[RTTY] 300hz or 500hz IF filter?

[Kok Chen]

On Aug 22, 2013, at 7:34 PM, Joe Subich, W4TV wrote:

> Absolutely incorrect as 250 Hz does not account for the necessary
> modulation sidebands or for the discontinuity (additional bandwidth)
> generated by the 1.5 bit stop.  Due of the half bit, the necessary
> bandwidth for 170 Hz shift RTTY approaches 170 + (2 * 90.9 * 1.2) or
> slightly over 370 Hz as the shortest element is now 11 ms.

Joe,

Kai is correct.  

Even in May 1964,  in his "Filters For RTTY" article in the RTTY Bulletin Vic Poor (SK) had mentioned that data filters has nothing to do with keying sidebands.

I had mentioned this in my RTTY Demodulator article, quoting Poor:

> As emphatically stated by Mr. Poor, the proper filter has nothing to do with being "wide enough to let the third (or fifth or umpteenth) harmonic pass." For any design bandwidth, the optimal filter has to meet his two conditions. 

(from http://www.w7ay.net/site/Technical/RTTY%20Demodulators/Contents/filters.html).

The two conditions that Poor mentioned were of course the first two Nyquist Criteria (up until today, Nyquist's third criteria is seldom used).

As long as there is no ISI, you can make the filters so narrow as to pass the fundamentals of a single keying sideband, as discovered by Harry Nyquist in 1928.

For OOK keying, the Raised Cosine is the narrowest filter that has no intersymbol interference (ISI).  For a steam RTTY signal, it falls to zero (zilch, nada) at 45 Hz away from the carrier.  So, an optimal RTTY filter has zero response outside of 170 Hz + 91 Hz = 261 Hz.  

The equivalent noise bandwidth of the Raised Cosines are narrow than that, giving the narrowest filter's equivalent noise bandwidth that is about 216 Hz (the number that Kai cited).

For optimal SNR, the Matched Filter is quite a bit wider and about 0.1 dB better than the Raised Cosine, which is the narrowest filter that has no ISI.

73
Chen, W7AY