Message-ID: <392072FD.26AF@xtra.co.nz>
Date: Tue, 16 May 2000 09:58:21 +1200
From: "vernall" <vernall@xtra.co.nz>
MIME-Version: 1.0
To: rsgb_lf_group@blacksheep.org
Subject: LF: Minimal ringing narrow filter
References: <001201bfbd78$d264a680$6d5c97d4@win95.swipnet.se>
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 8bit
Precedence: bulk
Reply-To: rsgb_lf_group@blacksheep.org
Sender: <majordom@post.thorcom.com>

I have changed the subject heading to better identify the thread:

C Andersson wrote:
> 
> >Could you post a brief text summary of your design?
> >73, Bob ZL2CA
> 
> PA0LQ filter descriptions:
> http://www.alg.demon.co.uk/radio/136/af_filter.htm
> http://home5.swipnet.se/~w-54761/pa0lq_filter.htm
> 
> 73
> Christer, sm6pxj

Browsing this material gives some interesting results.  The "secret" to
minimising ringing is in the stagger tuning of the four stages of
filtering.  It is a hybrid sort of arrangement (so is not a classic
Butterworth, Chebyshev, elliptic or Bessel ...).

The design from Harry PA0LQ is for a centre frequency of 1000 Hz.  For
an overall bandwidth of 30 Hz, the individual stages are tuned for 966,
982, 1018 and 1034 Hz.  The tuning resistors need to be accurately set,
but are of the order of 180 ohms.

GW4ALG made a variant of the PA0LQ design, centred on 800 Hz, and tuned
it by trial and error looking for minimal ringing when fed with a stream
of 800 Hz dits using a keyer used as a signal generator.  The tuning
resistors arrived at are 346, 385, 420 and 485 Hz.  The overall
bandwidth was not measured but a comment was made "the bandwidth would
seem to be less than 45 Hz" and the filter easily passes 25 wpm.

I do not have the basic design formulae for the filter stage used, but I
assume there is a critical relationship between individual Q for and
resulting bandwidth in a four stage stagger tuned arrangement.  It is
specifically the linear phase that is the desired result.  GW4ALG
confirms that with all stages netted to the same frequency, and was
hopeless for ringing and "normal speed CW would not pass through the
filter".

Some CW ops use a tone as low as 400 Hz.  While the stagger tuned filter
appears to offer a lot in the way of minimised ringing compared to other
designs, if the PA0LQ design is sought to perform at a somewhat lower
centre frequency, then I believe the R and C values of each stage need
to be scaled, as well as the low value "tuning resistors".  For a given
overall bandwidth, I would expect the stage Qs to each be lower as the
centre frequency becomes lower.  Then what stagger frequencies should be
used to obtain the desired linear phase?  Are there any filter design
experts on the reflector that could come up with a general design, so
that individuals could implement their own preferences for centre
frequency and net bandwidth?

As a further suggestion, a filter using four stages could have tuning
resistors (and capacitors?) selected by one or more CD4066 chips (four
electronic switches, has internal resistance, but is low enough to work
with 180 ohms net resistance) so that pre-set variations on frequency
and bandwidth could be selected as desired by DC control (no need for a
multi-pole switch).

Bob ZL2CA