At 06:56 AM 5/13/03, Tom Rauch wrote:
>How else would we set bandwidth in a communications system?
That would depend on which bandwidth we're attempting to set.
In most cases we'd be interested in setting the < -20 dB BW or the occupied
bandwidth at > 97% of total power. Design control over these bandwidths
would be through the shape of the keying waveform.
But if there was a requirement that necessitated setting the -6 dB BW or
the occupied bandwidth at < 97% of total power the only degree of control
available is through the keying rate. Not much can be accomplished with
envelope shaping so close to the carrier without also suppressing all of
the keying sidebands in the process.
>We could never detect the peaks and valleys in the bandwidth.
True enough. In order for a signal to remain recognizable as CW, the
receiver bandwidth must be wide enough to pass at least a few keying
sidebands. Requiring a filter wide enough to do that obviously renders
individual sidebands and nulls undetectable.
The peaks and nulls in the spectrum of the transmitted signal could be
resolved by using a narrower filter. But, after having passed through such
a narrow filter, a CW signal would no longer be readily discernible as such.
There's a corollary to this. If the receive bandwidth is narrowed to the
point that it becomes impossible to copy CW, an improperly shaped CW signal
no longer sounds "clicky" off frequency. (While it no longer sounds clicky,
what it does sound like is semi-continuos crud.) Just like being able to
copy CW, hearing distinct clicks requires the presence of several keying
sidebands within the receiver passband.
73,
Mike K1MK
|