Hi Matt and List,
Matt had a sound card that wasn't working with 700D. The symptom was
the poor scatter, and timing wandering about. As there was a
possibility of a bug in 700D I worked with him to investigate the
problem. After a few more tests I am now satisfied it's a hardware
fault - the card is randomly gaining or losing samples.
The bug is subtle, as other modems can run through the sound card.
Re (1), 700D uses a coherent (phase locked) demodulator that has 160ms
frames and it (not unreasonably) expects the sound card to sample
consistently over that period. 1600 used the FDMDV modem which has much
shorter frames (20ms), and non-coherent detection so is less sensitive
to this particular hardware fault.
Matt - the on-air experience from many users is that 700D is far more
robust that 1600.
-/-
Re(2) in OFDM each carrier is unfiltered PSK (hard keying is a good
analogy - the waveform changes abruptly) so has wide side lobes. In
contrast, the earlier modems had filters for each carrier. The OFDM
sidelobe energy is not required.
I added a somewhat buggy band pass filter to 700D to address this, and
Jim Ahlstrom (of Quisk fame) did some fine work to get it working
correctly, in particular for complex valued output signals.
The BPF is on by default in freedv_api.c, but freedv_tx() was switching
it off. Thanks for pointing that out Matt, I have checked in a fix to
SVN (rev 3742).
The BPF can be switched using the freedv_tx command line options or
Tool-Options on FreeDV GUI. I have attached a plot of the output
spectrum I get on Audacity using:
$ ./freedv_tx 700D ../../raw/ve9qrp.raw - | sox -t raw -r 8000 -s -2 -
~/Desktop/ve9qrp_700d.wav
- David
On 13/07/18 21:39, Matt Roberts wrote:
> Hi All,
>
> Earlier in the list, I had described finding a sound card that seemed to
> work fine for the 1600 mode, but had some intermittent errors on 700D,
> even on a very high SNR back-to-back test between computers.
>
> I did some diagnostic recordings, comparing the device to one that
> doesn't present the same symptoms, and sent them to David. He
> identified a subtle difference between the behavior of the two devices,
> and using that as a starting point, I have continued to investigate what
> exactly is the difference between the two devices -- and that
> investigation is still ongoing.
>
> However, during that conversation, I asked David a couple of questions
> that really belong on the list here, so I'm hoping David will be willing
> to provide a little commentary on either or both of these:
>
> 1. What details about the 700D mode's modem layer make it more
> dependent on things like frequency linearity, frequency stability,
> phase stability, etc., above and beyond those of older modes like 1600?
> 2. I noticed that the keying spectra of the 700D mode contains much
> more energy outside the 1kHz to 2kHz frequency "envelope" than the
> 1600 mode did. I can't post pictures to the list, but below are a
> couple of links to the different measured spectra. Is the 700D mode
> dependent on having this additional spectral energy to demodulate
> the symbols properly? E.g., is this similar to the difference
> between the hard-keyed RTTY spectrum distribution vs. that of a
> wave-shaped PSK signal? Or is there some other reason why the mode
> appears "broader"?
>
> 1600 Spectrum: http://www.kk5jy.net/snr-freedv/codec2-1600.png
>
> 700D Spectrum: http://www.kk5jy.net/snr-freedv/codec2-700D.png
>
> Thanks ahead of time to David for taking a look at these.
>
> 73,
> Matt
>
>
>
> ------------------------------------------------------------------------------
> Check out the vibrant tech community on one of the world's most
> engaging tech sites, Slashdot.org! http://sdm.link/slashdot
>
>
>
> _______________________________________________
> Freetel-codec2 mailing list
> Freetel-codec2@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/freetel-codec2
>
------------------------------------------------------------------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
_______________________________________________
Freetel-codec2 mailing list
Freetel-codec2@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/freetel-codec2
