[Elecraft] K2 clix

[Lyle Johnson]

> This reference is not discussing an on-off signal like CW, but
> rather a signal where two bits are represented by a 180 degree
> phase shift in a more or less continuous signal.  I believe this
> is actually quite different and that's why the optimum filtering
> arrived at in this discussion is something other than in the
> Gaussian curve as derived in the reference I cited.

Actually, Miller's article is discussing the baseband waveshaping which
cares not if the modulation is PSK (the phase shift you mention), FSK (the
G3RUH 9600 bps modem whish is in the article) or AM (which is CW - on/off
keying).

And so is Schmidt's.

If you compare the spectra for the similar waveshaping in both references,
you'll find they are essentially the same.

The "key clicks reference" is talking about the waveform of the CW "pulse"
not the RF carrier or how the RF carrier is modulated.  See figure 2, for
example, in http://fermi.la.asu.edu/w9cf/articles/click/index.html and
compare it with Miller's
(http://www.amsat.org/amsat/articles/g3ruh/108.html) figure 3 - they are the
same.

This is because both articles are talking about an exponentially filtered
baseband pulse (an RC filter).  And neither analyzes RF!

Similarly, Miller's figure 7 and Schmidt's Figures 6 and 7 are strikingly
similar.

We extrapolate both results to RF assuming the waveshape is not distorted.
The context of CW is on-off keyed RF, so we make the leap in our minds that
the spectra in Schmidt's article represents occupied RF bandwidths - which
they do it the modulation system is linear (preserves the baseband
waveshaping).  But nowhere does he analyze RF, just the keying waveshape.

Similarly, while Miller is using examples from PSK and FSK systems, he is
examining the baseband and explaining how to minimize occupied bandwidth.

As I said earlier, the relevance of Miller's work applies, even though the
context is different.  I suspect if you used the 9600 bps "packet" baseband
shaping he mentions on CW, you'd get minimum occupied bandwidth, which
includes clicks.  Much better than R-C filters, or even the cosine shaping
(which he calls "error function keying") advocated by Schmidt.

Enjoy!

Lyle KK7P