On Tue, 30 Nov 2004 18:59:25 +0000 (UTC), "Reg Edwards"
wrote:

How much, if any, is digital filtering better than the old-fashioned Xtl
filter.

Hi Reggie,

DSP, depending on the artistry of the programmer, can do very well
with removing periodic (coherent) noises like birdies and heterodynes.
It can also reduce some incoherent noise. Some filters are adaptable
in that they can recognized (mostly coherent) noises and automatically
reduce them.

As for filtering, DSP filters are capable of skirt selectivity that is
impossible to achieve with any other means. Here too, they are user
configurable (if the option is available) to tailor the skirt slopes
and bandwidth independently and maintain a very low pass band ripple.
All of this is performed through discrete Fourier math, with the
routines in software called IIR or FIR filters. IIR is Infinite and
FIR is Finite Impulse Response.

The software routine, tied to an ADC chip, takes samples and feeding a
delay line, taps that line to feed the same signal (in proportions)
back to the input. Some of this might be compared to the Analog
BiQuad filters but this delay line I speak of actually exists entirely
in software as an algorithm (although early examples in a hybrid
design did this in hardware with greater sophistication than the
BiQuads).

The downside is the quality of the audio has that hollow pipe sound or
exhibits obvious sampling clock signal fed into the audio stream. As
with all digital design, it must be preceded by analog filters to
prevent spectrum folding (high frequency components being translated
into low frequency products) and post filters to smooth out the
product (if the software can't cope).

The sampling rate is the limiting factor. Early DSPs sampling at
least twice the highest audio frequency expected (the Nyquist limit)
established the clock rate and the processor had to perform all the
instructions between those samples. Later DSPs pushed the sampling
rate higher such that you could design filters in the IF range. This
was the second generation DSP of about 10 years ago. Of course, that
IF range was in the 10s of KHz, perhaps as high as 100KHz.
Traditional Analog IF design in this range was already perfectly
suited to razor thin filters with steep skirts, so the advantage was
in the adaptability.

Cheap, external, audio DSPs easily make up for any receiver's lack and
the IF DSP is arguably worth the cost for the beginner.

73's
Richard Clark, KB7QHC