USRP inserts a peak in the carrier frequency

Hi,

I am using GNU Radio for 2 weeks. I always get in the receiver side a
signal
with a peak in the carrier frequency when I turn on the transmitter even
if
I send a signal made of zeros. You can see it in the picture
http://old.nabble.com/file/p33648622/usrp_carrier_peak.png

In spite of being handy to discover the frequency offset of my signal,
it is
influencing my results when I try to measure the power of the received
signal. I thought this was the DC offset, but I read somewhere that
USRP2
eliminates the DC Offset.

So, do you know what it is and how to remove it?

View this message in context:
http://old.nabble.com/USRP-inserts-a-peak-in-the-carrier-frequency-tp33648622p33648622.html
Sent from the GnuRadio mailing list archive at Nabble.com.

mleech wrote:

Mixers always have some amount of LO leakage at the output port.

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


View this message in context:
http://old.nabble.com/USRP-inserts-a-peak-in-the-carrier-frequency-tp33648622p33648872.html
Sent from the GnuRadio mailing list archive at Nabble.com.

On 07/04/12 10:12 AM, frankist wrote:

eliminates the DC Offset.

So, do you know what it is and how to remove it?

This is LO leakage. What is you setup here–two USRPs with XCVR cards?

Mixers always have some amount of LO leakage at the output port.
Depending on your signal bandwidth,
you can use an LO offset to move that leakage outside of your
applications passband, but there’ll still be
leakage–just not at a place that matters to you.


Principal Investigator
Shirleys Bay Radio Astronomy Consortium
http://www.sbrac.org

Hmm, I tried to use this to change the LO offset:
self.signal_source_0.set_center_freq(uhd.tune_request(self.freq+self.freq_offset,self.lo_off))

However, now I have a new peak and this peak doesn’t disappear when I
turn
off the transmitter.

I am using a USRP2 and XCVR2450

mleech wrote:

Mixers always have some amount of LO leakage at the output port.


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


View this message in context:
http://old.nabble.com/USRP-inserts-a-peak-in-the-carrier-frequency-tp33648622p33648876.html
Sent from the GnuRadio mailing list archive at Nabble.com.

Ok forget, I had to remove the lo offset from the transmitter and not
from
the receiver! (facepalm)

There are still some minor peaks but I think I can remove them using the
same proccess

frankist wrote:

mleech wrote:


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


View this message in context:
http://old.nabble.com/USRP-inserts-a-peak-in-the-carrier-frequency-tp33648622p33648955.html
Sent from the GnuRadio mailing list archive at Nabble.com.

Now I have another problem related to peaks in the fft plot, but in this
case, with no transmitter.

Why does a peak appears at the center of the fft plot for some
frequencies
and not for the others. I mean, I am using a USRP2 to receive just noise
and
plot its fft and when I center my receiver in the frequencies 5.0GHz,
5.5
GHz and 6.0 GHz I have no peaks but for the other frequencies in the
same
band I do have them.

In this case it is not related to LO leakage since I can’t get rid of
them
and I have no transmitter. Also, the peak always appears at the center
of
the plot

frankist wrote:

Hi,
influencing my results when I try to measure the power of the received
you can use an LO offset to move that leakage outside of your


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


View this message in context:
http://old.nabble.com/USRP-inserts-a-peak-in-the-carrier-frequency-tp33648622p33688986.html
Sent from the GnuRadio mailing list archive at Nabble.com.

On 15/04/12 09:27 AM, frankist wrote:

and I have no transmitter. Also, the peak always appears at the center of
the plot

Direct conversion receivers suffer from the so-called “DC anomaly” –
spectral features caused by imbalances and
LO leakage in the analog mixer. LO leakage will contribute to a small
DC offset appearing in the signals coming
out of the mixer, which leads to a “bogus” spectral feature around
0Hz. The USRP-family FPGAs have a
DC-offset removal algorithm in them, but it isn’t perfect, and the
degree of imperfection will change with
center frequency. This “feature” is usually very narrowband in nature,
and for many types of modulations,
it is merely a distraction rather than than a disaster. You can use
offset-tuning on reception to arrange for
the DC-anomaly to appear outside your passband.

I’ve attached a small flow-graph that allows you to see the effects of
DC offset, and phase/amplitude imbalance
in a complex-sampled system, using a 10Khz pure signal with additive
noise.

Oops. problem solved. I had to increase the LO offset in the transmitter
(not
the receiver)! Facepalm

Thanks for the support

mleech wrote:

Mixers always have some amount of LO leakage at the output port.


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


View this message in context:
http://old.nabble.com/USRP-inserts-a-peak-in-the-carrier-frequency-tp33648622p33648967.html
Sent from the GnuRadio mailing list archive at Nabble.com.

Hi. I deleted my last message and created a new thread where I explain
things
properly. I tried to do some offset tuning but the peaks didn’t move so
I
suppose it has another cause. Also, after using your file, it seems the
iq
balance is something that appears in both sides of the spectrum
(negative
and positive) for the same frequency which is not my case. But things
are
better explained in my new thread. I’ve also put some plots there

Marcus D. Leech wrote:

GHz and 6.0 GHz I have no peaks but for the other frequencies in the same
DC offset appearing in the signals coming
I’ve attached a small flow-graph that allows you to see the effects of
http://www.sbrac.org


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


View this message in context:
http://old.nabble.com/USRP-inserts-a-peak-in-the-carrier-frequency-tp33648622p33699885.html
Sent from the GnuRadio mailing list archive at Nabble.com.