[RTTY] What's cleaner: FSK or AFSK?

[Kok Chen]

On Feb 5, 2014, at 6:28 PM, Joe Subich, W4TV wrote:

> With the K3 it doesn't matter - the K3 DSP passes both AFSK and FSK
> through a narrow filter *after* the modulation stage.  

Unfortunately, the K3 has high transmit (PA) IMD, which keeps it from working at full potential.

As Joe mentioned, you can use waveshaped AFSK, or the FSK-D on the K3 -- both cases produce a nice narrow RTTY signal until the noise passband suddenly widens by 350 Hz or so, starting about 24 to 30 dB below the Mark or Space carrier levels.  

When you filter or waveshape an RTTY signal, there is energy overlap between Mark and Space carriers (unless you narrow the Mark and Space symbols themselves so the filtered tails don't overlap -- doing so however, will reduce the decoding SNR a little, so most RTTY developers are not willing to do it.  And you can't do that using a bipolar FSK keying signal, of course -- you need separate Mark and Space OOK keying signals).

The spectrum widening is caused by IMD from the PA and driver stages.

If you have a panadpater, you can easy see this effect -- what you will see is what looks like a nice, narrow RTTY signal that sits on top of a wider pedestal which is 25 to 40 dB lower than the carriers, depending on the amount of transmit IMD (the "pedestals" are mesa looking shapes that you see in Arizona :-).  It is harder to notice in a waterfall.  The signal also has to be more than 5 or 6 S-units above the noise floor for you to clearly see it.  Some of the spectrum captures that Andy K0SM published show them pretty clearly.

The lower the transmitter IMD, the lower down this "pedestal" starts to appear. 

This sudden IMD caused widening happens with all transmitters.  It is just that the K3's transmitter produces dirtier IMD than some other 12 volt PA (it has 8 dB worse IMD than the Ten-Tec 418 PA, for example, if you believe QST reviews).

73
Chen, W7AY