Hello, I was interested to test tx_ofdm.grc and rx_ofdm.grc from
gr-digital, at the same time using USRP N200 devices.
TX OFDM:
The changes that I did were, removing throttle block and adding an UHD
Sink instead, with center frequency 430 MHz, which is the frequency band
for which my department has license to transmit.
RX OFDM:
In this case I removed all blocks before throttle (included), and placed
there an UHD Source at the same center frequency.
Flow graphs attached in the files.
I monitored the output signal and the received signal. I also used a
spectrum analyzer to check the actual signal that is being transmitted.
Pics are attached.
In the tx signal measured by a spectrum analyzer, we can see that the
spectrum is not that we desire. I would like to ask if I am doing
something wrong with the configuration. Before transmission there was no
signal at that frequency and tx and rx antennas are suitable for the
carrier frequency.
In addition I tried to transmit a file instead of just random bits. In
reception I was not receiving the file. The config was just adding a
file sink at the end of the rx block and a file source at the beginning
of the tx block. Could you please help me with some tip?
Hi Gabriele,
have you tried adding more gain on the TX side?
Also, you will need to increase sampling rate. Right now, you’re using
100kHz, which is 1/1000 of the master clock rate of the N200. If I’m not
mistaken, the maximum decimation for the N200 is 512, and hence the
minimum rate would be 100MHz/512 = 195.3125kHz. You probably just want
to use 200kHz or something similarly practical.
I hope that helps!
Best regards,
Marcus
On 25.06.2015 06:07, Marcus M. wrote:
Hi Gabriele,
have you tried adding more gain on the TX side?
…or less gain?
Also, you will need to increase sampling rate. Right now, you’re using
100kHz, which is 1/1000 of the master clock rate of the N200. If I’m not
mistaken, the maximum decimation for the N200 is 512, and hence the
minimum rate would be 100MHz/512 = 195.3125kHz. You probably just want
to use 200kHz or something similarly practical.
Or even more, like 1-2 Msps.
I haven’t tried a combo of N-Series specifically with this example, but
I’ve tried X-Series + B-Series, so I’m confident it’s all OK. However,
OFDM is sensitive to distortions, so you need to play around with the
amplitude multiplier before tx’ing as well as the gain for your specific
setup to get a good transmit spectrum.
Since you have a spectrum analyzer, that’s pretty useful for getting
those settings right.
You might also try playing with the rolloff length and/or adding more
empty carriers left and right of your tx spectrum while experimenting
with settings.
Cheers,
Martin
Thanks Marcus and Martin. I finally increased the sample rate up to 1
Msps and decreased the amplitude multiplier. Now I can see a clear ofdm
spectrum with no distortion at all.
Did some of you tried file transmission from one end to the other? I did
it for GMSK before. Now when it comes ofdm and I open the receiving file
while transmitting there is no printed information in the received file.
In that configuration I replaced the random source by a file source and
a file sink after constellation decoder.
Again thanks for your help. It was so useful
Hi,
Gabriele, do you have news about this? I am trying to do the same.
And I have another question (if Marcus can help) … what is the
modulation
in this example (QPSK, QAM)?
Sorry for my bad english
Thanks
–
View this message in context:
http://gnuradio.4.n7.nabble.com/OFDM-example-from-gr-digital-with-UHD-tp54446p54503.html
Sent from the GnuRadio mailing list archive at Nabble.com.
Hi Gabriele,
if there is no output, chances are the receiver side packet detector
doesn’t see any valid packets.
Make sure tx and rx parameters match, and adjust the RX gain. Make sure
no under/overflows happen. There’s quite a lot of things that can go
wrong in a real-world transceiver, but these are the most likely to
happen in my experience.
Greetings,
Marcus
Hi Marcus,
Thanks so much for the answer. But like in the screenshots of Gabriele,
the
blocks OFDM Transmitter e OFDM Receiver are off. And I can’t visualize
four
symbols (QPSK) or 2 symbols (BPSK) in the output of QT Constellation
just a
big one (see the screenshot), is this correct?
Thanks again.
http://gnuradio.4.n7.nabble.com/file/n54530/ofdm_tx-_constellation.ofdm_tx-_constellation
–
View this message in context:
http://gnuradio.4.n7.nabble.com/OFDM-example-from-gr-digital-with-UHD-tp54446p54530.html
Sent from the GnuRadio mailing list archive at Nabble.com.
I am plotting the signal after the Multiply Const Block … the signal
that I
will transmitt with USRP.
Isn’t correct?
http://gnuradio.4.n7.nabble.com/file/n54534/ofdm_tx.png
–
View this message in context:
http://gnuradio.4.n7.nabble.com/OFDM-example-from-gr-digital-with-UHD-tp54446p54534.html
Sent from the GnuRadio mailing list archive at Nabble.com.
On 30.06.2015 10:41, Jose Perez wrote:
Hi Marcus,
Thanks so much for the answer. But like in the screenshots of Gabriele, the
blocks OFDM Transmitter e OFDM Receiver are off. And I can’t visualize four
symbols (QPSK) or 2 symbols (BPSK) in the output of QT Constellation just a
big one (see the screenshot), is this correct?
Are you plotting the output of the IFFT by any chance?
M
On 30.06.2015 11:06, Jose Perez wrote:
I am plotting the signal after the Multiply Const Block … the signal that I
will transmitt with USRP.
Isn’t correct?
If what you want to plot is your constellation, then no, it’s not. This
is all modulated, pulse shaped and all.
Cheers,
M
So … Is it possible to see the Constellation points like QPSK in this
flowgraph?
Sorry for so many questions!
Thanks Martin!
–
View this message in context:
http://gnuradio.4.n7.nabble.com/OFDM-example-from-gr-digital-with-UHD-tp54446p54537.html
Sent from the GnuRadio mailing list archive at Nabble.com.
On 30.06.2015 12:12, Jose Perez wrote:
So … Is it possible to see the Constellation points like QPSK in this
flowgraph?
Sorry for so many questions!
On TX, you can plot the complex symbols before they go into the carrier
allocator. On the receive side, you can do the same after the carrier
serializer. This won’t exactly plot 1 OFDM symbol at a time, if that is
what you want.
Also, in both cases, the output will be pretty boring. On TX, it’s just
your constellation; on RX, it’s already equalized, so it’s also just the
constellation.
M
Jose,
we’re talking about this flow graph, right?
http://gnuradio.4.n7.nabble.com/file/n54534/ofdm_tx.png
In this flow graph, you’re trying to find the individual symbols in time
domain; that’s not possible, they only exist clearly distinguishable in
frequency. That is why there is an DFT (in this implementation, the FFT)
in OFDM!
Best regards,
Marcus
