Re: [Discuss-gnuradio] Inconsistant Gain in X310

[Daigle, Andrew - 1008 - MITLL]

Marcus,

 

The FFT plots for power levels below and above this non-linear power response threshold in Channel 2 have been provided below. Here a 50.1 MHz tone was channeled into the USRP via a splitter. The USRP channels were set with a rate of 1M samples per second and a 50 MHz CF. I’ve tried this in multiple X310s with multiple different basic RX cards and they all have the same response and always in the second channel around ~-45 dBm.

 

 

Here Channel 1 acts exactly as you would expect with the peak 100kHz above DC. The Channel 2 response is almost identical to Channel 1 for powers below -50 dBm and then jumps quite dramatically around -44 dBm. At -40 dBm there is a significant constant difference between the channels. This behavior is mirrored in the time domain (below) where in the first viewgraph the two channels of IQ pairs are identical. Once the power exceeds the power threshold the second channel (purple here) varies wildly compared to channel one. It is almost as if I was saturating an amplifier and the signal is being coupled to a higher order mode? The thing that confuses me the most is that nothing here is changing other than the input power. There is nothing in the software or hardware that I can tell that would be affected by an increasing input power level. Especially one so low compared to the input power limit of -15 dBm? I would also expect the RF chains within the USRP to be identical for each channel so I don’t see why it is only happening in CH2.

 

 

Thanks!

 

-Andrew

 

From: discuss-gnuradio-bounces+address@hidden [mailto:discuss-gnuradio-bounces+address@hidden On Behalf Of Marcus D. Leech
Sent: Monday, November 24, 2014 4:58 PM
To: address@hidden
Subject: Re: [Discuss-gnuradio] Inconsistant Gain in X310

 

On 11/24/2014 10:26 AM, Daigle, Andrew - 1008 - MITLL wrote:

Ben,

 

I have tried injecting both a 50.0 MHz and a 50.1 MHz tone and setting the USRP with a center frequency of 50 MHz and a sampling rate of 1 MSps. The signal generator I am using is clocked with the same 10 MHz reference as the USRP (octoclock). I am currently setting the digital gain to zero (though I have tried a number of different values with no noticeable differences in terms of this non-linearity to input power). I have also tried inserting a LO offset, but haven’t had any real success with that either (outside of removing the small peak at DC). I tried disabling the DC offset but that didn’t do anything either.

 

Looking in the time domain at the IQ for each of the channels I see two things which appear strange to me. First, when I set the USRP center frequency to the same frequency as my tone (50 MHz) I would expect a flat line, but instead I see a triangular wave with a non-zero amplitude (actually it is quite large). When I offset the frequencies I see a very noisy sinusoid which has the wave on it. As the power increases, the channels scale evenly in both noise and intensity, but at a threshold power level (~45 dBm) it appears as though leakage occurs between the main sinusoidal wave and this secondary carrier signal.

 

 

Thoughts? Thanks for taking the time to look at this.

 

-Andrew

Andrew:

What does the following test flow-graph yield?   With the same signal input to both sides (via a splitter):

http://www.sbrac.org/files/test_x310_basicrx.grc

If you're injecting a -50dBm test tone at 50.1MHz, you should see it at 100kHz in the FFT display of each half, and within 1dB of each other.

 

From: Ben Hilburn [mailto:address@hidden]
Sent: Friday, November 21, 2014 7:36 PM
To: Daigle, Andrew - 1008 - MITLL
Cc: address@hidden
Subject: Re: [Discuss-gnuradio] Inconsistant Gain in X310

 

Hi Andrew -

 

What you describe is very strange. There is nothing in the USRP that changes dynamically based on the input power level (there isn't even AGC). The only thing I can think of is that somehow the built-in DC offset calibration is going haywire. Are you setting the center frequency directly to the frequency of your input signal? If you tune with an LO offset, do you see the same behavior?

 

Another question: what is your signal source? Can you lock the USRP and your signal source to the same reference, and then tune such that your input signal is directly "over an FFT bin"? Do you still see different behavior between the two channels?

 

Also, the decimation is done in the FPGA, and must be the same for each channel.

 

Cheers,

Ben

 

On Fri, Nov 21, 2014 at 7:17 AM, Daigle, Andrew - 1008 - MITLL <address@hidden> wrote:

Hello,

 

I am running into a small issue with the X310 USRP. My basic setup involves receiving from two antennas (or at this point CW tones from signal generators) using two basic RX daughter cards.

 

My problem is that when I pump in a -50 dBm CW tone to channel A the FFT plot makes sense in that it is stable with an expected value (floor ~-117 dB and peak -62 dB). As the signal increases the peak increases accordingly (i/e for a -40 dBm CW tone I see a floor of ~-117 dB and a peak of -52 dB). This is consistent all the way up to -20 dBm (I didn’t want to get too close to the -15 dBm limit written on the outside of the USRP). When I move this signal into channel B its FFT mirrors channel A at an input of -50 dBm, but as soon the signal is increased past -45 dBm the noise floor starts jumping from ~117 dB to ~93 dB rapidly and the peak jumps and stays stable 14 dB higher (-38 dB vs -52 dB for the same -40 dBm). Below a -50 dBm input; however, everything is identical between the two channels. I was thinking maybe the decimation rate is changing on the second channel based on the input power (but for some reason not the first). The thing is I am using code loosely based on the rx_multi_samples example and I don’t know if there is a way to specify the decimation rate of each of the daughter-cards as if I was using GNU radio companion (or even if this is even the cause of my problem and I should be looking elsewhere).

 

Any thoughts? Thanks!

 

-Andrew

 

_______________________________________________
Discuss-gnuradio mailing list
address@hidden
https://lists.gnu.org/mailman/listinfo/discuss-gnuradio

 

_______________________________________________

Discuss-gnuradio mailing list

address@hidden

https://lists.gnu.org/mailman/listinfo/discuss-gnuradio

-- 

Marcus Leech

Principal Investigator

Shirleys Bay Radio Astronomy Consortium

http://www.sbrac.org