New versions K1JT weak signal digital modes

Bob, G3WKW, has passed on this information from Joe Taylor K1JT:

"Date: Fri, 07 Aug 2015 16:28:19 -0400

Several people have asked for an update on development of the "Fast modes" in WSJT and WSJT-X.  So here's a brief summary.

First, a review of some relevant terms and motivations.  It's convenient to think of the various WSJT protocols ("modes") in two groups:

*Slow modes* -- JT4, JT9, JT65, and WSPR.  These modes are designed for communication with extremely weak signals -- often too weak to be heard. Target propagation modes include EME and long-distance troposcatter on HF-and-up bands, and QRP Dxing on the LF, MF, and HF  bands.  Relevant signal amplitudes are approximately constant over a minute and more, aside from so-called "libration fading" for EME. Transmit/receive sequences are 1 minute for JT4, JT9, and JT65, and 2 minutes for WSPR.

*Fast modes* -- JTMS, FSK441, ISCAT, and JT6M -- and now also *FSK315* (implemented in WSJT) and *JT9E* through *JT9H* (implemented in WSJT-X. These modes are made for communication with rapidly varying signals:for example, meteor scatter, ionospheric scatter, airplane scatter, and scatter off the International Space Station.  The decoders are designed take advantage of short enhancements of signal strength.  T/R sequences are 30 seconds (or sometimes even shorter).

Bill, ND0B, has implemented a trial version of FSK315 in WSJT.  Think of this mode as FSK441 slowed down to 315 baud; the bandwidth is therefore narrow enough to make the mode legal in the "CW and data" portion of the 10 meter band.  Bill and a few others have been experimenting with FSK315 and also ISCAT-A on 10 meters, under dead-band conditions, using meteors and ionospheric scatter propagation.

I have implemented experimental submodes of the JT9 protocol in the program branch WSJT-X v1.6.1.  As with JT4 and JT65, letters following the "JT9" designator indicate increased spacings between the FSK tones. Traditional JT9 (now also called JT9A) has tone spacing 1.736 Hz, so the signals used at HF and below have total bandwidth 9*1.736 = 15.6 Hz.  The widest of the new submodes, JT9H, has tone spacing 200 Hz and therefore bandwidth 9*200 = 1800 Hz.

When used with the standard 1-minute periods, the wide JT9 submodes should be useful for the same purposes as the wide JT4 submodes: microwave EME, for example, where libration fading can cause Doppler spreading of 100 Hz or more.  Used in this way, all JT9 submodes are "slow" modes; they use 1-minute T/R periods and keying rate 1.736 baud, and they send the full 85-symbol message protocol in 85/1.736 = 48.96s.

Optionally, the wide JT9 submodes can now also use "fast" keying rates equal to their tone spacing.  "Fast JT9H", for example, uses keying rate 200 baud, so the full message protocol is transmitted in 85/200 = 0.425s.  The message is sent repeatedly for the full Tx period, in the same way as done for the other fast modes.

The fast JT9 submodes should be very effective for meteors and ionoscatter propagation, especially on the 6 meter band.  Sensitivity should be similar to ISCAT, or perhaps slightly better.  Because JT9 includes strong forward error correction, decoding results are like those for all the slow modes: you should see messages exactly as they were transmitted, or nothing at all.

Tests of the fast JT9 submodes are currently under way, with excellent results.

   -- 73, Joe, K1JT"

Mixed results with WSPR-X and WSPR-15

The new version of WSPR works well with plenty of reports given and received on 472kHz with it over the last 12 hours using the 2 minute version (as before but now with a waterfall and better interface), but mixed results on the slower 15 minute slot WSPR-15 which can detect much weaker signals.

My problem is TX drift over a long 15 minute transmission. My small 10W transverter is in a tiny plastic box. When using a 2 minute transmission the PA barely gets warm so there is little heat transferred to the crystal oscillator. With the much longer TX burst the PA warms more and the crystal warms up several degrees, moving the oscillator frequency by some 10s of Hz. So, although I was able to copy signals in WSPR-15, I was not able to be detected by others overnight. There are also far fewer stations monitoring yet on WSPR-15.

The solution is to rebuild the transverter into a bigger (metal) box with (a) better heatsinking so the heat rise with a mismatch is lower, (b) a slightly more efficient PA - problem here is with weather changes the antenna match goes off optimum sometimes and the heat rises as it the PA becomes less class E, and (c) put the conversion frequency oscillator physically further away from the PA, currently it is less than 10cm.

If I fix these issues and am sure that I transmit a more stable signal in the narrower (25Hz wide) WSPR-15 slot then I have a better chance of succeeding with the more sensitive version.  With another 7-9dB S/N detection improvement, my signal could reach some of the more distant stations like TF3HZ.

WSPR-X and WSPR-15

A note from Joe Taylor K1JT informing us about a new version of WSPR.

WSPR-X and WSPR-15

Wed Jan 16, 2013 8:23 pm

Some of you may be interested in trying WSPR-X, a new version of WSPR that includes the slow mode WSPR-15. WSPR-15 uses 15-minute T/R sequences, rather than the standard 2-minute sequences; at MF and LF it is 9 dB more sensitive than WSPR-2, decoding signals as weak as -37 dB in the standard 2500 Hz reference bandwidth. WSPR-15 is not recommended for use at HF: the tone spacing is only 0.183 Hz, less than the Doppler spreading typical of many HF paths. The main intended application for WSPR-15 is for very difficult paths at 137 kHz and the new 472 kHz band. It may be interesting to try on 160 meters, as well. Note that the recommended WSPR-15 sub-band is a 25 Hz slice just above the 200-Hz WSPR-2 sub-band, i.e., 1600-1625 Hz above the standard "dial Frequency". This will be handled automatically if you set up WSPR-X in the normal way.

A brief online User's Guide for WSPR-X is posted at
http://www.physics.princeton.edu/pulsar/K1JT/WSPR-X_Users_Guide.pdf ,

and the Windows installation file is posted at
http://physics.princeton.edu/pulsar/K1JT/WSPRX_07r2948.exe .

At present, WSPR-X does not support I/Q audio (for use with direct conversion receivers and transceivers) and does not do “band hopping”. Otherwise most familiar WSPR features are present and working well.

For software enthusiasts: Unlike older versions of WSPR, the user interface of WSPR-X is written in C++ and the Qt programming framework. My expectation is that this shift will make for easier development and program maintenance in the future. At present a click-to-install package is available only for Windows. Source code for WSPR-X is available from the open-source SVN repository at berlios.de. Anonymous checkout of the full WSPR-X source code can be accomplished with the command svn co svn://svn.berlios.de/wsjt/branches/wsprx

User comments, suggestions, and bug reports will be very welcome!

-- 73, Joe, K1JT

First signals seen on JT9-2 mode

On 500kHz this morning PA0A and G3ZJO were testing with the new JT9-2 weak signal mode. Not yet copied Eddie, but PA0A was coming through well. This is a screenshot with his decoded signal overlaid.

500kHz first JT9-2 signal received

JT9: a new digital mode for MF and LF

Hot news from Joe Taylor K1JT on the RSGB LF Yahoo group this evening about a new 9-FSK digital communications mode (for 2-way QSOs rather than beaconing) optimised for MF and LF bands. With winter approaching and better LF conditions, this is an exciting development.

The wide graph display for JT9

"I invite you to try a new digital mode called JT9, designed especially for making amateur QSOs at MF and LF. JT9 uses the structured messages introduced in 2003 for the JT65 mode, now widely used for EME and for QRP operations at HF. JT9 can operate at signal levels as low as -27 dB (in a 2500 Hz reference bandwidth), with one-minute timed transmissions. It also offers slower transmissions of 2, 5, 10 and 30 minutes duration, and the slowest mode can decode signals as weak as -40 dB. With one-minute transmissions, submode JT9-1 has a total bandwidth of just 15.6 Hz -- less than one-tenth the bandwidth of a JT65A signal. The other submodes are narrower still: a JT9-30 signal occupies about 0.4 Hz total bandwidth.

Note that these JT9 sensitivity levels are comparable to or better than those of WSPR, which uses simpler messages and is not intended for making 2-way QSOs.  JT9 has much higher throughput and reliability than QRSS CW, including DFCW modes.

JT9 is implemented in an experimental version of WSJT called WSJT-X. Some further details can be found at http://www.physics.princeton.edu/pulsar/K1JT/WSJT-X_Quick_Start_Guide.pdf , and an early version of WSJT-X can be downloaded from
http://physics.princeton.edu/pulsar/K1JT/WSJTX_01r2695.exe.

Please note: WSJT-X is in an early development stage. A number of improvements and enhancements are already in the works, and others will surely be added. 

Your feedback will be much appreciated!

-- 73, Joe, K1JT"

SM6BHZ hears me on 500kHz WSPR

SM6BHZ heard me at last on 500kHz WSPR at 0556z this morning, so my best DX has now gone up to 951kms with the tiny station. He reported my WSPR signal at -26dB S/N. The distance beats my previous record on the band by almost 100kms. Gus, thank you so much for country number 6 on the band and Joe K1JT for designing the software to make this possible.

A map to end a fantastic day on 500kHz QRPp

This is the map that says it all: DX that even yesterday I would not have believed possible with a true QRPp microstation on 500kHz WSPR. I go to bed very happy, leaving the beacon running overnight in case any others manage to detect my signals.

K1JT, Joe Taylor, thank you for bringing us this wonderful software package.

WSPR versus CW - what's the improvement?

As WSPR is so effective I asked Joe, K1JT, what the difference is between WSPR and "ear-and-brain" CW. This is his very helpful reply:

"For an answer to your question about relative sensitivities of CW, WSPR, and some of the other modes implemented in WSJT, let me suggest going to the "References" link on the WSJT web site, http://physics.princeton.edu/pulsar/K1JT/refs.html ,
and select reference #11. Table 3 on page 9 gives the information you asked about. The difference given there is about 11 dB, in favor of WSPR over ear-and-brain CW. For most operators, the difference is more like 15 dB."