[600MRG] Inconclusive Pre-Sunset N1BUG JT9-2 Results

[Rik Strobbe]

Assuming that Paul is running the same power in JT9-1 and JT9-2 that SNR should be the same for both.
And as Paul mentions the JT9-2 treshold should be about 3dB better (lower).
In fact SlowJT9 uses the JT9 decoder from the WSJT-X suite. JT9-2 signals are resampled to JT9-1 and then fed to the JT9 decoder.
In the outcome of this decoder the SNR of the resampled JT9-2 signal are 3dB too high, so I subtract 3dB before displaying the decodes.

For those who want to give it a try: 
Put SlowJT9 in JT9-2 modus and in the Settings->Save menu select Save wav files. Now you can find the resampled WAV file is the save folder.
If you have a decode note the time, call frequency and SNR
Next open a command window, go to the folder where SlowJT9 is installed and continu to the subfolder wsjt. There you will find jt9_copy.exe. 
Now type:
jt9_copy -9 -d 3 c:\slowjt9\save\nameofthewavfile
The path to the WAV file (c:\slowjt9...) can be different on your computer. Dont forget the .wav extention.
If done right you should see the decoded messages. SNR should be 2 or 3dB better than the noted value. Frequency should be 2.22 times higher.

73, Rik  ON7YD - OR7T

________________________________________
Van: 600mrg-bounces at mailman.qth.net <600mrg-bounces at mailman.qth.net> namens N1BUG <paul at n1bug.com>
Verzonden: woensdag 12 december 2018 23:31
Aan: Andy - KU4XR; RSGB LF Group; 600 meter group
Onderwerp: Re: [600MRG] Inconclusive Pre-Sunset N1BUG JT9-2 Results

Thanks for the continuing reports Andy, and thanks to all other
participating stations.

Actually your SNR reports make perfect sense. SNR is reported as the
signal level in a 2500 Hz bandwidth. This should not change between
modes. Therefore with the same SNR measuring algorithm in the
decoding software, SNR should be the same for both modes when both
can decode the signal. What you are seeing appears to be just that,
with perhaps a bit of QSB influence as expected.

The slower mode should be able to decode 2 to 3 dB further into the
noise compared to the fast one. The theoretical limit for JT9-1 is
27 dB. JT9-2 should be able to decode 2 to 3 dB less than that, so
-29 to -30.

I actuality, I believe the SNR calculation algorithm is not quite
the same between the two programs, so I would be cautions trying to
compare SNR directly. Some of the results I have seen tend to bear
out my hunch that the respective relationship doesn't end up being
the same on all receiving systems. This has probably led to much
confusion as people try to compare the modes.

I believe a better measure of one mode against the other is simply
this: is the slower mode able to decode the signal more often or
under more marginal conditions than the faster one. That is why I am
running this test, in an effort to determine whether this is the
case. There is mounting evidence that JT9-2 is able to get the
message through more often than JT9-1.

73,
Paul N1BUG



On 12/12/18 4:53 PM, Andy - KU4XR via 600MRG wrote:
> I started seeing traces of signal in the waterfall at 21:00 UTC - 7 minutes ater sunset at N1BUG, and 82 minutes
> before Sunset at KU4XR.. The first decode of JT9-2 at 21:18 UTC is still an hour Pre-Sunset here.. Decodes of
> both JT9-1, and JT9-2 have occurred consistently since.. Skywave component of course affecting Pre-Sunset
> results.. Distance between stations is too far for daytime / groundwave reception..
> SNr reading is raising an eyebrow here, however I do not pretend to understand.. Both softwares receiving the
> exact same audio feed, from the same radio.. I expected to see Higher Number SNr for JT9-2 compared to JT9-1
>
> 2118 -28  0.8 1350 2  VVV N1BUG 2
>
> 2124 -26  0.8 1350 2  VVV N1BUG 2
> 2126 -25  1.8 1350 @  VVV N1BUG 1
>
> 2130 -26  0.8 1350 2  VVV N1BUG 2
> 2132 -26  1.8 1350 @  VVV N1BUG 1
>
> 73 all:
> Andy - KU4XR
______________________________________________________________
600MRG mailing list
Home: http://mailman.qth.net/mailman/listinfo/600mrg
Help: http://mailman.qth.net/mmfaq.htm
Post: mailto:600MRG at mailman.qth.net
Message delivered to rik.strobbe at kuleuven.be

This list hosted by: https://www.qsl.net
Please help support this email list: https://www.qsl.net/donate.html