[Elecraft] WSJT-X, AGC, Notch, and NB

[Jim Brown]

On 5/13/2019 2:53 AM, Richard Lamont wrote:
> The WSJT-X decoder has huge dynamic range (16-bit 48000 samples/sec
> input decimated to 12000 samples/sec, so over 90 dB), so it's best to
> use the RX as a dumb converter and let the decoder deal with it. It's
> the right tool for the job.

Yes, BUT -- while the theoretical range of a 16 bit system is 96 dB, 
non-linearities at the lower end of the dynamic range limit it to about 
90 dB. That's still a lot, and it's why I recommend setting the audio 
level as high as possible at the input of the A/D.


> This means keeping the bandwidth of the RX at least as wide as the band
> displayed on the waterfall and avoiding the use of anything in the K3
> that might improve the signal for aural CW/SSB reception such as notch
> filtering, equalisation, noise blanking and noise reduction.

I will offer an alternative view on use of the NB. Yesterday, AG6EE was 
in a rare grid in northern NV on 6M, and there was VERY strong impulse 
noise coming from his direction. Without the NB, WSJT-X could not decode 
him. WITH the NB, it did, and I made the QSO. I would NOT recommend 
general use of the NB, because it generates IMD that results in multiple 
decodes of the same signal, but there are exceptions to every rule. :)

> AGC should be avoided because the level of every FT8 signal in the
> passband will be modulated by fading on the strongest signal, which
> makes the decoder's job harder.

I find that this is not a problem with AGC-S, primarily because in my 
situation, the very strong signals are locals, not subject to QSB.

73, Jim K9YC