Re: [Discuss-gnuradio] Receive Sensitivity and Transmit Leakage?

Thanks for the feedback.

In regards to the first part, our receive signal from the BTS is at about 880 MHz and our
sampling rate is set to 4 MHz.
We are definitely using the correct antenna ports, assuming they are labeled correctly
on the hardware. The uhd_fft.py screenshot is attached - this is from a signal generator outputting at -60dBm.

In regards to the transmit leakage, I am referring to leakage out of the transmit port.
I see the CDMA spectrum sitting well under a dominant carrier on the spectrum analyzer
for the ranges I mentioned. Also in the IQ data in Matlab I see the data riding the
dominant carrier in the time domain. All the GRC program does is transmit the receive
samples at a different frequency. The CDMA spectrum hiding under the carrier is of the
correct bandwidth (~1.23MHz)


-----Original Message-----
From: Matt Ettus [mailto:matt@ettus.com]
Sent: Tue 8/2/2011 2:23 PM
To: Delgado, Christopher
Cc: discuss-gnuradio@gnu.org
Subject: Re: [Discuss-gnuradio] Receive Sensitivity and Transmit Leakage?

On 08/02/2011 08:10 AM, Delgado, Christopher wrote:
> Have some questions regarding received signal levels and what seems
> like LO leakage on transmit side. We are using N210 with WBX (UHD
> driver). This issue seems to be pretty independent of carrier
> frequency.
>
> 1) Receive sensitivity - We're cabling in a CDMA2000 carrier (1.2288
> MHz bandwidth) from a femtocell base station we have. The transmit
> power is about -80 dBm. We've done a simple test with GRC to receive
> this signal, write to a file and do PN correlations in Matlab to
> verify the integrity of the data. With the signal level of -80 dBm
> and the rx_gain in the GRC gui set to maximum (38 dBm) the I/Q data
> seems to be about 3 bit values (about +/- 4 or 5). These values seem
> a bit low for this signal level. On other SDR receivers we use (not
> usrp) with max gain, signals over the air from base stations are
> about -90-95 dBm and toggle about 6 bits on ADC. In short, I'm
> wondering if this is expected behavior or if there is another analog
> gain setting to change on receive side.

You didn't state what frequency you are using or what decimation, but
the noise figure should be under 6 dB, and that is the only true measure
of sensitivity.

You can always add more digital gain in the fpga, but I don't think you
need that.

Are you sure you are using the correct antenna input?  Also, the
grand-daughterboard is static sensitive.  In order to check to see if
yours is working, please put a known level sine wave in from a signal
generator, and send us a screenshot of the uhd_fft display along with
the settings you used.


> 2) Transmit leakage We've set up a simple program in GRC to take the
> received IQ from the same base station and retransmit them on a
> different frequency. The receive is in the 800MHz band, and we've
> tried transmit on both 500MHz and 1GHz. Looking at the output on a
> spectrum analyzer as well as just plotting IQ in Matlab, there is
> some strong carrier leakage, in fact the carrier is dominant over
> the signal for all practical ranges of signal level at the input.
> With the input at -30 dBm and the rx_gain set to max 38 dBm the
> signal begins to finally overtake the carrier however this signal
> level is obviously not practical. Furthermore, the gain parameter in
> GRC's UHD: USRP Sink block seems to do nothing based on observing
> live spectrum of the transmit signal. The DC offset of the receive
> data seems to be negligible - we're writing it to a file before it
> goes out. Wondering what could be causing such strong LO leakage.


I am confused here.  Is this carrier leakage out of the transmit port,
or leakage of TX into the RX?

Matt