--- In digitalradio@yahoogroups.com, Sven <sven9...@...> wrote: > > Hi, > I' ve searched the Internet for RTTY decoding methods and found the following: ... > Are there other known methods and who knows something about or can explain > the above decoders ? How do they perform in a noisy environment ? > I'm far from expert and up to date, but let me reveal how little I know.
The original concept for FSK was to use a separate filter and detector for mark and space and combine the outputs of the two detectors by taking the difference or by running them into oppositely-poled windings of a polar relay. This was worked on by Schmitt of Teletype and by Armstrong about the same time. Presumably they were unaware of each other's work. Both of them hoped that noise would affect both detectors equally and cancel out. Carson of AT&T showed that this hoped-for effect would not work, although he may have missed that the two-frequency scheme has an advantage over make-and-break keying because the transmitter is transmitting all the time; hence the transmitted power is higher with FSK than with make-and-break. Armstrong went on to develop FM as we know it, with a limiter followed by a discriminator. The next incarnation of FSK reception used the same principle. FM systems have a threshold property: with SNR above threshold they improve the SNR in the output signal, and with SNR below the threshold it gets even worse in the output. A professor at MIT proposed using positive feedback aound the input filter and limiter. This has the effect of sharpening the threshold. In the 1960s some hams got interested in returning to the limiterless two-tone schemes. Some of this was probably a revival of Armstrong's hope, unaware of Carson's rebuttal. I only became aware of the Armstrong and Carson material in the last year or two. In two-tone systems there is a definite advantage if the transmitting station is using "diddle" as this gives both detectors some signal to chew on all the time; there are no long pauses when only the mark signal is present and the space detector forgets how strong its signal was. Work continued on terminal unit designs as new ICs came along, such as the phase locked loops. There was also a scheme that I don't know if anybody ever tried, called frequency feedback. This uses a VCO heterodyned with the input signal and controlled by the discriminator output. It does not phase lock, but reduces the apparent shift of the signal so that a narrower filter can be used ahead of the discriminator. L-C filters gave way to active filters and then to switched-capacitor filters, the latter making it easy to vary the center frequency of the filter to accomodate odd shifts and arbitrary audio frequencies. There may have been some work done with DSP using some of the DSP-engine development kits that I am unaware of. Then about 1996 K6STI announced his RITTY program, using the ordinary 486 or Pentium PC with a sound card to do DSP. One of his innovations he called the "digital flywheel". This took advantage of the constant character rate, if the sender was using diddle or was sending from a file, to lock to the character rate and use matched-filter detection. He continued developing and improving this software for the next four years or so. I believe he achieved about the best that can be done for FSK reception. He pulled the product off the market when some people cracked his copy-protection scheme and also when the original sound-card PSK31 software came out and was offered free. Today you find the vast majority of rag-chewers have switched from RTTY to PSK-31 because the latter usually performs better for a given SNR. RTTY continues to be popular for contests and DX because of its more rapid turn-around and because you don't care that much about errors when you already know pretty much what the other station is going to say to you. I'll confess I don't have any idea what algorithms the various RTTY demodulators are using today. Jim W6JVE
