[RTTY] Fw: ARLB006 NTIA: No Objection to Additional Data Modes on60 Meters [now another FSK/AFSK thread]

[Kok Chen]

On Mar 30, 2012, at 12:45 PM, Jay WS7I wrote:

> Joe wide filters work better, trust Chen, but you have to use his 
> software.  

Wait, wait... (oops, that is from another show, but on NPR radio, not TV :-).

I think that you need to qualify the statement with "as long as you don't saturate the sound card by using a wide filter."

Once the sound card clips, it is "game over" for any software demodulator.

However, as long as the sound card does not clip, it is better to trust the software to do the proper filtering, where it can apply optimal Nyquist filters (Match Filters are also Nyquist, incidentally).

The problem with arbitrary filters that don't satisfy the Nyquist criteria (and this includes crystal I.F. filters) is that they generate Inter-symbol Interference (ISI).  A Baudot bit from a station is smearing into its *own* next bit's position, causing errors under low SNR conditions even when there is no QRM.  

By using a wide filter, you avoid adding in the group delays from crystal filters, and achieve the design criteria of the software author.

When there is enough QRM, you will need to sacrifice demodulator sensitivity, however.  So it is not a case of never using narrow filters.

73
Chen, W7AY