Hi Marcus, Hmm, I think a duplexer is quite challenging because it would need an unphysical Q. This works for cellular with dedicated up-link and downlink.
For a full-duplex system with same up/downlink (as in RFID), we have for example uplink = A*cos(2*pi*900MHz*t) and downlink = B*m(t)*cos(2*pi*900Mhz*t) with m(t) ~ cos(2*pi*400kHz*t) Desired signal: 900.4 MHz, undesired signal: 900 MHz. Thanks, Lukas > Gesendet: Montag, 04. Januar 2021 um 23:49 Uhr > Von: "Marcus D Leech" <patchvonbr...@gmail.com> > An: "Lukas Haase" <lukasha...@gmx.at> > Cc: USRP-users@lists.ettus.com > Betreff: Re: [USRP-users] X310/UBX as a monostatic transceiver (e.g. RFID > reader)? > > If the TX and RX frequencies are a few 100kHz different then what you want is > a duplexor arrangement where the TX frequency is strongly attenuated ahead of > the RX. > > This is how repeaters work when multicoipled to a single antenna. The TX > frequency is often attenuated 90dB in the RX path. > > Sent from my iPhone > > > On Jan 4, 2021, at 11:33 PM, Lukas Haase <lukasha...@gmx.at> wrote: > > > > Hi Marcus, > > > >> On 01/04/2021 12:25 PM, Lukas Haase wrote: > >> In an RFID system, the RX is *designed* to be up-close-and-snugly with > >> the other ends TX. There's NO reason to have an input > >> stage that is sensitive in the usual RF sense. So, completely > >> different engineering goals than in a typical RF system. > > > > That's not quite true. Sensitivity *is* important (in the RF sense) because > > it defines operating distance. > > Yes, many people claim RFID systems are forward link limited (which was > > true) but state-of-the art tags can have a sensivivity of -24dBm which > > corresponds to over 20m free space, 900 MHz. > > The response from the tag at the reader is around -80dBm. The R2000 chip > > that I mentioned has a sensitivity in excess of that. > > > > Indeed, this is the reason why *sensitivity* (due to self interference) is > > the limiting factor in RFID. > > > >>> I have thought of a limiter already. This could be an option. > >>> It's true, I haven't found limiters with lower power levels. > >>> > >>> Two questions here: > >>> > >>> - How/why would they add to the noise figure? > >> Any limiter diode has shunt capacitance. Which means that the degree to > >> which input power is shunted to ground is proportional > >> to the input frequency and shunt capacitance. ANY attenuation > >> (whether resistive dissipation or shunt-to-ground pathways) in > >> front of the first gain stage adds *DIRECTLY* to the noise figure of > >> that stage. So, let's say you have a nice small-signal LNA with > >> a notional noise figure of 0.5dB, and you put 10dB of loss in front > >> of it--the noise figure now becomes 10.5dB. For RFID type applications > >> this doesn't matter that much--see my "up close and snugly" > >> comments. But for "ordinary" RF receive chains, you generally want > >> to minimize noise figure while maximizing gain and linearity. > >> > >> There are exceptions--for example at HF (below 30MHz or so), the input > >> noise is *utterly dominated* by galactic background noise and > >> atmospherics--there's no point in having an input stage with a noise > >> figure below perhaps 5-10dB. So for HF, input stages tend to > >> be optimized for linearity at higher input levels--because even > >> distant signals can be quite strong at HF--particularly on the lower end. > >> > >>> - The large self-interfere would result in clipping (hard > >>> nonlinearities). Is this any problem for the LNA (gain desensitivisation > >>> etc.) > >> Well. Yeah. That's what the P1dB parameter is all about--the input > >> level at which gain is compressed by 1dB. > >> > >> The overall take-away here is that generic radios (whether they be SDRs > >> or others) should be thought of as *components* in an > >> overall *engineered RF system*. That may mean things like relays to > >> shunt the RX pathway during TX, circulators, attenuators, > >> diplexors, filters, RF-plumbing-in-general. > > > > Yes. > > > > To clarify limiter vs non-limiter. > > My self interfering signal from TX can be up to 20-25dBm. The desired > > signal is a modulation bandwidth (few 100 kHz) away and the receiver should > > be as sensitive as possible to that signal. > > > > Now I have the two options: > > > > 1.) Adding a normal attenuator of 40dB. This ensures the USRP input is > > always less than 25-40=-15dBm but it also reduces my desired signal by > > 40dB. In other wirds, my noise figure worsens by 40dB, as you mentioned. > > > > 2.) Adding an RF limiter with flat leakage +5dBm (I found SKY16602-632LF). > > Afterwards I add a 20dB attenuator to get the max to 5-20=-15dBm. > > > > If I understand your explanation correctly, there is no real difference > > between these two (because the limiter would still crushes signal and with > > it desensitizes the desired signal on top). > > Is this correct? > > > > Would you prefer one option over the other? > > > > Thanks again, > > Lukas > > > > > _______________________________________________ USRP-users mailing list USRP-users@lists.ettus.com http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com
