On Wed, January 22, 2025 11:56 am, steve bennett via GreenKeys wrote:
> GKs,
>   Too cold to play outside today so here's how I started the day.Created
> this AFSK visualization using the waterfall display in the Goldwave audio
> editing app.It was been slowed way down (obviously) so you can watch the
> frequency shifts.
> https://imgur.com/a/lXlR5Je
> -Steve______________________________________________________________


That's Great! I notice that even when there is steady tone, there is a
signal visible to the left of the main signal. Any idea what that is?

Then, the screen gets pretty exciting when there is keying. In my DSP TU,
I use DDS to generate the FSK and just shift from one frequency to the
other without running the keying signal through a low pass filter (slowing
the transition). If we consider the FSK to be an FM signal modulated by a
square wave, we get sidebands due to the BR/2 harmonics present in the
square wave. If we consider the FSK to be two AM on-off keyed signals, we
also get sidebands on both sides of the tone frequencies due to the
fundamental keying rate (BR/2) and due to the harmonics of that. These are
FSK "key clicks."

The question becomes, should we have a filter between the keying and the
FM modulator? If so, what should its bandwidth be? Since there is also a
filter on the receive side, these filters cascade and attenuate sidebands,
including, possibly desired sidebands.

In my latest Heavy Metal RTTY presentation (
https://w6iwi.org/rtty/Heavy%20Metal%20RTTY241030.pptx ), I look at
demodulator tone filter bandwidth starting at slide 36. As shown there,
using a tone filter bandwidth equal to the baud rate results in the
sideband at +/- BR/2 on each side of the tone being attenuated 3 dB
resulting in about 70% modulation of the tone. Increasing the bandwidth to
1.7 * BR makes the attenuation at BR/2 close to 0 dB, so the sidebands are
not attenuated, and we get 100% modulation.

https://w6iwi.org/rtty/TuReferences.html#:~:text=Certain%20Topics%20in%20Telegraph%20Transmission%20Theory%2C%20H.%20Nyquist%2C
suggests that the tone filters should have a bandwidth of 1.2 times the
baud rate which would still attenuate the desired sidebands some.

So... should we have a low pass filter between the keying (loop detect)
circuit and the FM modulator? If so, what should its bandwidth be?

Harold
https://w6iwi.org




--
Not sent from an iPhone.