Peak near center freq for noise signal, how to fix it

GNU Radio
1

i use the following command to capture some samples of noise

uhd_rx_cfile -a serial=4759a751 -N 100000 -f 2.48G --samp-rate=2M
noise.dat

then using matlab i read the samples
y=read_complex_binary (‘noise.dat’);

and i do plotfft
plotfft(y,2000000)

the plot shows a peak near the center. I don’t know why. what could i do
on
the samples y so that the peak not appeared?

i attach the figure of plotfft

2

On 04/26/2013 12:52 PM, vegihat vegihat wrote:

the plot shows a peak near the center. I don’t know why. what could i
do on the samples y so that the peak not appeared?

i attach the figure of plotfft

Discuss-gnuradio mailing list
[email protected]
Discuss-gnuradio Info Page
That is the residual DC offset.

UHD allows you to use offset tuning to arrange for the DC offset to be
outside your passband, but you’d have to modify uhd_rx_cfile to use that
feature of UHD.

http://files.ettus.com/uhd_docs/manual/html/general.html#two-stage-tuning-process

3

On 04/26/2013 12:56 PM, Marcus D. Leech wrote:

plotfft(y,2000000)
Discuss-gnuradio Info Page
–
Marcus L.
Principal Investigator
Shirleys Bay Radio Astronomy Consortium
http://www.sbrac.org

Discuss-gnuradio mailing list
[email protected]
Discuss-gnuradio Info Page
Actually, just noticed that uhd_rx_cfile now has a --lo-offset option,
so you don’t even need to modify the code, just specify the LO offset –
typically I set
it to a little bit more than half my sampling bandwidth.

4

if i have understand i need to use the --lo-offset of uhd_rx_cfile

i have gave various values to --lo-offset and only one value has moved
the
DC offset (–lo-offset=1G) and the plotfft gives many small spikes
(maybe
this is caused to aliasing, but i am not sure)

which is the rule (value of --lo-offset) to set the DC offset out of my
band?

2013/4/26 vegihat vegihat [email protected]

5

I remember that some DC is manually added into the frequency point which
can be divided by 5Mhz or 10Mhz? Besides the DC at the your central
freq,
be aware of that if the lo offset setting makes your bandwidth cover
these
frequency point, you still can see the peaks.
Hope I am not wrong, at least It seems that I observed that thing
before.

On Fri, Apr 26, 2013 at 1:44 PM, Marcus D. Leech [email protected]
wrote:

_____________**
Discuss-gnuradio mailing list
[email protected]

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

–

Alex,
Dreams can come true just believe.

6

Well as you advice me, i set the --lo-offset=1M , for the following
command

uhd_rx_cfile -a serial=4759a751 -N 100000 -f 2.53G --samp-rate=2M

–lo-offset=1M noise.dat -v

However the peak is still there. I give below the output of

uhd_rx_cfile.py.
As you can see the Rx DDC frequency is -2M (if I have understood
correctly, it should be 1M,right? )

Any idea what is the wrong..

linux; GNU C++ version 4.4.5; Boost_104200; UHD_003.005.002-47-g4a860d74

– Opening a USRP1 device…
– Using FPGA clock rate of 64.000000MHz…
Using mid-point gain of 45.0 ( 0.0 - 90.0 )
Motherboard: USRP1 (4759a751)
Daughterboard: RFX2400 (no serial, RX2, A:0)
Rx gain: 45.0
Rx baseband frequency: 2.528G
Rx DDC frequency: -2M
Rx Sample Rate: 2M
Receving 100k samples
Writing 32-bit complex floats
Output filename: noise.dat

2013/4/28 Alex Z. [email protected]

7

Veghihat, Hope this can help you
http://lists.ettus.com/pipermail/usrp-users_lists.ettus.com/2012-October/005376.html

On Mon, Apr 29, 2013 at 8:24 AM, vegihat vegihat
[email protected]wrote:

Rx gain: 45.0

case:

Alex,
Dreams can come true just believe.

–

Alex,
Dreams can come true just believe.

8

On 04/26/2013 02:27 PM, vegihat vegihat wrote:

if i have understand i need to use the --lo-offset of uhd_rx_cfile

i have gave various values to --lo-offset and only one value has
moved the DC offset (–lo-offset=1G) and the plotfft gives many small
spikes (maybe this is caused to aliasing, but i am not sure)

which is the rule (value of --lo-offset) to set the DC offset out of
my band?
What I do is set my LO offset to about half my bandwidth, so in your
case:

–lo-offset 1.0e6

–
Marcus L.
Principal Investigator
Shirleys Bay Radio Astronomy Consortium