A) There's really no reason to "boost" the selectivity of the BC-453.
They were once seen as the ultimate in selectivy, at a time when
there were few options. If the IF is too wide, then you need to
adjust it.

Agreed. In fact I am looking for extra selectivity in terms of image rejection,
not in terms of smaller channel bandwidth. The BC-453 selectivity is enough for
me!

B) The Q-Multiplier works by adding regeneration, but the prime
point is just before it oscillates. Which means that when you need
it to oscillate to provide a beat note, you lose control of selectivity.
And the proper point is so sharp that external factors will kick
the circuit into oscillation. Which is where the superregen came in,
keeping it at that crucial point of highest gain without the need for
crucial tuning (but incidentally bringing other problems into play).

There is no need to push the regenerative converter into oscillation for the
purpose of generating the beat note. I can easily get the beat note from the
BC-453 built-in BFO.

C) If you have a self-oscillating converter, which seems like you
are talking about, note the regen to the point of oscillation is
at the oscillator's frequency. That's not going to do a thing on
the signal frequency, and it's not going to do a thing on the
IF frequency. A mixer by definition has a different frequency on
the input from the output, which does not make for proper regeneration.

As I wrote in my initial post, I am supposing to have a separate local
oscillator that feeds the conversion frequency into the converter stage.
Therefore the converter is not a self-oscillating one. Nevertheless the
converter is regenerative because its plate coil is linked to its grid coil, so
that it would tend to self-oscillate if one does not properly control the
cathode resistance. Such circuit is called Q-dyne, see QST 1938

.. Unless they were modified, most BC-453's were used in multiple conversion
schemes. Either tapped into the 455KHz IF of an existing receiver (and
if the existing receiver was single conversion, leaving its image problem
intact), or as a tuneable receiver with some sort of converter ahead of
it. The better ones used two stages of conversion, to get around the
image problem that arose with a high signal frequency dropping immediately
down to 455KHz.

I know, but my purpose was to get good image rejection with just a single stage
(too easy with two!) and I was wondering whether the Q-dyne approach would do
the job.

73

Tony I0JX