With PSK31 on Short Wave

 

First things first; my first confirmed DX QSO in PSK31 mode was from om Nestor from Buenos Aires. The QSO was made 22 June 2004 on the 15 meter band (QRG was 21.082 MHz). Not that special one should say. But if you take into account the circumstances ( 20 Watt HF power and a simple dipole aerial -inverted V- under the roof) one might be slightly amazed . My signal arrived with a solid S599 in Argentinia. For experienced HAM radio amateurs, having worked with PSK31 for a longer period, nothing to be amazed of. For me, however, it was a small miracle that - although assisted by that great mirror above us (free electrons in the ionosphere) - the tiny PSK signals could cross the Atlantic Ocean without problems.

Simulations show that PSK31 is a real QRP mode. The program NTIA/ITS for example shows the incredible opportunities of PSK31; below is shown what your chances were for making PSK31 QSOs with only 5 Watts (!) in the 40 m band on February 2001 around 16 uur UTC. The PSK31 transceiver " under test " was located in Germany. As you can see, the chance for making a QRP-link within Europe was 90% (red area). Parts of Australia, New Zealand and the Pacific could be reached with some extra luck (yellow areas; 75%) ! This kind of pictures made me very curious. The proper time to collect some background information about this interesting mode and get things prepared to go QRV with PSK31 !

 

PSK31, an introduction

PSK31 is a - from RTTY derived - modus being developed in 1999 by Peter Martinez, G3PLX. PSK means Phase Shift Keying. Where other modes switch on- and off the carrier wave (CW), vary the carrier wave frequency (FM), or shift the carrier wave (FSK), PSK31 does not spoil valuable HF power because of its smart principle. BPSK (Binary Phase Shift Keying) is being generated by changing the polarity of the carrier wave. This happens with a ' keying rate ' of 31.25 baud. Therefore, theoratically the bandwidth of the BPSK31 signal will be 31.25 Hz (!). In order to facilitate 'life QSO's' this choosen bitrate turns out to be the practical minimum value. This bandwidth is much smaller than the bandwidths for any other mode of telegraphy (digimode). The fast click created by the polarity change results in a relative wide frequency spectrum. Therefore, the signal must be filtered to the proper bandwidth. After this filtering process, a double side band (DSB) signal with suppressed carrier ("two tone signal") comes available. The RTTY signal is also a two tone signal with a greater frequency separation compared to the PSK31 signal.

The efficient PSK31 signal  shown as graph

In addition to the 31.25 Hz bandwidth, the use of the Varicode is a second characteristic of this modus. The code was developed by Peter Martinez. In fact, it is based on the old morsecode but some enhancements were added. The signlength of the characters is variable (as in the morsecode). The average signlength is 6.5 bits per sign. The Varicode alfabeth was derived by Peter from a huge pile of English texts. The often used character ' e ' has the code 11 (short !). The less often used [ sign has the code 111110111. Useful to know for practice; as a result of the construction of the Varicode alfabeth, the transmission rate for the small characters (non-capitals) is twice the rate of the capitals. In order to transmit Varicode at a speed of about 50 words per minute, a bitrate of appr. 32 bits per seconde is required. The value of 31.25 was choosen because this is an easy value to derive from the 8 KHz sampling signal of DSP systems. Practice has shown that 31.25 Hz is an acceptable bandwidth to facilitate ' life QSO' s ' ("chatting").

PSK31 signals can often be received on following frequencies (not accepted in the official band plans for 2004):

160 m

1838 KHz

80 m

3580 KHz

40 m

7035 KHz

20 m

14070 KHz

17 m

18100 KHz

15 m

21080 KHz

12 m

24150 KHz

10 m

28120 KHz

 

A related mode to BPSK31 is QPSK31 (Quaternary Phase Shift Keying). Compared to the BPSK31 signal, QPSK31 has a doubled information capacity without the disadvantage of a double bandwidth and loss of speed. A second carrier wave (shifted in phase) is added. The price that must be paid is the fact that the available HF power must be divided amongst both channels. Practice has also shown that QPSK frequency stability must be twice as good compared to PSK31 (QPSK within 4 Hz). For the old transceivers no option ! Two stations often start to work with BPSK and switch to QPSK, if the conditions allow it.

The designer has refrained from adding an error correcting algoritm into the PSK31 modus (as in e.g. the PACTOR modus). This algoritm should slow the transmissions thus spoiling the possibility of life two-way contacts. 'Breaking-in' , as commonly done in ' live QSO's ' , should become impossible as well.

 

PSK31; operating practice

For regular PSK use, a transceiver with linear amplification characteristics must be used. In addition, a PC with a minimum of 16 MB RAM memory and a processorspeed of 100 MHz is required. The soundcard must a 16 bits type. No extreme requirements these days I should say. I am using a Siemens Celeron 266 MHz with 64 MB RAM memory. The PSK31 software works smoothly on this platform. The choice for PSK31 software is great at this moment. For Windows, DOS or Linux users plenty of software can be found. Just look for it with Google and type in "PSK31 AND software". I have choosen for the program MixW ( www.mixw.net ). Be aware, this is no freeware or shareware type of software. But it is a splendid piece of software making the use of the digimodes very easy. (Besides PSK31 also modes like RTTY, CW, PSK63, QPSK31, Throb, Pactor, Hell and Fax are supported). The waterfall display plays a keyrole rol in selecting the proper signal(s). The impression of the pograms working area below shows that there are about 8 PSK31 signals within a 2 KHz wide band. When listening to the audio of this signal one might determine only one or two seperate PSK signals. This shows the strength of the decoding part of the software. Sometimes only noise can be heard, but at the same time, the waterfall display shows a number of clear PSK31 signals. Just one click on such a track (lost in the noise) and readible text will be generated (in most cases). Pure magic when glancing at this phenomenon the first time.

MixW also provides a built-in Logbook. Very handy. The number of succesful QSO's is counted automatically. It also tells you whether you have workend with a station before. In addition, the macros allow the possibility to program standard texts. In the example here, under the "CQ CQ"-button the user has programmed the text shown in the RX screen (upper screen) in red and the same text is visible in black in the TX screen. Very useful that items like UTC, callsign counterstation and Log # can be integrated into the macros. The "BYE"- button for example can generate a piece of text with the current date, UTC and Log # . Very nice. Also for your counter station. Do not forget to align your cursor behind the last character in the TX screen, otherwise your transmission will stop somewhere in the middle of a sentence (the actual position of the cursor). I have experienced this some times.

I have choosen to use the PC's parallel port for the interface between SW transceiver and PC for routing the control signals. This port was not used on my PC yet. The circuit on the left side shows the way I made the connections. The components have been mounted into the 25-contacts D-connector. I have used BC547B instead of the shown 2N4401 transistors. I did not include the 'paddle' part. Do not forget to select the parallel port in the setup of the program. Linking via the serial port is possible as well. See info at ' hardware ' on the MixW website: www.mixw.net

The signal path from microphone input up till and including the power amplifier must amplify linearily. When offering big signals to the microphone input, the input circuitry will produce distorted signals. The distortion will be amplified in a linear way and the resulting PSK signal will be 10 times (!) wider than usual. All the advantages of the PSK31 mode will dissappear just like that. In the waterfall display those wide signals are frequently noticed. The solution to attentuate the input signal in the proper way is very easy; use a 2 resistor network (1 : 100 voltage divider) as shown below.

 

 

The audio signals have been connected as shown on the right side. Use shielded wires ! I have used a 4K7 potmeter instead of the two resistors. This enables, in combination with the Windows volume control, the selection of any desired HF TX output. I found that 20-30 Watt HF power for this digimode works quite comfortable. The signal coming from the LINE OUT of the soundcard via the potmeter is connected directly to the (sensitive) MIC input of the SW transceiver.

At the moment of setting up my configuration it was not clear whether my old tubetransceiver from 1972, a TS515 would support the severe frequency stability recommendation of 8 Hz for PSK31. The handbook told that the short term stability is 10 Hz (after being switched on for at least 30 minutes). After having worked with this digimode a couple of months (with about 100 PSK31 QSOs end of July 2004, within Europe and Eurasia) I can conclude that the setup works just fine. The Japanese designers of yesterday really did a great job !

Inverted V dipole for 40, 20, 15 and 10 m band under the roof (with a balun in the middle of the dipole).

 

The built-in AFC (Automatic Frequency Control) facility of MixW enables my counter stations to lock onto my signal automatically. This covers part of the stability insufficiencies of my rig. But sometimes things really do not work. At times the 10 Hz stability (shortly after switching on the transceiver) requirement is not met and my counterstation has to click on my unsteady trail again.

I can forget QPSK31 with its 4 Hz recommendation !

 

Sources:

August 2004