Re: [MT63] MT63-like modes for EME?
- Tomi Manninen wrote:
> I'm not really an expert but I think the idea of using MFSK for EME workThat makes sense. I haven't been able to get solid values on what the
> is correct. I have no real EME experience but from what I have gathered
> the path has huge amounts of fluctuation, it's even worse than a bad HF
> channel where phase coherence is often lost even during a typical symbol
> time leading to PSK type modulations failing completely... Like PSK31 does
> under heavy multipath propagation.
coherence time is for an EME path (depends heavily on frequency, I believe),
so I can't make any guesses, yet. With luck I'll have an operational
station in March and I can try it for myself. :)
> Parallel modems, using OFDM, really only make thing worse in this respect.Yep, that makes sense. They'll produce a lot of symbol to symbol
> In OFDM systems each individual carrier uses longer symbol time and
> thus makes it more prone to phase incoherence.
(adjacent channel) interference, then, wouldn't them? Hmm..
> MFSK modems are well suited to non-coherent demodulation and reallySo you don't think they symbol time needs to be kept shorter than the
> benefit from long symbol times.
channel coherence time?
> > JT44 looks okay, but the redundancy is very simple--simpleWhere would this software be available? :)
> > repetition. So, it's not cross correlated like a convolutional
> > code would be. The base FSK system seems sound, but the coding
> > on that seems a bit less optimal.
> I was recently somewhat involved (mostly giving some advice in DSP
> programming etc.) in realising LinWSJT, an implementation of the WSJT
> modes (so far only JT44 is done) for Linux, by Jonathan G4KLX/HB9DRD. I
> would definitely say that the mode would benefit from an efficient coding
> method. I would propose some R=1/3 block code, preferably one that allows
> soft decoding.
- On Sun, 1 Dec 2002, David Willmore wrote:
> > MFSK modems are well suited to non-coherent demodulation and reallyThat's what I think but then again I might be completely wrong... :)
> > benefit from long symbol times.
> So you don't think they symbol time needs to be kept shorter than the
> channel coherence time?
> Where would this software be available? :)Jonathan's web page is at
LinWSJT can be found under "Software" ...
Tomi Manninen Internet: oh2bns@...
OH2BNS AX.25: oh2bns@...
KP20ME04 Amprnet: oh2bns@...