Re: [Discuss-gnuradio] Works with GR 3.6, breaks with 3.7

[Luke Berndt]

This is awesome! I just pulled master and I am recompiling it right now. Thanks you both so much for the help.

I clearly still have some tuning work to do either way, but it would be great to get move up to 3.7. I will let you know if this clears it up. I have my fingers crossed that it will.

On Nov 4, 2014, at 8:19 PM, Andy Walls <address@hidden> wrote:

On Mon, 2014-11-03 at 22:13 -0500, Luke Berndt wrote:

Thanks Andy! Good catch - I made the changes you suggested. I was just
doing a simple back of the envelope calculation to come up with
channel size. It does look like cleans things up and adding the
waterfall graphs does make it easier to see. Unfortunately, it does
seem to change the decoding. I am still getting the same amount of CRC
errors.

Did the overall signal look right, like something that should be decodable? 

So I put the output of the FLL into a quad demod block and compared its
output to the PLL Freq det output.  They are pretty much in line, except
the quadrature demod block output shows occasional burst of noise for a
few symbol times where the transmission must have stopped and restarted.
The PLL freq detect block just makes something up during these times.
So that seems generally OK.

Anything else I should try?

Your PLL freq detector block output is swinging between -2 and 2.  You
might want to reduce the max and min freq by a factor of 2, so that the
clock recovery block sees inputs limited to +/- 1.0.  See Nick's
recommendation here:
http://lists.gnu.org/archive/html/discuss-gnuradio/2014-10/msg00473.html

So after i did that, what I really noticed is that the correlator is
just not marking preambles properly.  See the top plot of the attached
to window shots:  Preamble_marked_late_twice.png and
Preamble_marked_early.png

If the deinterleaver and/or crc block isn't searching around for where
the preamble really is, well that would explain it.

Regards,
Andy

- Luke

Hi Luke,

I have not built the SmartNet blocks yet, but:

Your low pass filter looks way too wide and you'll get aliases when you
decimate by 185.  2 Msps / 185 = 10.81 ksps, so the Nyquist frequency is
5.405 kHz.  I winged this is as a low pass filter and things looked
better:

firdes.low_pass_2(1, samp_rate, 4500, 1000, 60, firdes.WIN_HANN, 6.76)

The 60 dB down may be overkill, you can make it smaller for a filter
with less delay.

I found an offset slider value of -15k made things look about centered
manually.

You might want to put a waterfall sink before and after the FLL
Band-Edge filter to observe how it is making the spectrum wobble around
a little. If you change your offset slider, you can see the FLL
band-edge filter centering things back up; so that looks like it is
working.

Regards,
Andy

<Preamble_marked_late_twice.png><Preamble_marked_early.png>