First tests of New progress in WSJT
Copy from: https://lists.berlios.de/pipermail/wsjt-devel/2007-September.txt 

From joe at Princeton.EDU  Fri Sep 21 21:57:35 2007
From: joe at Princeton.EDU (Joe Taylor)
Date: Fri, 21 Sep 2007 15:57:35 -0400
Subject: [Wsjt-devel] Experimental DBPSK and 4-FSK modes in WSJT
In-Reply-To: <46D1FD2C.80205@iol.pt>
References: <46D1FD2C.80205@iol.pt>
Message-ID: <46F4222F.6080507@princeton.edu>

Dear Colleagues,

No doubt you have noticed the recent flurry of "commits" to the WSJT repository.  I should bring you up-to-date on what's happening.

I have been playing with two experimental new modes for WSJT called JT2 and JT4.  JT2 is a possible replacement for JT65A/B, for use at HF through 144 MHz.  JT4 is a possible replacement for modes JT65B/C, intended for EME at 432 MHz through 10 GHz and also for rainscatter at 10 GHz.  I emphasize that both are presently experimental; if they don't work out well, they will not become permanent parts of WSJT.

The basic specs are as follows:

JT2 uses 72-bit messages with the same message structure and source encoding as in JT65.  Instead of 64-tone FSK for the data and one additional tone for sync (as in JT65), JT2 uses 2-tone FSK for the sync and binary PSK for the data. A long-constraint convolutional code (K=32, r=1/2) augments the 72 user information bits to a total of 207 channel symbols.  The keying rate and FSK tone separation are 4.375 Hz, so a transmission takes 207/4.375 = 47.3 seconds, nearly the same as for JT65.

On reception, sync is established by using noncoherent detection of the FSK information.  The synchronizing frequency offsets are then removed, and differential PSK techniques are used to convert the encoded data to soft channel symbols. Finally, a sequential decoder using the Fano algorithm decodes the user message.

JT4 uses 4-tone FSK modulation to convey both a sync bit and a data bit in each channel symbol, following the suggestion made by Stewart Nelson nearly two years ago.  The sync pattern and the encoding/decoding of user information are the same as those used in JT2, except that noncoherent FSK detection methods are used for data as well as sync. Distinct submodes (JT4A, JT4B, ...) will use tone spacings from a minimum of 4.375 Hz up to a maximum of perhaps 315 Hz.

Some possible advantages of JT2 and JT4 relative to JT65 are as follows:

JT2 has a very narrow total bandwidth, just 8.75 Hz.   The WSJT simulator shows it to have a S/N performance about equal to JT65B, or perhaps slightly better.  Its narrow bandwidth could be a big advantage -- especially for random operation -- because 100 or more JT2 signals could easily fit into a transceiver's SSB bandwidth without mutual interference.  Some straightforward software enhancements could then give WSJT many of the remarkable capabilities that MAP65 now provides for JT65, including the decoding of all of JT2 signals in the receiver passband (perhaps 2.5 to 5 kHz).  JT2 could be a very attractive and useful mode at HF through 144 MHz.  If activity were concentrated in a few 5 kHz band segments, random operation could become easy for all 144 MHz EME operators.  The main uncertainty in my mind is whether 2-meter EME signals will have adequate phase stability to make PSK modulation effective on that path. This needs to be tested.

At 432 MHz and higher frequencies, differentially demodulated PSK will most likely fail ... and then JT4 may be an attractive alternative.  The JT4 tone separation can be as low as 4.375 Hz, making the total bandwidth 17.5 Hz: about 1/10 the bandwidth of JT65A, or 1/20 of JT65B.  Larger tone separations will be needed at microwave frequencies; I believe something like 150 Hz should be adequate for 10 GHz EME.   The S/N performance of JT4 may be 2 or 3 dB worse than JT65B/C under ideal conditions on the lower UHF bands, but on real EME paths -- especially at 1296 MHz and above -- JT4 may actually be better than JT65.  At 10 GHz no present WSJT mode can be used for EME, because libration-induced and differential Doppler shifts broaden the reflected signals by much more that the JT65C tone spacing.  With its wider spacings JT4 should be effective there and it can provide significant gain over CW.  JT4 should also work well for rain scatter.

Anyway, those are the basic facts and my present ideas ...

WSJT r537 has implemented JT2 for both Tx and Rx and JT4 for Tx only.  I will probably try some on-the-air tests before proceeding much further.

As always, comments and suggestions will be very welcome!

-- 73, Joe, K1JT
 


From lz1bb at mail.bg  Sat Sep 22 02:00:38 2007
From: lz1bb at mail.bg (Harry Popov)
Date: Sat, 22 Sep 2007 03:00:38 +0300
Subject: [Wsjt-devel] Experimental DBPSK and 4-FSK modes in WSJT
In-Reply-To: <46F4222F.6080507@princeton.edu>
References: <46D1FD2C.80205@iol.pt>  <46F4222F.6080507@princeton.edu>
Message-ID: <1190419239.4670.16.camel@ham>

Joe,

Good news. I already compilled but no partners to test JT2/4. Just a question. As far as I know PSK requires Linear PA. My experience is from PSK31 mode. What about JT2? Should it still work with non linear amplifiers in class "C" like JT65?

And I see too many JT4s as modes. My understanding is that one should be careful. Yes it is good for testing but remember anarchy with FSK441 wich finished when variations were removed.

73, Harry LZ1BB

 

From joe at Princeton.EDU  Sat Sep 22 02:49:44 2007
From: joe at Princeton.EDU (Joe Taylor)
Date: Fri, 21 Sep 2007 20:49:44 -0400
Subject: [Wsjt-devel] Experimental DBPSK and 4-FSK modes in WSJT
In-Reply-To: <1190419239.4670.16.camel@ham>
References: <46D1FD2C.80205@iol.pt> <46F4222F.6080507@princeton.edu>
<1190419239.4670.16.camel@ham>
Message-ID: <46F466A8.5070701@princeton.edu>
 

Harry --

Please treat these new modes as experimental only.  You should not assume that anything in r237 right now will stay in the program.  In no sense has JT2 or JT4 been "released" for general use.  The large number of JT4 submodes, in particular, are there for testing purposes only.

Yes, using a Class C amplifier PSK modes can cause excessive spectral broadening.

It's good that you can compile the program.  No surprises there.  These modes are not yet ready for general on-the-air use.

-- Joe, K1JT

 


From lz1bb at mail.bg  Sat Sep 22 02:43:44 2007
From: lz1bb at mail.bg (Harry Popov)
Date: Sat, 22 Sep 2007 03:43:44 +0300
Subject: [Wsjt-devel] Experimental DBPSK and 4-FSK modes in WSJT
In-Reply-To: <46F4222F.6080507@princeton.edu>
References: <46D1FD2C.80205@iol.pt>  <46F4222F.6080507@princeton.edu>
Message-ID: <1190421825.4670.20.camel@ham>

There must be a bug. While monitoring on JT2. Each time while decoding program is going down with this message:

WSJT Version 5.9.7 r537 , by K1JT
Revision date: 2007-09-21 16:17:03 +0300 (Fri, 21 Sep 2007)
Run date:   Sat Sep 22 00:36:12 2007 UTC
Using PortAudio.

Audio     Input    Output     Device Name
Device  Channels  Channels
------------------------------------------------------------------
   0       16        16       /dev/dsp
   1        2         2       Sound Fusion CS46xx: CS46xx (hw:0,0)
   2        0         2       Sound Fusion CS46xx: CS46xx - Rear
(hw:0,1)
   3        0         2       Sound Fusion CS46xx: CS46xx - IEC958
(hw:0,2)
   4        2         2       front
   5        0         2       rear
   6        0         2       iec958
   7        0         2       spdif
   8        2         2       skype
   9        2         2       radio
  10        2         2       radioconv

Default   Input:  0   Output:  0
Requested Input:  1   Output:  1
Opening device 1 for input, 1 for output.
Audio streams running normally.
******************************************************************
Fortran runtime error: Array reference out of bounds for array 'ccfred',
lower bound of dimension 1 exceeded (in file 'sync24.f', at line 142)
[
73, Harry LZ1BB



From joe at Princeton.EDU  Sat Sep 22 03:01:25 2007
From: joe at Princeton.EDU (Joe Taylor)
Date: Fri, 21 Sep 2007 21:01:25 -0400
Subject: [Wsjt-devel] Experimental DBPSK and 4-FSK modes in WSJT
In-Reply-To: <1190421825.4670.20.camel@ham>
References: <46D1FD2C.80205@iol.pt> <46F4222F.6080507@princeton.edu>
<1190421825.4670.20.camel@ham>
Message-ID: <46F46965.6020506@princeton.edu>

Harry --

As I said, the program is not ready for actual use.  I know about the array bounds problem.  My posting earlier today was only to let the development group know something about what I have been doing.

-- Joe

 _______________________________________________
 Wsjt-devel mailing list
 Wsjt-devel at lists.berlios.de
 https://lists.berlios.de/mailman/listinfo/wsjt-devel