On Sun, 15 Jun 2014, gareth wrote:

If one were to be building an RX around a scrapped Xtal filter,
that would certainly resolve the SSB bandwidth requirement,
but what of CW?

How about then using one of the USB / LSB CIO Xtals as a
crystal-plus-phasing control to circumvent the paying out
of shekels?

But then what do you use for the BFO?

Cascading the filters makes sense. The single crystal filter will be
nice and sharp, but have a lousy skirt. Keeping the SSB filter inline
ahead of it means that will limit the skirt selectivity of the overall
receiver.

Given the synthesizers and DDS boards around, a different consideration is
to forget about matching SSB and CW filters, and just build a separate IF
strip running in parallel with the SSB strip. Just program the right
offset into the synthesizer when changing modes. Terminate the mixer with
a broadband amplifier, and then feed the two IF strips in parallel.

Then you can find a decent CW filter, the frequency doesn't matter so long
as it's high enough that image rejection isn't a problem. Or, pick a
frequency where crystals are nice and cheap, and build a ladder filter
that will have better shape than a single crystal filter. It's
apparently easier to build a narrow filter than a wide one. Or, there was
a time when those single crystal filters (with the phasing controls) were
cascaded to improve skirt selectivity. Instead of the IF transformers,
they'd use a triode or bipolar transistor, the cathode or emitter wasn't
bypassed so there was the needed 0 and 180degree outputs for the crystal
and phasing capacitor, just cascade a few of those. Or another scheme,
though likely not as good selectivty, used a string of IF stages with the
cathode or emitter bypassed not with a capacitor, but a crystal or ceramic
resonator. Certainly a very easy filter, I'm not sure how narrow it gets.

The reality is needing crystals can cause rejection of some ideas or
otherwise decent filters. One could go with an HF range SSB filter, then
convert that down to 455KHz where a mechanical or ceramic filter provides
CW selectivity. But then you have to buy a crystal to make that
conversion down to 455KHz, not likely to find the right crystal except for
paying for a custom ground one. It's easy to find them to get from
10.7MHz to 455KHz, but there aren't many SSB filters at 10.7MHz. If you
can get the synthesizer to also output the needed frequency, then it
becomes less messy crystal wise.

Or, there was that scheme seen in the early days of SSB, where upper or
lower sideband was selected by using a conversion oscillator, one
frequency above the IF, the other below, so one would cause inversion. A
neat trick, and with the right frequencies they'd get away with one
crystal properly multiplied, but that sort of thing is difficult nowadays
unless one is willing to pay for the crystals.

A bit fiddley in the switching from USB / LSB / CW though,
but the other CIO should be VXOable to pull it into range.
(ISTR that VXO statistics suggest a shift of 1kHz per MHz
should be possible)

That just sounds messy. YOU complicate things so you can use the BFO
crystal for the filter, then have to make do with another BFO crystal.

Michael

(Train of thought triggered by a pile of bits left over from a
scrapped FTDX560 from 30 years ago, and an IF frequency
of 3.18MHz.)