[Elecraft] Tight Skirts...

[Geoffrey Mackenzie-Kennedy]

After having fingers burnt while designing very 'srtrong' receivers, may I
add comment.

IF filters with very tight skirts do allow the recovery of very good SSB
audio or CW. To do this the filter must NOT have sharp corners at the top of
its 'response', the corners should be a little rounded. i.e. the response at
the top, often the top 6db, should "approach Gaussian". My understanding of
this is that the Group Delay through a filter with SHARP corners varies
greatly especially at the corner frequencies, resulting in "raspy" SSB or as
you put it, trash on CW. After the slight rounding at the top, the skirts
can drop at a rate of knots. Cohn style Ladder filters are poor when Group
Delay is considered. Other types of Ladder filter at mid HF using 14
crystals can yield SSB bandwidths, <0.2db ripple, symmetry, 6/60db shape
factors of 1.2 or sometimes a bit less, and stopband attenuation of 100db.
How the filter is built and terminated are very important considerations,
and for my part I prefer ladder filters with an even number of crystals for
reasons of "mesh tuning".

SSB and CW bandwidth filters can also be built at low VHF, but the CW ones
difficult for mass production.

How much filter stopband attenuation is useful at a given offset is
determined in the greater part by how much noise is produced by the LO
driving the mixer ahead of the filter and post mixer amplifier (if the type
of mixer requires the isolation). Reciprocal mixing is one bugbear. To take
advantage of 100db stopband attenuation the LO must be quiet.

The effect of noise/signals on the skirts is tied to the linearity of the
filter, and Ron, here we could spend hours!!

The filter between the ears is super, it also tracks!

73,
Geoff.
GM4ESD

............................................................................
................
Merlin, W3ICT wrote:

...nobody else I've observed on this net seems to be as  annoyed as I about
the bad
skirts, high ripple, bad ultimate selectivity,  etc. when the K2 is used
with the
KSB2. ...On sideband, tight skirts (the electronic kind, gentlemen!) are
terribly
important, maybe more so than on CW.  There is nothing neater than
listening to
a weak SSB signal through an IF system that has a 1.2:1 shape  factor.

--------------
On March 04, 2005, Ron AC7AC wrote:


My SSB operating is only casual but I can say with half a century of
operating behind me that I absolutely LOATHE steep-sided filters on CW. They
produce a lot of "trash" in the bandpass caused by the way they modulate
signals on edge of the bandpass. Of course, signals varying at all in
frequency, including noise, are amplitude-modulated by the filter slope,
with causes further sidebands, etc.

A lot of work has been done by both engineers and mathematicians working out
the best shape to minimize these effects, but this ol' operator simply likes
listening to a big chunk of the band at one time and letting my grey-filter
between the ears do the work for me.