|Subject:||LF: PSK31, PSK08, Spectrum Lab|
|Date:||Wed, 6 Feb 2002 12:41:25 EST|
Thanks to all for the reports. Concerning PSK31 and PSK08:
In fact there is an implementation of PSK08 in Spectrum Lab, but it's not very user friendly and not really optimized (in fact, it is a crude version of PSK31 with two additional decimation stages before the decoder). I have compared its performance in the presence of white Gaussian noise and got a much lower error rate, but white noise is not a real world signal.
So the practical "gain" of PSK08 against PSK31 in an LF-environment with static crashes and 'carrier type' noise and will be less.
In contrast to John's (G4CNN) observations, I did not notice missing or wrong characters in Jims transmission, the character error rate in his signal here in JO42 was less than one percent. This is certainly a result of the different types of noise at different locations, as digital transmissions with a "relatively high" symbol rate suffer much more from short crashes than others, and there is no error correction method in Peter's varicode set which was intended for fast keyboard-to-keyboard communications where it does an excellent job.
(John, I hope you didn't use Spectrum Lab as a PSK31 decoder. It will not display a warning if the computer is not fast enough for a high-res waterfall and the PSK decoder running at the same time, so I highly recommend using one of the "profi"-programs like Digipan, WinPSK, etc etc. One of them may even have PSK08 implemented).
If you want to do some experiments with the implementation of PSK08 in Spectrum Lab, you can download it from www.qsl.net/dl4yhf/spectra1.html . But beware ! The current version has the charm of a building-site (*), the proper function of the PSK08 decoder has not been verified since a long time and its almost certain that there are undiscovered bugs in it.
(*) more about that soon, if other threads don't keep me busy.
|<Prev in Thread]||Current Thread||[Next in Thread>|