[RTTY] PSK31 is faster (Was FD RTTY Question)

[Kok Chen]

On Jun 27, 2012, at 11:33 AM, Bill Turner wrote:

> Isn't the "real" objection to PSK that is prone to garbling due to 
> multipath? It might be ok for a domestic contest like Field Day, but how 
> about a DX contest?

That is correct.

When there is no multipath, PSK31 beats out RTTY by a couple of dB, assuming both modes use the same average power.

However, as multipath increases, PSK31 errors build up rapidly.  Using more power won't help either since it is the phase of a signal at a previous bit period vs the phase of the signal at the current bit that cause errors.  PSK31 uses phase differences to encode data bits (dibits in the case of QPSK31),

This is why increasing power with PSK31 is completely counter productive after some point.

When conditions get worse with RTTY, you can mostly increase power and get through.  

PSK31 tends to be more "binary."  When conditions are good, you can use very little power to get through to DX.  But when multipath is bad enough, no matter how much power you use, you asymptote at a constant amount of error rate.  This can be shown easily with HF Channel simulators by turning the AWGN noise off, but keeping the Doppler shift and Doppler spread terms on -- i.e., emulating a gigawatt station from across an auroral path HI.

Since PSK31 is so narrow, it also gets wiped out by selective fading.  With RTTY, when one tone has selectively faded, a software demodulator that has an ATC circuit will be able to copy with just one tone, albeit with error rates that are equivalent to a signal that is 3 dB wear or so.  Not all RTTY software demodulators implement ATC or DTC, though.

Wider shift RTTY is even better to counter selective fading, but even 170 Hz shift is good for most HF cases.  Again, only if your demodulator implements ATC/DTC.

Other than the inefficiency of two tone FSK, RTTY's weakest point is arguably the Baudot encoding that requires LTRS and FIGS shifting.  Especially with contest exchanges where you shift between FIGS and LTRS very often.  And even more FIGS needed when you engage USOS.  

Not only do the shifts add extraneous (non-printing) characters to slow you down, they also cause a single bit errors to turn into burst errors (affecting multiple characters).  

73
Chen, W7AY