a 180deg signal when I applied RF to the osc in pin.

    I think someone really has to compare signal-quality with 2 equal LNBs, one singlepase feeded and one biphase feeded.

    I would do this, but sorry, right now no time for it ....

    Another interesting subject is, that in the app-note they say, that oscillator lines should not go along with IF lines to avoid spurious IF signals.

    They show sideband signals in the diagram.

    Hmmmm .... but that's what we do when we use the single coax way to apply the external XO ....

    In all LNBs I modified up to now I used the two coax mehode.


    Armin DF1QE

  • DF1QE The application note doesn't say anything different I wrote when I opened this thread, or the other way round, I didn't wrote anything different than the application note says.

    Sometimes statements -or information- may sound unbelievable, and we don't understand them, but in this case my friend was right. (he must be, part of his work is designing LNB's for the VSAT industry ; -)

  • Ok, don't take me wrong.

    I did not say that it could not be right.

    But I'm a man who likes to understand why things are the way they are.

    And the question is : Is it right for any chipset ?

    That's why I said someone schould check it out ...

    Anyhow, if I get the time next days / weeks I will do that and let you know the results.



  • Concerning the chipset, that's a valid question. My friend has a Philips background and the app note (for us Dutch people that is still from a 'Dutch' company { Philips Semiconductors was merged into NXP, NXP was sold to Chinese and remainders including FreeScale [ex Motorola] -now formally called Nexperia- are in Nijmegen, Netherlands <-- do you get it ? (I don't ; -) } and my test LNB happened to contain a Philips/NXP/Nexperia/<whatever> chip (TFF1015).

    Other tests with other chipsets greatly appreciated : -)

  • As promised I made a test with 2 equal LNBs, one feeded Osc in-Pad only and one feeded Osc in- and out-Pad.

    Damned, Mc Murphey was my friend in this test ....

    I had 3 very cheap Single LNBs from Megasat "single gold" on stock.

    One of it I shot to death in an earlier experiment.

    They were extremely static-sensitive at the antennas, because the antennas are directly connected to HEMT-FETs.

    Ok, so 2 left for this experiment :-)

    I wound a ferrit bead trafo with enamelled copper wire as PA3FYM described and mouted it like the upper described Methode 1a.

    First Tests in the lab gave good results, but then sudden the PLL stopped working at all .... So what happened ?

    Mesurements showed something horrable : The isolation between primary and secondary windings got bad and so 12V powersupply was directly lead to the oscillator IC. Ouch !

    So here a warning : Methode 1a is very dangerous ! You have to assure a good isolation between primary and secondary winding.

    So the second LNB was shot to shot to death too :-(

    Ok, no problem, they are cheap and I ordered another 3 LNBs.

    Well, one LNB left on stock so let's do it again.

    This time I smoothened the sharp edges of the ferrite bead, inside with a drill and outside with a file. Then I wound it with 0,1mm cotton sheathed enamelled copper wire. Actually I would prefer teflon coated wire, but I do not have it so thin. I made some isolation measurements and all was ok. Then I mounted it in the LNB. This time it works fine and I had to wait for the delivery of further LNBs.

    When they arrived the typical shock came on me as to many other hams who modify LNBs ... It was the same Package but a complete different design.

    Even the chips were different. The old PLL was with a 3566 IC, the new one with a 3568 IC.

    To keep my test comparable I had to modify equal LNBs ...

    This is the original LNB antenna side with antennas grounded.

    (see the very interesting designed filter, I've never seen it this way before)

    And the component side with removed x-tal

    So again I wound 2 more ferrite beads as described before.

    And mounted it as above.

    At both LNBs I placed a 1n chip C at osc.in pin for DC decoupeling.

    In one LNB I connected the other end of the winding to the osc.out pin and in the other LNB I connected it to ground. All x-tal ballast Cs were removed.

    Then both LNBs were succsessively connected to a triplexer to which I feeded 12V DC and 25 MHz from my "indoor TCXO" which I desribed in other article in the forum. Between TCXO and triplexer I placed a 60 Ohm variable attenuator to give diffrent 25 MHz level. The output of the triplexer lead to an Avantest spectral analyzer.

    This were the signals found with an oscilloscope at the osc. in and out pins at the "symmetrical" version.

    No wonder that the level on the in pin is lower because the level on the out pin (other side of the transformer) is in 180 deg. phase shift and works against the transformated signal.

    And this were the signals found at the in pin and to ground version.

    The out pin is left open.

    I find it interesting that the phase shift differs from 180 deg in this version.

    Anyhow to get the pll to lock, in the symmetrical version much more Ref level is needed. And to get a clean noise- and jitterfree signal I even had to remove the internal pi attenuator in the indoor TCXO.

    Here's an analyzer screenshot of the unmodified LNB with x-tal

    (sorry I forgot to set the marker right, so the correction)

    That's the symmetrical version with 100 mV ext. Ref (at the coax)

    A lot of noise to see !

    And here the same with 1V ext. Ref

    Nice low noise and clean signal, as it should be.

    (sorry I forgot to set the marker right, so the correction)

    The "in pin only" version looks the same, but with much less drive power needed.

    So now to the real test in the dish :

    The Oscar100 CW-Beacon with the symmetrical ref feed.

    And the Oscar100 CW-Beacon with the in-only ref feed.

    The "in-only version" signal might look sharper and the noise lower, but that's not true.

    This pictures are momentary captures. When I watch them for a longer time they are very equal. Also all other signals on Oscar100 look equal.

    So here is my fazit : At the LNB Type I modified, it makes absolutely no difference in signalquality to feed ref symmetrical or not.

    The only difference is, that the symmetrical feed needs much more ref power.

    Anyhow it ist still a very nice and easy idea to place a trafo for decoupling the ref. I will keep using the in-only methode.

    Does anyone know where to get 0.1 or 0.15mm teflon coated wire ?

    best 73s

    Armin DF1QE