USRP Clipping?

GNU Radio
1

I am using N210 USRP’s with WBX daughtercards connected with a loopback
kit
implementing a 8-PSK waveform. When I start transmitting, I get a clean
constellation. After a short time however (~5 sec) it appears that the
receiver temporarily loses phase lock. The eight point constellation
degrades such that many points appear in a circle with the same radius
as
the points. The FFT of the signal shows that the noise floor temporarily
rises. It’s as if the amplifier begins to clip or some other non-linear
process is occurring. After such a ‘burst’ the constellation appears
clean
again until another burst occurs. Has anybody else observed this
behavior?
Any ideas to what could be the underlying issue?

I am using an multiplier to set the signal amplitude to 0.3 and have the
gains of the USRP’s set to around 7dB. If I use a higher amplitude or
greater gains, performance degrades quickly. Are there any rules of
thumb
for these settings? Why does the performance degrade with increased
values?

2

On 09/19/2013 10:45 AM, Isdren Gineer wrote:

could be the underlying issue?
[email protected]
Discuss-gnuradio Info Page
You’re either getting ADC clipping, or your receiver is entering a
non-linear operating region where harmonic distortion starts to happen.

3

Can you describe for us what the “loopback kit” comprises? Theres
hopefully significant attenuation in the signal path.

4

The loopback https://www.ettus.com/product/details/LPBK-KIT kit is a
30dB
attenuator and an SMA cable sold by Ettus R…

5

OK, good, I’d never noticed that on the price list!
So 30db is a good value to protect WBX from physical damage in a
loopback configuration. It does however leave you with a potentially
very high signal level at the RX and so it makes sense that your setup
appears to be operating at the edge of linearity, despite the modest
configured gains. My first guess, assuming that your TX is working as
you believe is that your setup is subject to some additional
environmental strong and intermittent RFI, though I concede in a cabled
loopback setup that theory is harder to support. I’ve seen very similar
FFT results when working weak signals without enough out of band
filtering. Are you working near a RADAR? Laptop with WiFI right next to
the USRP? Does a further reduction in gains solve your problem?
-Ian