- Transmitted on GB3SCC 5.76GHz
- and on 10GHz MoonBounce

Generate DFCWi from the Soundcard Dual Frequency CW, or DFCW, is a technique that evolved from SLOWCW, or QRSS. This is where a CW signal is transmitted at a very slow symbol rate, typically at speeds from 1 to 80 seconds per dot. This allows reception in very narrow bandwidths - down to tens of mHz - using a visual spectrogram, or waterfall display. The first example of SLOWCW being used on the amateur bands was my QSO with G3PLX on 73kHz in July 1997. Since then the technique has grown and is in widespread use on the 137 and 500kHz bands, with a with amateurs experimenting on HF with ultra-low power

Sending conventionally coded CW using this technique is wasteful of time, in that dashes take three times as long to send as dots, but carry equal information content. Therefore, as a spectrogram display can discriminate in frequency, it was postulated that if a dash were sent at a slightly different frequency to a dot, but of equal length with a bit of mental agility it would be possible to read the 'CW' off the screen with little more effort than for conventional CW coding.

So DFCW was born, and several software authors wrote code for generating the mode, with several schemes for actually generating the frequency shift. These often used the serial port handshaking lines (RTS and DTR) for controlling a variable frequency oscillator, or a DDS source that could be programmed directly. Other writers produced Spectrogram / waterfall type software specifically optimised for monitoring QRSS. 'ARGO' - by I2PHD is a favourite amongst LF operators.

As there is plenty of visual differentiation between dots and dashes, and the length of symbols is constant, the need for a full dot gap between symbols is a bit wasteful, and most QRSS is now sent with a much-reduced gap between symbols, ranging from nothing at all (multiple sequential dots or dashes are inferred from the length of the trace on screen) up to 25% - 50% of a dot length.

Recently, after a fraught time trying to resolve weak, broken and frequency scattered CW on 10GHz, John, G0API, wondered if some data-type mode relying on a PC and could be adopted for such microwave signals. The nature of the scattering is such that dashes frequently get broken up into multiple dots, and the spectral width of the carrier is widened to a few tens. or even hundreds, of Hz in heavy rain. To differentiate dots from dashes on a spectrogram-type display, DFCW is a natural choice if the frequency shift is made wide enough to cope with scattering.

Then a brainwave struck! Why use just the two tones with breaks of signal between symbols? Why not replace the 'key-up' state with a third tone so it becomes obvious when the signal has really dropped out. Now, the duty cycle of the transmission is at 100% and a tone will always be heard if the path is open, and the DFCW can be read off a screen (or copied by the practiced ear!) with the added advantage of clearly having a key up refeernce line on the screen as well.
As this is DFCW with an additional Idle Tone, let's call it DFCWi

A comment on the Microwave Reflector resulted in favourable comments on this idea, and a suggestion from GW4DGU to use tones related musically (he is a musician) and in the hundreds of Hz to 1kHz region where the ear is optimised for listening. Suggested tones were:
A reference frequency which carries the space information , 'dash' frequency 587.33Hz above that and 'dot' frequency at 783.99Hz above the reference.
As a test, I modifed my PIC beacon keyer to generate the DFCW data and used a TP5089 DTMF encoder chip in single-tone mode to generate the three tone frequencies. They aren't exactly Chris's suggested ones, but are near to them, ie: Idle 694.8 Hz Dot 1206.0 Hz Dash 1331.7 Hz

Symbol length was set at 500ms as a compromise between a speed fast enough to be readable by ear, and slow enough to be decoded with an FFT bin size of a few Hz, to give an advantage of a few dB Signal/Noise over that possible by listening alone. The Intersymbol gap was set at 10%, or 50ms to minimise time wasted, but still give an indication of symbol boundaries and supply a clearly defined tone for listening.

Listen to the pure, noise free transmitted signal --- TxSignal

With added simulated, or synthetic, Gaussian noise at the Full Nyquist Bandwidth of5.5kHz, with S/N ratios of -10 and -20dB:
-10dB S/N & -20dB S/N -----------Illustrated here ---------->

These S/N ratios equate roughly to -7dB and -17dB in the reference 2500Hz (SSB bandwidth), and are equivalent to to +7dB and -3dB in 100Hz which is roughly the noise BW of a practiced CW operator's Ear.

Listen to them, view on your waterfall display of choice (Spectrogram, Spectrum Lab , ARGO or Spectran (etc) and let us know what you think...

DFCW-I Generation Software

DFCWi.EXE is a simple piece of software for generating DFCW messages from the default soundcard. Tone frequencies, symbol length, intersymbol gap and amplitude can be changed as required. The software is written in VB6, so you will need to have the runtime library for this already installed on your computer. Based on experience with other software, I have found that many users already have the necessary libraries.

Download DFCWi.exe
The latest Version has 0/90 degree Stereo data for Direct Conversion I/Q upconverter drive and an option for continuous looped messages.
If the software generates an error message saying the libraries cannot be found, download and install UWBCNMON , after which DFCWI should run properly.
The programme generates four small .WAV files corresponding to a complete symbol at each of the three tone frequencies and a fourth for the inter-symbol gap. The DFCWi is formed by repeatedly playing these in the appropriate order.

GB3SCX 10GHz Beacon now sends DFCWi

On 19 September 2007 a DFCW Ident was added to the transmission from GB3SCX on 10GHz beacon
Parameters are :
~ Symbol interval - 0.5 second
~ Idle frequency 10368.905MHz,
~ Dot Tone +400Hz, ~ Dash Tone +600Hz
On 14 Aug 2009 this was replaced by a JT4G signal, but DFCWi continues on the GB3SCC 5.76GHz beacon

Screen capture of the signal as received at my QTH 70km away over a double-obstructed path. It was decoded using Spectrum Lab

This screen capture was decoded using Spectran which is a lot easier to get going than Spectrum Lab

In this trace, note how the first symbol of the 'B' (second letter) has been obliterated by a rapid fade. The absence of a corresponding idle tone at this point indicates this is a fade, rather than the letter 'S' having been sent.

Both traces show the DFCW data reading "GB3SCX IO80UU"

Here is a simultaneous audio recording and screen plot captured using Spectran. and one of 10GHz Rainscatter


On 27 September, DFCW was succsessfully transmitted via Moonbounce on 10GHz. The signal was transmitted from the 1.8 metre EME dish at the Flight Refuelling Amateur Radio SOciety G4RFR. The signals were received (2.5 seconds later!) by Brian, G4NNS, in Andover. Transmit power was about 13 Watts, and the receive Signal/Noise ratio was just barely adequate for normal CW. The DFCWi signal was sent by holding the transmitter microphone against the speaker on a PC running the DFCWi.exe Software
Brian recorded this EME DFCW transmission. The voice that can be heard about half-way through belongs to G0API on the 2m talkback link - and not via the Moon!

Back to Index