Re: [Discuss-gnuradio] RFX-400 RF sensitivity

[Greg Troxel]

  We are using the RFX-400 board (without housing) with an appropriate
  aerial connected to TX/RX.

I'll assume appropriate means a 1/4 lambda vertical without a real
ground plain.  Still, that's ok. 

  Using various examples (usrp_fft.py,
  usrp_wfm_rcv.py and usrp_nbfm_rcv.py) we attempted to receive a signal
  over-air, generated by a handhold portable radio unit. Unfortunately
  We were unable to receive our signal.

I did this using usrp_nbfm_rcv.py and a NBFM signal from an amateur
440 MHz transceiver, with a loaded 1/4 wave (stock antenna from a
Yaesu VX-1R, to be precise) as the receive antenna.  With 0.5W into a
similar antenna a few feet away, I got a full quieting signal.  I was
able to copy signals from a repeater that was ~10km away.  The
quieting was noticeably worse than with a conventional receiver (Yaesu
FT-709R, made ~1986, specified receive sensitivity roughly 0.21 uV for
NBFM, 10 dB SINAD postdetection).  This is about -120 dBm.

  We then connected a signal
  generator directly to TX/RX and experimentally determined that the
  RFX-400 has an RF sensitivity of approximatively -105 dBm.

Without specifying the receive modulation and especially the detection
bandwidth, this is not very meaningful.  NBFM demodulation for 10 dB
SINAD is quite different from MDS (minimum discernable isgnal) at 500
Hz bandwidth for CW.  I'll assume you mean the level that caused
perceived quieting for NBFM.

  Is this consistent with other peoples findings?

That matches my subjective impression pretty well.  I would have
guessed -110 dBm for NBFM at minimum workable SNR, but I didn't fire
up my HP 608E, which takes an hour to warm up and stabilize.  I should
do this some time; there's humor in using such a device to
characterize a software radio.

  Is it possible that
  the high speed digital signaling on the USRP motherboard is
  significantly reducing the sensitivity of the daughter-board?

Yes, it is possible.  You should examine the circuit diagram and
determine what kind of noise floor you would expect, given the NF of
the first RF amplifier and losses before it; your experiment is not
enough evidence to conclude that there's any such interference.