de M0BMU

I received G4JNT's Jason beacon again last night, and messed around further with the gain levels, etc. I found the best results were achieved with the AGC set to manual, the RF gain cranked up to maximum, and the audio gain set so that the sound card ADC had a fairly high input level applied, but not enough to overload it. Under these conditions, the IF signal was quite severely clipped all the time (see the attachment). The audio sounded very distorted of course. Backing off the RF gain so that little or no clipping occurred made the signal unreadable by Jason, and almost invisible on a waterfall display - whilst with the "Jimi Hendrix style" gain settings, the waterfall was 'O', and Jason could copy perfectly - the difference was very striking and counter-intuitive.

I'm now fairly sure it is not the sound card ADC that is involved in the effect, but rather that clipping the signal under QRN conditions improves the SNR after further processing. This is the opposite to the optimum setting required for receiving QRSS when there are multiple signals in the passband; in this case the gain must be reduced until no clipping occurs, otherwise "blocking" affects the weaker signals.

I will look out for SM6LKM's 7MHz beacon on the new frequency later today when I get a chance.

Let me see if I can say something meaningful.

If you clip in the IF, the noise energy of the clipping process is spread
over a larger set of frequencies, thus the energy/Hz is lower than if you
clip at audio.  Since Jason is a non-linear mode, it will benefit from a
'limiting amplification' or crank up the gain and filter out the out of
band distortion products--this works better at RF than at baseband as the
sums and differences are out of the passband.  This is basically what
you're doing by cranking up the RF gain.  Cranking up the gain at baseband
will leave the distortion products in band--where they will interfere with
the desired signal.

You could implement something like this at baseband, but you'd want to
very narrowly filter the input *first*, then amp/clip, then filter
again.  Well, the last filter may be implemented by the Jason receiver
program, of course.  The 'anything outside of the yellow lines is ignored'
thingey.

But, I could be wrong.

Cheers,
David N0YMV