The CT inductor the 2A3's grids are attached to is the divice in question. Ground its centre, and drive one end. The signal at the other end is 180-degrees out of phase. It is interleaved to make sure the two are of equal magnitude and phase-opposite. It does work well if driven by a low impedance source.If I didn't like a straight diff amp, as drawn for Galahad, it is an attractive option. The stage as drawn also has room for improvement( read that: well tweaked complication ). It has to look possible for the lower skill/confidence levels too. ONe can either spend a lot of time with support, or a lot of time simplifying. I think the end result of the former results in better amps...:) The end result of the latter is greater profit, so that's a non-issue in this forum.
cheers,
Douglas