So I have the stock python scripts (tx_ofdm and rx_ofdm) located in the
gr-digital examples and put USRP sources and sinks in the appropriate
spots. I left sampling alone (100K for TX, 3.2M for RX) and had the
TX/RX
frequency at 500 MHz. The USRPs used are N210s with WBX daughterboards.
I
tested the USRP used for TX out by outputting a signal to another and
looking at it on the GNU Radio spectrum analyzer. Also, this is through
SMA
cables and a 30 dB attenuator, no over the air transmission. I placed
file
sinks after each block in the process; each file is a binary file. After
multiple runs I felt like something wasn’t working after looking at the
binary files.
When I execute, only the file sinks placed after the Schmidl & Cox OFDM
Synchronizer freq_offset and detect have data in them, at least their
files
have some size to them, implying something has been written. The file
sinks
after FFT (or the header/payload demux) have no data, 0 KB.
When I discovered that the receive didn’t work, I took the grc file
another
user made back in October with both TX mod and RX demod in the same
flowgraph to see if that would work (files attached). Again same issue,
all
the binary files after the FFT have no data so something is getting
screwed
up in my demod.
Has any user on this list run into this issue before or can offer some
insight to why the receiver is failing?
Thank you,
Jon
Attached is the python script, grc file, and screen shot of the TXRX
combo
in GRC.
Hi Jon,
Disclaimer: I haven’t run your code, but based on eyeballing the png
file:
It looks like your “delay” block is not configured properly. The delay
should be fft_len + cp_len which is 80 (instead of 72).
Best,
Aditya
And if you are running over the air, try setting the TX and RX sampling
rate to have the same value.
In any case, I would first run in loopback mode (as shown in your png
file)
and save the over-the-air transmissions for later.
On 03/19/2014 12:25 AM, Jonathan F. wrote:
So I have the stock python scripts (tx_ofdm and rx_ofdm) located in the
gr-digital examples and put USRP sources and sinks in the appropriate
spots. I left sampling alone (100K for TX, 3.2M for RX) and had the
TX/RX frequency at 500 MHz. The USRPs used are N210s with WBX
daughterboards. I tested the USRP used for TX out by outputting a signal
to another and looking at it on the GNU Radio spectrum analyzer. Also,
this is through SMA cables and a 30 dB attenuator, no over the air
transmission. I placed file sinks after each block in the process; each
file is a binary file. After multiple runs I felt like something wasn’t
working after looking at the binary files.
Maybe you’re probably screwing up your tx signal by clipping, although I
can’t say for sure from your flow graphs (where exactly do you put the
UHD sink?) See
http://video.fosdem.org/2014/AW1125/Sunday/Tutorial_OFDM_Packet_Transceivers.webm,
starting around ~27 mins for an explanation, or search the mailing list
archive.
Martin
Thanks for the responses.
When I use the USRPs the UHD sink is placed after the OFDM Cyclic
Prefixer,
right after the Mutliply Const (50m). I haven’t changed any settings
because I remember reading in a few previous emails that the most common
question people ask is “Did you change anything?”
Sadly, the delay wasn’t the case, at least for the loopback mode.
I reduced the amplitude of the signal by an additional 0.02, so the
multiply constant is at 0.03 as opposed to 0.05. It brought the clipping
well within +1, -1. Before the signal clipped past -1 but never rose
above
+1. No effect but I am not ruling out that it wouldn’t help.
The video is interesting enough that I am going to continue watching
after
the first half, maybe I can figure out more from there.
Jon
A good way to test is to use the hier blocks for OFDM (you still need to
reduce amplitude), they will set up everything in a way that will work
for sure. See the loopback example in gr-digital/examples/ofdm.
M
So I found the loopback script, tied the ends together with a multiply
const block and was successful in modding and demodding that sinusoidal
waveform. Added a UHD source and sink, success again. Then I split the
process into two separate GRC files, a TX and a RX script. Everything
works. I am extremely happy right now to see OFDM work. Thank you for
the
nudge in the right direction.
So the next question I have is why I have success with the hierarchal
blocks in the loopback script but not so much with rx_ofdm or that
custom
loopback I posted? Even with a binary file source with the modulated
signal, that rx script won’t work. I wouldn’t say too terribly important
but it is interesting to learn how OFDM works. But for what I am working
on, those hierarchical blocks will suffice.
Thanks again.
Jon
On 03/20/2014 05:14 PM, Jonathan F. wrote:
modulated signal, that rx script won’t work. I wouldn’t say too terribly
important but it is interesting to learn how OFDM works. But for what I
am working on, those hierarchical blocks will suffice.
Well, it’s hard to say from a distance, but you could check the output
on the header path after the header/payload demuxer to see what’s going
on.
M