[RTTY] 200 hz filter on RTTY

Kok Chen chen at mac.com
Mon May 31 21:42:17 PDT 2010


On May 31, 2010, at 8:04 PM, Don Hill AA5AU wrote:

> Not much you can do about that although I have heard
> rumors of a modem that could actually copy Baudot hearing only one of the two tones, but this could have been a HAL myth.

No, definitely not a myth, Mark-Only (MO) and Space-Only (SO) works quite well.  I have used it during times when Mark-Space copy is not possible because QRM is overlapping one, but not both of the tones.

MO and SO is selectable right in the "Data Control" part of the front panel of the ST-8000 (I believe also on the ST-6000'e front panel).  It should also be available on software modems since it is so trivial to implement.

Basically, you filter everything (QRM and all) away except for a single tone and its keying sideband.  At that point, the FSK (frequency shift keyed) signal looks like an OOK (on-off keyed) signal.  What this means is that you lose about 3 dB in terms of SNR when there is no selective fading, but otherwise a perfectly decodable signal as long as your modem knows how to handle it.

MO and SO is useless when there is selective fading, however.  

> I think I remember someone, maybe Chen, saying a 200 Hz filter would not work on RTTY or something to that effect.  Maybe I don't
> have the story just right, but it appears to be working great.  Too tight for contesting, but works when required.

If you have a true brickwall 200 Hz filter, you will not be able to reliably copy 45 baud RTTY.  

In practice, an amateur "200 Hz filter" is never 200 Hz wide, of course.   

The RTTY shift is 170 Hz.  With a 200 Hz filter, that leaves 15 Hz on each side of the two FSK carriers.  RTTY baud rate is 45.45, that means that the keying fundamental is 22.7 Hz.  If you really only have 15 Hz on each side of the RTTY shift, a Baudot character that has alternating 1s and 0s will be decoded wrongly with finite probability even under good SNR condition.

Nowadays, it is easy enough to check the rig's real actual width by just measuring band noise (even better, injecting something like the Elecraft NGEN noise generator into the receive antenna connector of the rig) with a sound card spectrum analyzer, especially if the software lets you average the spectrum.  You won't need expensive RF spectrum analyzers and tracking filters like in the bad old days anymore.

The narrowest brickwall filter that would copy 45 baud RTTY reliable is something like 170+(45.45/2)*2 = 215 Hz.  (The 170 is the shift, the "45.45/2" is the fundamental of the keying signal, and the "*2" represents the two outer edges of the FSK signal.)

A 250 Hz filter will therefore copy RTTY signal quite reliably as long as the SNR is good.   And you even have 35 Hz of wiggle room :-).

However, when SNR becomes poorer, you want the passband to accept at least the 3rd harmonic of the keying sideband to capture most of the energy of the modulation (this is the "Eb" part of "Eb/No").  To pass the third harmonic of the keying sideband, the filter needs to be 170 + (45.45/2*3)*2 Hz wide (i.e., about 305 Hz).   I suspect that many "250 Hz" I.F. filters are in reality wider than 300 Hz :-).

Theoretically, when there is neither multipath nor Rayleigh fading, to get best weak signal copy, you actually want a filter passband that is very soft (drops off as 1/f from each of the two tones, where f is scaled to the fundamental of the keying signal).  This is the so called "AWGN matched filter" -- many software modems, including RITTY implements a matched filter for weak signal work.  

When the band is stable (no selective fading or flutter), I prefer using a receiver filter that is at least 500 Hz wide and let the matched filter in the decoder to pull out the really weak ones.

73
Chen, W7AY



More information about the RTTY mailing list