Spent some time tonight fiddling around with my idea for a bootstrapped Mullard 5-20-style front end (AF gain + inverter stages) for a 6550 U-L amp. The simulation behaved more or less exactly as I expected it to (modulo my triode models not liking a CCS in the inverter tail for whatever reason), giving an open loop gain of ~70 for the gain stage, and ~6 for the inverter using 6SN7 and ~5 using 12AU7 (with no additional circuit changes). With -20dB of feedback, using the 6SN7, that would allow me to drive the amp to full power with ~750mVrms.
(Note in this drawing, there are a couple of extra nodes to make probing the simulation more convenient.)
(Note in this drawing, there are a couple of extra nodes to make probing the simulation more convenient.)