A DSP filter can have very linear phase response across the pass band.
My DSP-59+ handles 20 wpm CW down to 50 Hz bandwidth without too much
damage or ringing.
The phase response is a design variable. The more coefficients used in
the filter synthesis, the better the possible phase response at the cost
of the greater the time delay between input and output. It is also
possible to design DSP filters with the same Chebychev approximations as
used for analog filters and with the same rotten time response. Such
filters tend to have less average time delay, but not constant across
the pass band. Phase response across the pass band is not critical for
audio since our ears are not sensitive to the relative phase of
harmonics and components of speech. But data is very sensitive to phase
and a Chebychev filter can destroy data using more than the central
third of the passband. And rings for a long time when hit by a spike.
The ladder filters Tentec has been using for decades don't have the
sharp shoulders of the Chebychev filter and do have a better phase
response as the result and tend to pass lighting as clicks instead of
crashes. I've created ladder filters both ways. One time I built one to
sharpen up a panadapter. My first design with equal value capacitors
would ring easily when swept, then I knuckled down and designed on with
a Bessel amplitude response for maximally flat phase response and I
could sweep it at least ten times faster before ringing. I notice that
in the old Q-5er (BC-453 when operated with the IF cans under coupled.
It will take a lot more rapid repeated pulses to fill up that receiver
than the typical crystal lattice filtered receiver or a mechanical
filtered receiver. As far as I can tell my old Collins mechanical
filters are the ringingest possible filters, useless in the real world.
73, Jerry, K0CQ
--
Entire content copyright Dr. Gerald N. Johnson. Reproduction by
permission only.
|