[Lowfer] 495.9 last nite

[JD]

"In WSPR (and other digital modes) i-f filters narrower than the 'standard' 
~2.5 kHz used for
USB reception only serve to artificially inflate the reported s/n ... weak 
signal reception is
not enhanced."

Very true, Jay.  (With a few exceptions, perhaps, where broadband noise 
would be sufficient to shred the signal badly enough through AGC action.)

There is another way of looking at reported S/N, though.  Measuring noise 
over full voice bandwidth, when the information and the decoding technique 
actually require much less than that, is just a way of inflating the 
perceived performance of the modulation method.

It makes folks feel good about copying signals that are "below noise," even 
though that's only comparing apples to oranges...or maybe more aptly, one 
apple sitting in a tiny basket beside a bushel of irrelevant oranges.  Were 
I considering buying the apple to eat, I wouldn't especially care about 
what's in the bushel of oranges, only whether the little basket contained 
any bugs or other foreign matter along with the apple I'm interested in.

If software developers wanted to be more honest about performance of their 
pet modes, I suggest they should define the necessary detection bandwidth 
(I'm willing to let them include not just the BW needed to retrieve the 
information itself under ideal conditions, but also any allowance that is 
necessary for the algorithms to track signals with real-world drift once 
acquired) and then compare the signal against the noise energy detected in 
only those relevant bins.

John