Hi Charlie,
Many thanks for sharing your test results -- they are much appreciated.
Following up on them, I found an issue in the JT65 DS routine that can
*sometimes* hinder finding a valid decode. It has been fixed in
revision 6568.
To test in a way consistent with your tests, I ran two instances of
WSJT-X r6568 with an audio loopback cable. Doppler spread = 40 Hz and
S/N = -24 dB, as in your tests. With DS enabled but Avg not enabled,
here's what I got on 18 recorded test files (6 each with 3 different
messages). All the decodes are correct.
1813 -23 0.1 1001 # K1ABC W9XYZ EN37 d14 9
1814 -23 0.0 1000 # K1ABC W9XYZ EN37 d14 *
1815 -21 -0.0 999 # K1ABC W9XYZ EN37 d13 *
1816 -22 0.1 1000 # K1ABC W9XYZ EN37 d14 7
1817 -23 0.1 998 # K1ABC W9XYZ EN37 d14 *
1818 -22 0.1 1000 # K1ABC W9XYZ EN37 d16 7
1819 -24 0.0 1001 # K1ABC W9XYZ -24 d12 *
1820 -21 0.1 1000 # K1ABC W9XYZ -24 d12 *
1821 -21 0.1 1000 # K1ABC W9XYZ -24 d15 *
1822 -22 0.1 1001 # K1ABC W9XYZ -24 d15 *
1823 -22 0.0 999 # K1ABC W9XYZ -24 d14 *
1824 -22 0.1 1000 # K1ABC W9XYZ -24 d14 *
1825 -22 0.1 1001 # K1ABC W9XYZ RRR d14 *
1826 -24 0.0 999 # K1ABC W9XYZ RRR d12 *
1827 -22 0.0 1001 # K1ABC W9XYZ RRR d14 *
1828 -27 0.1 999 # K1ABC W9XYZ RRR d14 *
1829 -24 0.0 1001 # K1ABC W9XYZ RRR d14 *
1830 -21 0.1 1001 # K1ABC W9XYZ RRR d14 *
With Avg enabled and DS not enabled, I opened the files sequentially,
starting at the one for UTC 1813. I clicked "Clear Avg" after each
group of 6. Again the decodes are all correct; they needed either 2 or
3 message repeats in order to get a decode. These lines appeared in the
right-hand ("Average Decodes") window:
1817 -23 0.1 998 # K1ABC W9XYZ EN37 f32
1818 -22 0.1 1000 # K1ABC W9XYZ EN37 f33
1821 -21 0.1 1000 # K1ABC W9XYZ -24 f23
1822 -22 0.1 1001 # K1ABC W9XYZ -24 f23
1823 -22 0.0 999 # K1ABC W9XYZ -24 f33
1824 -22 0.1 1000 # K1ABC W9XYZ -24 f32
1827 -22 0.0 1001 # K1ABC W9XYZ RRR f22
1828 -27 0.1 999 # K1ABC W9XYZ RRR f23
1829 -24 0.0 1001 # K1ABC W9XYZ RRR f32
1830 -21 0.1 1001 # K1ABC W9XYZ RRR f33
Finally, I enabled both Avg and DS and repeated the above procedure for
each group of 6 files. Here are the results from the right-hand window:
1815 -21 -0.0 999 # K1ABC W9XYZ EN37 d24 *
1816 -22 0.1 1000 # K1ABC W9XYZ EN37 d24 *
1817 -23 0.1 998 # K1ABC W9XYZ EN37 d33 *
1818 -22 0.1 1000 # K1ABC W9XYZ EN37 f33
1821 -21 0.1 1000 # K1ABC W9XYZ -24 f23
1822 -22 0.1 1001 # K1ABC W9XYZ -24 f24
1823 -22 0.0 999 # K1ABC W9XYZ -24 f33
1824 -22 0.1 1000 # K1ABC W9XYZ -24 f32
1827 -22 0.0 1001 # K1ABC W9XYZ RRR f22
1828 -27 0.1 999 # K1ABC W9XYZ RRR f22
1829 -24 0.0 1001 # K1ABC W9XYZ RRR f32
1830 -21 0.1 1001 # K1ABC W9XYZ RRR f33
All of the above results make sense. There should be no difference in
the way that messages with calls+loc, calls+rpt, calls+RRR, or calls+73
are treated -- except, of course, that only calls+loc applies to
callsigns other than the one entered in "DX call". The FT decoder
always takes precedence. If it succeeds, the DS decoder is not called.
-- Joe, K1JT
------------------------------------------------------------------------------
Transform Data into Opportunity.
Accelerate data analysis in your applications with
Intel Data Analytics Acceleration Library.
Click to learn more.
http://pubads.g.doubleclick.net/gampad/clk?id=278785471&iu=/4140
_______________________________________________
wsjt-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/wsjt-devel
