[GreenKeys] FSK shifts

[Harold Hallikainen]

On Tue, August 24, 2021 7:33 pm, Jim Haynes wrote:
>
> You'll want to read the article by K6STI in the April 1996 issue of
> QST, pages 78-79
>
>  	---

Just read that article. Interesting. ARRL members can search for it at
http://www.arrl.org/arrl-periodicals-archive-search .

In general, he suggests that the change from 850 Hz shift to 170 Hz shift
was so the tones would fade together. Fixed threshold detectors would then
work better than if the tones faded separately. With the introduction of
"automatic threshold control," however, that advantage disappeared.

A simple automatic threshold control can be achieved by just AC copuling
(high pass filtering) the combined demodulated data (where a mark tone
drives the signal positive, and a space tone drives it negative). If the
tones are imbalanced, such as during a selective fade, the resulting DC
shift is filtered out. This is done in the TU-170 (see V24 ast
https://w6iwi.org/rtty/tu170/Flesher%20TU-170%20Terminal%20Unit.pdf ).

However, with 170 Hz shift, the tones tend to fade together (which was an
advantage with fixed threshold detectors) while with 850 Hz shift, they
fade separately. If, for example, mark gives a 1V output, space gives a 0V
output, the data going to the slicer would be 1 Vpp centereed about 0 V
(shifting between -500 mV and +500 mV). The slicer at 0V would properly
decode this. Ideally, noise is added equally to both channels, so it
cancels out.

It would be interesting to calculate out (I'd probably use LTSPICE) how
FSK behaves with selective fading. Assuming selective fading is a single
reflection with almost equal amplitude to the original signal, we could
simulate the sum of three equal signals. One is the reference, the second
is down 170 Hz, and the third is down 850 Hz. Te "direct" signal would be
combined with a delayed sum that results in a full cancellation of the
reference signal. How far down are the other two signals?

http://mai.hallikainen.org/org/FCC/FccRules/2020/97/307/section.pdf
authorizes up to 1 kHz shift at 300 baud.

Are the bands so crowded that this could not be done today?

We used to use 2.125 kHz and 2.925 kHz tones for AFSK with 850 Hz shift.
But most SSB transmitters would severely attenuate a 2.925 kHz tone. When
we changed to 170 Hz shift, we kept the mark at 2.125 kHz and put the
space up 170 Hz. Europe, however, put space at 2.126 kHz and mark down 170
Hz. I note that Navy LF LSB uses a 1 kHz center and shifts +/- 85 Hz (see
https://w6iwi.org/rtty/ ). I wonder if it would make sense to use a center
frequency of 1 kHz and shift up and down 85 Hz for 170 Hz shift and 425 Hz
for 850 Hz shift.

Harold




-- 
FCC Rules Updated Daily at http://www.hallikainen.com
Not sent from an iPhone.