Hello everyone,
I am trying to capture a few seconds of over the air HDTV on a file and
then
decode using the files in gr-atsc module according to the instructions
in
the README at gr-atsc/src/python. I am using the following command to
capture the signals off the air: /usrp_rx_cfile.py -R B -d 10 -f 587M -g
75
test_file, where 587MHz is the center of the 584-590 MHZ UHF band. I can
successfully capture about 20 seconds of data without any overruns.
After I
set up the pipes and start the equalizer, it is unable to find field
syncs.
At some gain settings it initially detects field sync but loses it later
on.
Am I lacking enough gain/sensitivity in the front-end of the TV_RX? If
so
would an lna at the antenna help? Or could I be doing something else
that
prevents the equalizer to find field sync? I believe a number of folks
must
have decoded over the air hdtv signals using the TV_RX and USRP. Any
suggestions and help appreciated.
Thanks,
Nirali
On Wed, Oct 10, 2007 at 03:23:51PM -0500, Nirali P. wrote:
Am I lacking enough gain/sensitivity in the front-end of the TV_RX? If so
would an lna at the antenna help? Or could I be doing something else that
prevents the equalizer to find field sync? I believe a number of folks must
have decoded over the air hdtv signals using the TV_RX and USRP. Any
suggestions and help appreciated.
Thanks,
Nirali
What does the output of usrp_fft -d 8 -f 587 look like?
Set it to average, then post a link to a screenshot. It should be
mostly flat, with the pilot tone visible at the bottom end.
[In looking at interp.py, it could be modified to run something like
4x faster by using gr.interp_fir_filter_ccf instead of what’s in
there. Also, decimating by 10 seems a bit agressive. By 8 would give
you more usable signal, though would require changes in the interp.py
code.]
Eric
On Wed, Oct 10, 2007 at 01:56:11PM -0700, Eric B. wrote:
What does the output of usrp_fft -d 8 -f 587 look like?
Should be
What does the output of usrp_fft -d 8 -f 587M look like?
Eric
Hello Eric,
[In looking at interp.py, it could be modified to run something like
4x faster by using gr.interp_fir_filter_ccf instead of what’s in
there. Also, decimating by 10 seems a bit agressive. By 8 would give
you more usable signal, though would require changes in the interp.py
code.]
I have gotten a set of RAID 0 drives installed and now I can write with
a
decimation factor of 8 without any overruns. I would like to modify
interp.py to appropriately interpolate and lowpass filter the 8 Msps
signal
from the USRP. Would it be sufficient if I interpolate by 2? Or do I
need a
higher, maybe fractional interpolation rate? I am also unsure of what
the
filter cut-off frequency would need to be after the proper
interpolation. If
you have any suggestions it would be a great help!
Thanks,
Nirali
On Wed, Oct 10, 2007 at 06:34:40PM -0500, Nirali P. wrote:
Your thoughts appreciated.
FYI, if you turn on the “Average” option (right click in fft display),
it’s
even easier to see what’s going on.
Hi Eric,
Attached is the screenshot for the spectrum at 587Mhz. I will soon get a
link so next time I won’t attach!
It looks like there is some multipath towards the upper end but not a
deep
notch or anything severe that would prevent field sync. The SNR does not
seem that bad either at 57dB gain. I will try to move my antenna to
another
place to see if I can get a completely flat spectrum.
With a gain setting of 50 I get slightly higher SNR and the field sync
stays
on for a longer time before the equalizer loses it.
Your thoughts appreciated.
Thanks,
Nirali
What does the output of usrp_fft -d 8 -f 587M look like?
Eric