Investigating transmit gain in UBITX V3 using LTSPICE, Aug 28, 2018

Caveats:

all modeling done at 13.8vdc

For simplification:

q90 collector circuit modeled as parallel feed with 400 ohm load

q911-q912 collector circuit modeled as parallel feed with 200 ohm load

q92,q93 and q96,q97 collector circuit modeled as parallel feed
                    with 50 ohm load,each side

observations:

in the bidir amps the 2n3904 gain rolls off 1.2 db from 5MHz
to 45 MHz...not real bad; I believe 2n3904s are quite usable
in the bidir amps at 45 MHz.

In the TX:
q90 rolls off less than .5 db from 5 to 30 MHz

same for Q911-Q912

same for Q92,Q93,Q96,q97

so, adding q90 rolloff, and q911-912 rolloff, and q92,q93,q96,q97
rolloff I get (.5+.5+.5)= 1.5 db rolloff from 5 to 30 MHz.

My most recent ubitx v3 has 10W out on 40M, but less than 2W out on 15M.
That is more than 7dB drop.

I take it from the above that replacing the TX driver transistors with
higher Ft devices may not substantially improve the 15M performance.

WARNING: q911,q912,q92,q93,q96,q97 all dissipate at least 296 mW. That
means 250 mW transistors are not acceptable.

So, in searching for a better Ft device for the TX chain:
the BFR106 has the dissipation, but the 3.5 to 5GHz Ft invites instability
the MMBT10     lists dissipation at 225 mW, so it may not be a good candidate
the MMBT2369 lists dissipation at 225 mW, so it may not be a good candidate
the DTX2222A has the dissipation, but the Ft is not better than 2n3904, and
the footprint is not compatible with the board.

So at this point, I do not see a real winner for 2n3904 replacement.
I am sure open to suggestions on this front.

All of the above really leads me to question the power amp circuit..
and I do not have the wherewithall to model that successfully.

Any ideas there?

Rod KM6SN