I promised earlier that I would do a short presentation of the JT65 and JT9 modes. Well, here it is.
Light technical presentation
JT65 was created by K1JT, Joseph Taylor, a radio astronomer and also a Nobel prize winner in 1993, mainly for EME (Earth Moon Earth) communication or moon bounce at VHF frequencies and up. Although meant for EME, it grew popular among HF operators and today it has many users on the shortwave bands. The fact that it is a weak signal mode makes it also ideal for QRP operation.
The mode itself, on air, sounds like a melody being played. A transmission is made using AFSK (Audio Frequency Shift Keying) modulation and consists of 65 different tones in which one, the lowest, is a synchronization tone. A transmission occupies approximately 177 Hz of bandwidth in the SSB passband but the actual efficient bandwidth used by a transmission is actually much smaller than that. I will return to this later on.
Fig. 1 JT65 signal in a waterfall view
Figure 1 above shows a JT65 signal viewed as a “waterfall” image. On the left the sync tone can be seen as the clear line and to the right of the sync tone the 64 data information tones are visible as white dots.
The mode is time controlled, using even and odd minutes for transmission and receiving thus showing output to the operator once a minute. This makes operation very strict and not as free as with other digital modes on HF. The actual transmission or receiving period takes 48 seconds. During these seconds a pattern of tones is transmitted which correspond to the message generated by the software. After the 48 sec the decoder at the receiving end starts processing the received tones . The processing takes a few seconds, depending of the computer used, and the output is shown to the operator as text messages. The operator then has about 10 seconds to choose an action before a new reception interval begins or a transmission starts. A typical QSO takes about seven minutes. For just monitoring the traffic a new receiving interval begins every beginning of a new minute.
Fig 2. A waterfall image of monitoring JT65. Notice the stations almost exact timing in starting the transmissions.
As you can see the JT mode is dependent of correct timing. The software uses the computer clock for keeping the time synchronized with an accuracy of seconds. A drift as little as two seconds makes your transmitted signal undecodable to other operators and the received signals useless. As the common home computer normally isn’t that accurate in timing the use of an NTP client software is recommended and sometimes even necessary.
In figure 2 we can see an example of a waterfall view of a number of J65 transmissions. The waterfall view is divided in to minute “rows” by red horizontal lines were each red line represents the start of a new minute. All transmissions starts after the red lines with a nearly exact timing.
As it is a weak signal mode the decoder is capable of decoding signals with a very weak signal levels. The decoder can work with signals as low as -22 dB SNR (Signal to Noise Ratio) nearly without any error and down to -24 dB with about 50% decoding error. The capability for weak signal decoding is partly made possible thanks to the use of a forward error correcting technology (FEC) called Reed Solomon. The code is commonly used in CD-ROM and Hard Drives for producing extremely low error rates in data recovery. But it isn’t only the algorithms of the decoder that makes the magic. During the 48 second transmission the message is transmitted over and over again. Up to 80% of the transmission can be lost still producing a correct decoded message. This adds to the strength of the protocol.
Other than this there is the fact that the transmitted signal Is extremely narrow in bandwidth. As I mentioned above, the JT65 mode uses 177Hz in signal bandwidth. This is the bandwidth that the transmission needs “free” for its use. In the actual transmission each tone uses only a small portion of the bandwidth almost making it as narrow as a CW transmission.
This is also a strength of this mode as it makes the use of high power unnecessary. You can work long distances with QRP equipment.
As with JT65, it was created by K1JT as development of JT65 for use on the LF/MF and HF bands. The JT9 mode are in many ways similar to JT65 as the share the same basic concept. They are both using the 48 second transmission period, decoder and time sensitivity but there are some differences.
Fig. 3 A JT9 signal in a waterfall image
The bandwidth of JT9 is much narrower using only 15.6Hz of the SSB passband with a nine tone AFSK transmission. One tone is for synchronization while the eight remaining is for data. This allows more JT9 transmissions to occupy the same bandwidth as a single JT65 transmission. The decoder is capable of decoding signals as low as -22 dB SNR nearly without any error and down to -26 dB with about 50% decoding error. This adds some 2dB compared to JT65.
Other than this the two modes is just the same to the operator and just as JT65, JT9 is perfect for QRP use.
If you want to go deeper into the protocols of JT65 the creator have made a document describing the mode more in detail. The PDF document can be found here.