Benchmark_xx.py Transmits Images at +/- 0.5MHz When Center Freq Is 2.4GHz [USRP N200 and XCVR2450]

Hi All,

I have a pair of USRP N200s with XCVR2450 daughterboards running the
latest
GNU Radio version with UHD. This is an odd problem, but I’m wondering if
anyone has seen this before.

When I run the benchmark_rx.py/benchmark_tx.py specified at 2.4GHz, the
transmitter is transmitting images of lower magnitude at +/- 0.5MHz, and
the
receiver isn’t able to decode any radio packets. Some frames are
received,
but they are all “ok=FALSE” with the lengths lower than the length of
the
actual packet sent. These images persist when I set the XCVR2450’s to
tx/rx
within its specified frequency range.

note: the benchmark scripts are from
~/gnuradio/gr-digital/examples/narrowband/

However, when I run the benchmark scripts at “–freq 0” on both sides,
the
transmitter says it is automatically set to “4800MHz” (although this is
outside of the XCVR2450’s range), and the receiver says it is
automatically
set to “2400MHz.” Somehow, the transmitter is transmitting at 2.4GHz and
the
receiver is tuned to 2.4GHz. The signal is clean, and there are no
images in
sight. I used both uhd_fft.py and a spectrum analyzer to check.

Does anyone have any idea what’s going on?

-Jason

View this message in context:
http://old.nabble.com/benchmark_xx.py-Transmits-Images-at-%2B---0.5MHz-When-Center-Freq-Is-2.4GHz--USRP-N200-and-XCVR2450--tp33763401p33763401.html
Sent from the GnuRadio mailing list archive at Nabble.com.

On 05/02/2012 07:42 PM, jasonatran wrote:

but they are all “ok=FALSE” with the lengths lower than the length of the
actual packet sent. These images persist when I set the XCVR2450’s to tx/rx
within its specified frequency range.

Do you have a screencap of the problem spectrum? Maybe from the
perspective of the receiver.

note: the benchmark scripts are from
~/gnuradio/gr-digital/examples/narrowband/

However, when I run the benchmark scripts at “–freq 0” on both sides, the
transmitter says it is automatically set to “4800MHz” (although this is
outside of the XCVR2450’s range), and the receiver says it is automatically
set to “2400MHz.” Somehow, the transmitter is transmitting at 2.4GHz and the
receiver is tuned to 2.4GHz. The signal is clean, and there are no images in
sight. I used both uhd_fft.py and a spectrum analyzer to check.

Well XCVR2450 has a dual band. I think you are seeing some weirdness
with the code either trying to auto select the middle frequency, and the
actual frequency being coerced.

-Josh

Hi Josh,

Thanks for the response!

Do you have a screencap of the problem spectrum? Maybe from the
perspective of the receiver.

I’ll generate a screencap tomorrow morning when I get back to the
equipment.

Well XCVR2450 has a dual band. I think you are seeing some weirdness
with the code either trying to auto select the middle frequency, and the
actual frequency being coerced.

This is exactly what I’m thinking as well. I’ve been trying to find the
trail of relevant pieces of code with no avail.
But the bigger question I’m trying to figure out is why the images occur
when the frequency is set at 2.4GHz
as opposed to having no images when the code is setting the frequency to
2.4GHz on its own.

–start code snippet—
uhd::tune_result_t set_center_freq(double freq, size_t chan = 0){
return set_center_freq(uhd::tune_request_t(freq), chan);
}
–end code snippet–

-Jason


From: Josh B. [email protected]
To: [email protected]
Sent: Wednesday, May 2, 2012 9:14 PM
Subject: Re: [Discuss-gnuradio] benchmark_xx.py Transmits Images at +/-
0.5MHz When Center Freq Is 2.4GHz [USRP N200 and XCVR2450]

On 05/02/2012 07:42 PM, jasonatran wrote:

but they are all “ok=FALSE” with the lengths lower than the length of the
actual packet sent. These images persist when I set the XCVR2450’s to tx/rx
within its specified frequency range.

Do you have a screencap of the problem spectrum? Maybe from the
perspective of the receiver.

note: the benchmark scripts are from
~/gnuradio/gr-digital/examples/narrowband/

However, when I run the benchmark scripts at “–freq 0” on both sides, the
transmitter says it is automatically set to “4800MHz” (although this is
outside of the XCVR2450’s range), and the receiver says it is automatically
set to “2400MHz.” Somehow, the transmitter is transmitting at 2.4GHz and the
receiver is tuned to 2.4GHz. The signal is clean, and there are no images in
sight. I used both uhd_fft.py and a spectrum analyzer to check.

Well XCVR2450 has a dual band. I think you are seeing some weirdness
with the code either trying to auto select the middle frequency, and the
actual frequency being coerced.

-Josh

On 05/03/2012 01:08 PM, Jason T. wrote:

center freq set to 2.4GHz. images +/- 0.5MHz (uhd_fft.py)
http://img801.imageshack.us/img801/3015/imagesonuhdfft.png

I believe that you are seeing the LO of the XCVR2450. The default
behaviour when tuning the transmitter on this board is to tune the LO
with an offset such that it stays out of your passband.

So I dont think is is harming the deframing and demoding of packets. You
might try messing around with some other parameters, gain, symbol rate,
modulation type.

Pictures of Clean Signal:

center freq set to 0; Tx “defaults” to 4.8GHz but is actually 2.4GHz
(uhd_fft.py):
ImageShack - Best place for all of your image hosting and image sharing needs

(oscilloscope version)
http://img801.imageshack.us/img801/4141/img20120503124257.jpg

I think whatever screwy thing is happening to make the tune request clip
to 2.4GHz is affecting the logic that decides the LO offset.

I’m not sure at what level the coercion is happening. Pardon me, I am
away from my USRP :slight_smile: If you run one of the examples that comes with UHD
try tx_waveforms --freq=4.8e9 --rate=1e6 and see the warning it prints
about the tune frequency.

-josh

Hi All,

Here are images of what I’m seeing in my setup.

Pictures of Images/Spikes:

center freq set to 2.4GHz. random image (oscillioscope)
http://img585.imageshack.us/img585/5871/img20120503124151.jpg

center freq set to 2.4GHz. images +/- 0.5MHz (uhd_fft.py)
http://img801.imageshack.us/img801/3015/imagesonuhdfft.png

Pictures of Clean Signal:

center freq set to 0; Tx “defaults” to 4.8GHz but is actually 2.4GHz
(uhd_fft.py):

(oscilloscopeversion)
http://img801.imageshack.us/img801/4141/img20120503124257.jpg

-Jason


From: Jason T. [email protected]
To: “[email protected][email protected]; “[email protected]
[email protected]
Sent: Wednesday, May 2, 2012 10:28 PM
Subject: Re: [Discuss-gnuradio] benchmark_xx.py Transmits Images at +/-
0.5MHz When Center Freq Is 2.4GHz [USRP N200 and XCVR2450]

Hi Josh,

Thanks for the response!

Do you have a screencap of the problem spectrum? Maybe from the
perspective of the receiver.

I’ll generate a screencap tomorrow morning when I get back to the
equipment.

Well XCVR2450 has a dual band. I think you are seeing some weirdness
with the code either trying to auto select the middle frequency, and the
actual frequency being coerced.

This is exactly what I’m thinking as well. I’ve been trying to find the
trail of relevant pieces of code with no avail.
But the bigger question I’m trying to figure out is why the images occur
when the frequency is set at 2.4GHz
as opposed to having no images when the code is setting the frequency to
2.4GHz on its own.

–start code snippet—
uhd::tune_result_t set_center_freq(double freq, size_t chan = 0){
return set_center_freq(uhd::tune_request_t(freq), chan);
}
–end code snippet–

-Jason


From: Josh B. [email protected]
To: [email protected]
Sent: Wednesday, May 2, 2012 9:14 PM
Subject: Re: [Discuss-gnuradio] benchmark_xx.py Transmits Images at +/-
0.5MHz When Center Freq Is 2.4GHz [USRP N200 and XCVR2450]

On 05/02/2012 07:42 PM, jasonatran wrote:

but they are all “ok=FALSE” with the lengths lower than the length of the
actual packet sent. These images persist when I set the XCVR2450’s to tx/rx
within its specified frequency range.

Do you have a screencap of the problem spectrum? Maybe from the
perspective of the receiver.

note: the benchmark
scripts are from
~/gnuradio/gr-digital/examples/narrowband/

However, when I run the benchmark scripts at “–freq 0” on both sides, the
transmitter says it is automatically set to “4800MHz” (although this is
outside of the XCVR2450’s range), and the receiver says it is automatically
set to “2400MHz.” Somehow, the transmitter is transmitting at 2.4GHz and the
receiver is tuned to 2.4GHz. The signal is clean, and there are no images in
sight. I used both uhd_fft.py and a spectrum analyzer to check.

Well XCVR2450 has a dual band. I think you are seeing some weirdness
with the code either trying to auto select the middle frequency, and the
actual frequency being coerced.

-Josh


Discuss-gnuradio mailing list
[email protected]
https://lists.gnu.org/mailman/listinfo/discuss-gnuradio