Topband: Receivers

[Richard Karlquist]

> It would be interesting to know what the actual shape of this filter is.
> As Tom said, it is physically impossible for a "brick wall" filter of this
> bandwidth to avoid ringing.  If the shape factor is large, however, there
> could be enough sideband energy allowed through the skirts, to give some
> kind of inter symbol transition.
> 73, Greg ZL3IX

For minimum phase filters, it is true that a brick
wall filter inevitably has extremely delay distortion,
resulting in ringing.  However, for non-minimum phase
filters, there is no such delay constraint.  Analog filters
are almost always minimum phase, although they can
be non-minimum phase (I built some of those during
the analog era).  DSP filters are frequently FIR types,
which are inherently non-minimum phase.  This gives
the designer many more degrees of freedom.

This is not to say that non-minimum phase is a panacea.
There will still be some ringing associated with them,
some of it in the form of pre-echos.  Also, there is
probably no single "best" filter, rather it depends
on the transmitter keying filter and the CW speed as
well, and perhaps on the S/N ratio.

If you are interesting in the topic of ringing, it is
useful to read about "Gibb's phenomenon", usually
covered in texts on Fourier series (eg Bracewell).   Gibb
discovered that a square wave passed through a perfect
brick wall filter (ie no phase or amplitude distortion)
has a certain amount of irreducable ringing.  Making
the filter wider only makes it ring faster, but it doesn't
reduce the amplitude of the ringing.

Rick N6RK