JT2 and JT4 info

WSJT version 5.9.8, r558 is an experimental release that contains two new operating modes, JT2 and JT4.

JT2 uses 2-tone FSK modulation for synchronization and differential BPSK for the encoded user information. In this way both a sync bit and a data bit can be transmitted with each channel symbol. Symbols are sent at a rate of 4.375 baud, and tone separation for the 2-FSK modulation is 4.375 Hz.

JT4 uses 4-tone FSK, so it can also also include both a sync bit and a data bit in each symbol. Again the keying rate is 4.375 baud, and a number of different tone spacings are offered.

JT2 and JT4 are both designed for making QSOs under extremely weak-signal conditions. They use the same message structure and source encoding as that used in JT65. Forward error correction is accomplished with a long-constraint convolutional code (constraint length K=32, code rate r=1/2), which makes these modes behave rather like JT65: at the receiving end you either get the exact message that was transmitted, or you get nothing at all.

The performance of JT2 may eventually be as good as JT65, or slightly better, at 144 MHz and below (but see below for a current status report). The JT2 signal is very narrow -- its total occupied bandwidth is just 8.75 Hz -- so dozens of JT2 signals can be scattered at random in a 2500 Hz SSB transceiver bandwidth with no more than a few accidental collisions. If it can be made to work up to its potential, JT2 could be a great boon to random digital EME operation on 2 meters, especially for those without panoramic wideband receivers and software like MAP65. JT2 may also be attractive for use on the VLF, MF, and HF bands under weak-signal conditions.

JT4 is presently more robust than JT2. For purposes of experimentation, a number of bandwidths have been made available
through submodes tentatively called JT4A through JT4G. The JT4A submode has 4.375 Hz tone spacing and 17.5 Hz total bandwidth -- twice the bandwidth of JT2, but still only 5 percent of the JT65B bandwidth. The measured performance of JT4A is some 0.5 to 1 dB worse than JT65. At the other extreme, the JT4G submode uses 315 Hz tone spacing and 1260 Hz total bandwidth. The wider JT4 submodes are designed to be useful for EME on the higher microwave bands, and for rain scatter at 10 GHz.

I invite anyone to test these new modes. Please be aware, though, that nothing about JT2 or JT4 is considered fixed at this time; the details of the signals, and especially details of the the decoders, will probably continue to evolve. It is possible that JT2 and JT4 will not become permanent features of WSJT. I am NOT recommending any wholesale changeover from JT65 to JT2 or JT4 for general use, at the present time.

If you learn something about JT2 or JT4 from on-the-air experience or bench testing, or if you have some relevant knowledge or opinions about these experimental modes, please let it be known -- either to me, or on a relevant reflector. We will all benefit from your input.

Current Status

Tests made so far, including a number of EME QSOs on 144 MHz, can be summarized as follows:

1. JT2 works very well at HF. To be competitive with JT65 for EME it will need a better phase-tracking algorithm than I have so far implemented. In short, JT2 (in WSJT 5.9.8 r558) works quite well with stable signals. It can be frustrating to use because seemingly adequate transmissions too often fail to decode, especially if there are frequency drifts. This will need to be improved if the mode is to become widely used at 50 or 144 MHz.

2. JT4 is quite robust. Tests with the WSJT simulator show that its S/N performance is about 0.5 to 1.0 dB worse than JT65B. A submode should be chosen such that the width of a CW carrier (caused by path modulation and/or oscillator instabilities) is less than the tone spacing.

Some parameters of the experimental modes are as follows:

Mode    Spacing    BW    S/N limit    Possible Applications
                (Hz)        (Hz)       (dB)
---------------------------------------------------------------------
JT2       4.375      8.75        -24          HF; EME at 50, 144 MHz
JT4A    4.375      17.5        -23           EME at VHF/UHF
JT4B     8.75        35.0        -22
JT4C    17.5         70.0        -21
JT4D    39.375    157.5     -20           2.3, 3.4 GHz EME
JT4E    78.75       315.0     -19           5.7 GHz EME
JT4F    157.5       630.0     -18           10 GHz EME
JT4G    315.0      1260.0   -17

The S/N limits are approximate and refer to conditions in which frequency spreading is less than the tone separation. The "Possible Applications" are suggestions only; on-the-air experience is needed to determine what works best.

In operation you will notice that WSJT's "red curve" has 3 side lobes of diminishing amplitude on either side of the main peak. This behavior is normal, and a natural result of the embedded sync mechanism used in JT4.

Neither the JT2 nor the JT4 decoder is optimized at present. They still need AFC and other features such as message averaging, matched filtering, and fine-tuning of some decoding parameters. They work pretty well already, however; both JT2 and JT4 have been used to make QSOs from 1.8 MHz up through 144 MHz EME. More on-the-air tests are now needed, in order to make progress.

If you make on-the-air tests of these experimental modes, please send me your results or report them on this reflector. As always, all comments, suggestions, reports, and wave files will be very welcome!

The URL for experimental program download is:

http://physics.princeton.edu/pulsar/K1JT/WSJT598.EXE