[Moon] EA9LZ 6cm

[charlie at sucklingfamily.free-online.co.uk]

Hi John

There are a number of factors that could have been at play here.

VK7MO and I did an extensive comparison of QRA64 and JT4F some time ago,
and published the results in Dubus 3/17.  These involved the use of both
simulation and lots of on-air testing (on 3cm).  The conclusions showed a
clear advantage for QRA64, for libration spreading values of up to about
200Hz. Yesterday on 6cm we were well below this spreading.

One of the things we learnt from the study was the need to look at large
numbers of test files (thousands) to reduce effects of statistical
variations - small sample sizes can easily give misleading results.

The performance of QRA64 is also dependent to some extent on the settings
used.  There is a link in the User Guide at the end of Section 8.4 that
describes some of these factors.  The linked document also describes how
to set up the waterfall to be more like WSJT10's, which makes seeing the
tones somewhat easier than with the default waterfall settings in WSJT-X,
but this of course does not affect decoding performance.

A common reason for failure to decode is when you are tuned so that the
lowest (leftmost) tone of the QRA64 signal falls outside the range of Ftol
you have selected. Choosing a low value for Ftol, while this can sometimes
help decode very weak signals, does increase the requirement for tuning
precision.  You commented on the logger that you could hear the tones
audibly.  In this case QRA64 in our experience has no trouble whatsoever
in decoding.

When comparing the performance of QRA64 and JT4 when no prior knowledge of
callsigns is available, QRA64 has an even larger advantage.  Hence it
really comes into its own for random operation.  With JT4 running with
correlation decoding (Deep Search) and a small sample size the difference
may not be so apparent.

By 'above the line' on WSJT10, are you referring to the difference between
a single line decode and a decode obtained by averaging?  QRA64 does not
have an averaging capability, at present.

When the JT4 decoder needs to use correlation to obtain a decode, the
decoded text includes a quality factor.  Smaller numbers mean that the
decoder had less confidence that the decode was a correct one, and if you
see a ? in the decode then it is normal practice to wait for another
decode to see if it is the same, after which you can be more confident
that the decode was a correct one.  With QRA64 it is our 100% experience
that you can rely on a decode being correct.

73

Charlie



> Hi Zdenek, well done! Great to finally get to you on 6cm last night for a
> second band.
> Only sorry I screwed up the day for 13cm as we would have made it 3 bands.
>
> Your decision to switch to JT4F  was inspirational!
>
> I was running both WSJT10 (RX) and WSJT-X (RX and TX) in parallel during
> the JT4F QSO and at no time did I either of the softwares fail to decode
> you, whereas with QRA64D I got just 2 decodes in the hour or so of trying!
> That said you did seem stronger when you were running JT4F. Maybe our
> minor
> tracking errors just got "in synch" for the JT4F QSO.
>
> Having slept on it, I wondered if it might have been the deep search in
> JT4G, but all the decodes were "above the line" on the WSJT10 screen. I
> don't know enough about Joe's excellent software to know if that really
> means it was a pure decode without deep search? Maybe one of our digimode
> experts may be able to tell me or have a view on their experiences
> comparing the two modes at really low levels? I was expecting QRA64 to be
> much better.
>
> 73 John
>
>
> On 10 October 2017 at 08:34, Zdenek SAMEK-OK1DFC via Moon <
> moon at moonbounce.info> wrote:
>
>> Hi EME gang, "game is over" , last evening I have loged 20 QSOs. Looks
>> that
>> everybody who was interesting have EA9, has EA9. Now will wrap up all
>> the
>> stuff and tomorrow morning I have ferry back to Spain. Then only drive
>> 3000
>> km and looking forward to see my sweet home.  Thanks to all of you for
>> QSOs,
>> many very nice emails and looking forward to meet you again via Moon.
>> 73!
>>
>> --
>> Zdenek - OK1DFC
>> www.ok1dfc.com
>> QRV EME 144 MHz - 10 GHz
>> WAC 432 - 1296 - 2320 MHz
>> WAZ #9 - WAS 432 #29
>> DXCC 432 - 1296 MHz
>> _______________________________________________
>> Moon mailing list
>> Moon at moonbounce.info
>> /mailman/listinfo/moon
>>
>> Join eQSL.cc  https://eqsl.cc/qslcard/Index.cfm
>>
> _______________________________________________
> Moon mailing list
> Moon at moonbounce.info
> /mailman/listinfo/moon
>
> Join eQSL.cc  https://eqsl.cc/qslcard/Index.cfm
>