[USRP-users] problem with USRP example on X310 with CBX-120

Jim Hunziker jhunziker at bcisensors.com
Wed Oct 29 15:29:53 EDT 2014


OK, as Frank said in a separate email, the cal patch didn't seem to fix our
problems. The only thing we're using to control the X310 (and CBX-120) is
txrx_loopback_to_file.

Attached is a zip file with a log of what we're doing and some images to go
along with it. Our X310 isn't hooked up to anything, and we're using CAL as
the receive antenna. (We've tried this same test with a loopback cable and
RX2, but it doesn't make a difference.)

The fft.png image is an FFT of the received data, zoomed to baseband and
the 1 MHz signal that was up- and down-converted to 5.6 GHz. You can see
the 1 MHz signal at the correct frequency. The other images are showing the
quality (or lack thereof) of the received signal. Not long after the
capture starts, the 1 MHz signal is no longer detectable.

In the test.log file, you can see how we're running txrx_loopback_to_file
and what gets spit out as it runs. You can also see the commands that I'm
using (in Python, with NumPy, though the commands are nearly the same as
their MATLAB equivalents) to generate the images.

The problem does not seem to occur (or at least not as badly) when we
capture the transmitted data with external test equipment. It also doesn't
seem to occur when we capture a signal from a signal generator with the
X310. All tests, even these last two, were done using
txrx_loopback_to_file. So our best guess is that something weird is
happening inside the X310 when it is being used to do a full duplex TX/RX.

If you like, I can also provide usrp_samples.dat.

-- 
Jim Hunziker
BCI Systems and Software Engineering
jhunziker at bcisensors.com

On Mon, Oct 27, 2014 at 4:05 PM, Jim Hunziker <jhunziker at bcisensors.com>
wrote:

> I found the problem. I had to run dos2unix on the patch first. I'll get
> back to you on how it affects our tests. Thanks!
>
>
> --
> Jim Hunziker
> BCI Systems and Software Engineering
> jhunziker at bcisensors.com
>
> On Mon, Oct 27, 2014 at 1:32 PM, Michael West <michael.west at ettus.com>
> wrote:
>
>> The patch should work for any version of UHD from 3.7.0 forward.
>>
>> Michael
>>
>> On Mon, Oct 27, 2014 at 7:48 AM, Jim Hunziker <jhunziker at bcisensors.com>
>> wrote:
>>
>>> Thanks. It looks like this patch isn't against 3.7.3. We're not running
>>> against the master branch right now. Is this patch intended for 3.8-rc?
>>>
>>>
>>> --
>>> Jim Hunziker
>>> BCI Systems and Software Engineering
>>> jhunziker at bcisensors.com
>>>
>>> On Fri, Oct 24, 2014 at 5:55 PM, Michael West <michael.west at ettus.com>
>>> wrote:
>>>
>>>> Hi Jim,
>>>>
>>>> An experimental UHD patch was sent to your colleague to address the
>>>> calibration issue.  Please let me know if you need the patch sent to you
>>>> directly and I will get it to you.
>>>>
>>>> Best regards,
>>>> Michael
>>>>
>>>> On Thu, Oct 23, 2014 at 10:40 AM, Jim Hunziker via USRP-users <
>>>> usrp-users at lists.ettus.com> wrote:
>>>>
>>>>> Marcus Leech responded to me off-list that I was using gain settings
>>>>> that were too low. Setting them to 10 greatly improved my signal, though
>>>>> there's still a carrier component that's bigger than my signal and not
>>>>> getting removed. A colleague tells me he's working with Ettus regarding
>>>>> this problem.
>>>>>
>>>>>
>>>>> --
>>>>> Jim Hunziker
>>>>> BCI Systems and Software Engineering
>>>>> jhunziker at bcisensors.com
>>>>>
>>>>> On Wed, Oct 22, 2014 at 4:42 PM, Jim Hunziker <
>>>>> jhunziker at bcisensors.com> wrote:
>>>>>
>>>>>> It looks like some of my problem was plotting abs(sig_c) rather than
>>>>>> real(sig_c), but the signal still doesn't look great.
>>>>>>
>>>>>>
>>>>>> --
>>>>>> Jim Hunziker
>>>>>> BCI Systems and Software Engineering
>>>>>> jhunziker at bcisensors.com
>>>>>>
>>>>>> On Wed, Oct 22, 2014 at 12:15 PM, Jim Hunziker <
>>>>>> jhunziker at bcisensors.com> wrote:
>>>>>>
>>>>>>> Hi. I have an X310 with a CBX-120 and a GPSDO, and I'm running the
>>>>>>> txrx_loopback_to_file example with UHD version 3.7.3 like this:
>>>>>>>
>>>>>>> ./txrx_loopback_to_file --tx-args addr=10.10.10.90 --nsamps=250000
>>>>>>> --tx-rate 2500000 --rx-rate 2500000 --tx-freq 56000000
>>>>>>> 00 --rx-freq 5600000000 --rx-ant CAL --wave-type SINE --ref internal
>>>>>>> --wave-freq 1000
>>>>>>>
>>>>>>> (It makes no difference if I use the RX2 antenna and a loopback
>>>>>>> cable.)
>>>>>>>
>>>>>>> With a sampling rate of 2.5 MHz and a sine wave of 1 KHz, I expect
>>>>>>> to see a sinusoidal cycle every 2500 samples. I'm plotting them with GNU
>>>>>>> Octave and the following commands:
>>>>>>>
>>>>>>> fid = fopen("~/uhd/host/build/examples/usrp_samples.dat", "rb");
>>>>>>> [sig, count] = fread(fid, Inf, "int16");
>>>>>>> sig_c = sig(1:2:end) + j * sig(2:2:end);
>>>>>>> plot(abs(sig_c));
>>>>>>>
>>>>>>> The signal seems very noisy, and it's pretty much gone towards the
>>>>>>> end of the capture. Am I doing something wrong?
>>>>>>>
>>>>>>> --
>>>>>>> Jim Hunziker
>>>>>>> BCI Systems and Software Engineering
>>>>>>> jhunziker at bcisensors.com
>>>>>>>
>>>>>>
>>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> USRP-users mailing list
>>>>> USRP-users at lists.ettus.com
>>>>> http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com
>>>>>
>>>>>
>>>>
>>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ettus.com/pipermail/usrp-users_lists.ettus.com/attachments/20141029/8fe54acc/attachment-0002.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: usrp_problems.zip
Type: application/zip
Size: 142739 bytes
Desc: not available
URL: <http://lists.ettus.com/pipermail/usrp-users_lists.ettus.com/attachments/20141029/8fe54acc/attachment.zip>


More information about the USRP-users mailing list