Hi Matt,
Those were the sort of bugs I spent a few weeks chasing down with the
test teams before 1.3 release. Especially on Windows. The goal is 700D
should hang on indefinitely, indeed it did last time I tried a similar test.
But doing this on a general purpose machine in uncertain state (some
Hams still use XP) running other software on cross platform code has
it's challenges.
Questions:
1/ Are the two PCs you are using running Linux? Which distro?
2/ Are the PCs doing anything else? Network activity, web browsers, mail
fetching, any hard disk activity?
3/ Fresh checkout and build of FreeDV GUI and codec2-dev?
At the bottom of Tools-Options are a bunch of FIFO counters that were
included to debug just such problems, also a portaudio buffer field you
can adjust.
https://svn.code.sf.net/p/freetel/code/freedv-dev/USER_MANUAL.txt
Rev 3689 is bleeding edge, and I am messing with the sound driver code,
but that doesn't explain the problems with earlier versions. I'll try
to repeat your test once I've finished the current work.
Thanks,
David
On 17/06/18 06:23, Matt Roberts wrote:
As I'm continuing to tinker with 700D, I am at the point where I'm
running FreeDV on two full PC computers, with back-to-back audio
cables. This setup works flawlessly for 1600, but when I switch to
700D, I get occasional bursts of errors in the received data stream.
There's no band noise, because the setup is just two PCs connected
together. There shouldn't be much frequency offset other than the clock
differences between the sound cards. The sound cards I'm using are
solid, and I am actively using them with other DSP projects without any
issues. I can run 1600 for an hour or more without a single bit error
with the same hardware.
When I switch to 700D, the decoding runs okay for a while, then I'll get
a burst of a few bit errors, then a second later a few more, then a few
more, then it will go back to behaving. The audio becomes very
distorted, and the out-of-band text at the bottom is also corrupted
during the bursts, which are each sub-second in length.
The waterfall doesn't show any signal distortion when this happens. I
still have nice smooth rabbit trails up the screen, and a great SNR
(20-ish dB).
Has anybody else seen this with back-to-back testing of FreeDV and 700D?
The codec I'm testing against is the dev branch r3689, but it also
happens with the stock 0.8 library.
------------------------------------------------------------------------------
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
