[Elecraft] Re: Displaying the keying waveshape on Spectrogram.

[Dale Boresz]

Mike S wrote:

>Without the keying speed being specified, the results aren't very meaningful. The amount of energy in the sidebands will vary with keying speed.
>  
>
While it's true that increased keying speed requires increased bandwidth 
for intelligence retrieval, I don't believe variations in the keying 
speed alter the spectral content at all. In any case, while running this 
test, I tried a series of 'dits' at about 10 wpm, and again at about 50 
wpm, both averaged over about a 30 second time interval, and the 
resulting plots were identical.

As another example, a common method of measuring the frequency response 
of a loudspeaker is to drive the speaker with a single fast rise/fall 
time spike, and capture the resulting ping over a time interval of some 
tens of milliseconds. An FFT is applied to this time-domain sample, and 
a resulting spectral plot in the frequency domain is produced. I believe 
this is analogous to what Spectrogram is doing, except that averaging is 
required because it can't sample fast enough to capture all of the 
spectral information contained in a single 'dit'.

Anyway, that's my empirically determined theory, but I'll be the first 
to admit that it is very possibly flawed, and I'd appreciate any 
clarification from someone more knowledgeable of signal sampling theory.

In any case, for myself, it'll be a nice baseline with which to compare 
the results of the 'fix' when it becomes available.

73,
-Dale / wa8sra