|Subject:||Re: [Discuss-gnuradio] self-interference with wbx|
|Date:||Thu, 20 Dec 2012 17:41:05 -0800|
Thanks for your answer. I do expect some of the Tx power to leak into my Rx chain, and my application can live with that. What I don’t understand, is why this power is much higher when transmitting packets than when transmitting a continuous tone.
I am using a USRP-2 with a WBX daughterboard that is operating in full duplex mode. Both the Tx and Rx gain are cranked up to their maximum values (tx gain 25 dB, rx gain 30 dB). The Tx and Rx frequencies are about 70 MHz apart (Tx->964 MHz and Rx->892MHz). I’m having a little trouble with the self-interference:
when the Tx chain is sending periodic packets (with 0s in between), this creates very high noise in my Rx chain (up to 0.01 on gnuradio companion scope)
when the Tx chain is sending a continuous pilot tone, there is no noise in my Rx chain (noise is lower than 0.0005 on gnuradio companion scope)
I would like that, when transmitting periodic packets, the noise would be as low as when transmitting a continuous pilot tone. I understand that periodic packets would create noise in a wider band than the periodic tone, but still, my Tx and Rx frequencies are very far apart. Is there some automatic gain in the USRP that is playing tricks here? If so, is there a way to solve this?
Any input is appreciated,
Thanks a lot,
PS: for our application, we had to turn the DUC cordic off. But even with the DUC cordic enabled, I still observe similar trends, so I don’ suppose that would matter. We also had to put the Clock source to “external” to avoid the automatic clock correction (but again, putting it to “Default” didn’t change a thing)._______________________________________________Discuss-gnuradio mailing list
I a fully-engineered full-duplex system, there'll be a duplexor to provide better isolation (70-80dB) for the RX gain an mixer. Your RX front-end is
"seeing" all of the TX energy you're transmitting, and very likely being drive into non-linear operating territory -- just because your mixer is tuned
70Mhz away from the TX, *does not mean* that your RX LNA is "tuned" only to your tuned frequency. You might able to get away with just using
a notch for your TX frequency in front of your RX, but without some actual RF plumbing/systems-engineering, you'll run into "desense" issues like
Any such duplexing arrangements are clearly, *necessarily* application specific. Which is why there's no duplexor or other filters on the daughtercards.
Since they're used for a *huge* variety of applications, there's no way to engineer them to always "do the right thing" regardless of application.
--Marcus LeechPrincipal InvestigatorShirleys Bay Radio Astronomy Consortium
Discuss-gnuradio mailing list
|[Prev in Thread]||Current Thread||[Next in Thread]|