[Elecraft] Dual Pass Band

[Jim Brown]

On 1/6/2014 10:34 AM, Joe Subich, W4TV wrote:
> That information is based on input from K7AY (cocoaModem) and G3YYD
> (2-Tone).  The software decoders do a better job when supplied with
> audio at least 400 Hz wide without frequency shaping from the IF (DSP)
> filter.  The Dual Band Pass filters in general introduce significant
> phase delays and amplitude inequality which results in poorer decoding
> accuracy in modern FFT based demodulators - although the frequency
> shaping may be of value in older amplitude (filter/detector) style
> demodulators. 

Actually, it's based on things a lot more fundamental than that. In the 
world of pro audio, we learned nearly four decades ago that non-flat 
amplitude response is accompanied by phase shift (distortion). The ear 
does not like phase distortion, and neither do decoders. Filters produce 
distortion, not only when their amplitude response is changing (the 
slopes), but also in the passband.

When i first encountered the dual-filter approach separate from the 
decoder itself, it looked to me like a bad idea for this reason. It was 
good to see it confirmed by G3YYD in the doc for his 2Tone RTTY decoder. 
BTW -- Nobel Laureate Joe Taylor, K1JT, author of the WSJT digital 
protocols, said exactly the same thing during a lecture I heard him do 
at Pacificon several years ago.

73, Jim K9YC