LNB REF signal (TCXO inside the LNB & Ref fed by the duplex) has an effect on the output signal.

  • LNB REF signal (TCXO inside the LNB & Ref fed by the duplex) has an effect on the output signal.

    Recently, I measured and compared the spectrum of the output signal of LNB using different REF modes.

    First measure the TCXO installed inside the LNB.The signal looks clean.Low base noise, good signal-to-noise ratio


    The measurement then provides a REF signal to the LNB through a diplexer.




    The output signal becomes very noisy.The signal-to-noise ratio is significantly lower.Has anyone done the same measurement?

    I hope everyone can see what I post with translation software :) 73 DE BG0AUB

    • Official Post

    The measurement then provides a REF signal to the LNB through a diplexer.

    Can you explain this in a bit more detail: where do you get the new external REF from? Many wideband measurement devices use YIG oscillators due to their wide tuning range but their phase noise is not comparable to a good TCXO and hence should not be used as reference directly.

  • Yes - that's expected if the reference is not of very high quality or it's not being fed in at the right level. Incidentally that LNB looks more like a DRO based one than a PLL based one but I assume it is a PLL. The quality of the 25MHz is critical. Any noise on it is multiplied many times. For DATV use you are often better with a standard unmodified PLL LNB. This often shows up in the MER on the beacon.


    Mike

  • Feng showed me the inside of his LNB. It has an RDA3566 chip, which doesn't have a 'symmetrical oscillator' (aka kalitron), so 'balanced' feeding does not work. It works perfectly for the NXP chips like I revealed/published in 2019 (or so).

    For the RDA-chips there is an 'input' and 'output' pin (where injected levels differ around +20 dB (relative to the input).

  • Feng showed me the inside of his LNB. It has an RDA3566 chip, which doesn't have a 'symmetrical oscillator' (aka kalitron), so 'balanced' feeding does not work. It works perfectly for the NXP chips like I revealed/published in 2019 (or so).

    For the RDA-chips there is an 'input' and 'output' pin (where injected levels differ around +20 dB (relative to the input).

    I tested this balanced method only this year. Here the chip is RT320M.