The parallel resonance helps us to a point, though an ideal inductor would generally be better.
For example, if the inductor is self resonant at 14mhz, we'd see much more power out on 20m
than we see on 10m.? I'd prefer to keep gain vs freq somewhat controlled and predicatable.

The caps are easier to add, easier to obtain, have a higher self resonance:
? ??
Maybe add 220pF caps across all six of the 22 ohm emitter resistors, and forget the inductor?
Then add a variable cap somewhere around Q90 that can be tweaked to give the flattest response,
that setting may vary given your particular 2n3904 transistor characteristics.?
?
But the inductor in series with the negative feedback is a good idea.
If this can all be done with just 3 extra components and get good enough results across
the different uBitx's out there, I'm fine with that.

One other issue:
As Henning notes in post 45035,?the chokes L8 and L9 are best wound on a single core:
See the discussion below figure 4 on this webpage, where he discusses how the bifiliar
approach works, though he does not discuss the individual choke solution:
? ??
I was seeing significantly worse results in how the uBitx final worked compared
to the WA2EBY final in my LTSpice simulation, could be due to these chokes.
? ??/g/BITX20/topic/9615903
I may have to play with that further, not obvious to me how or if the individual chokes
at L8, L9 would impact the results..

Jerry, KE7ER


On Thu, Apr 19, 2018 at 11:00 am, Jerry Gaffke wrote:

Good question.
It's a parallel resonance:??
? ??