J,
Ken's ideas about parafeed SE are right on the money. I've got
some buddies who build hardcore SET hi-fi amps, and I can assure
you, a well executed parafeed can sound killer. But it takes a lot of
fiddling to get it right and a lot of their early efforts did not sound
any better than my benchmark Dynaco ST-70. Plus, these guys
spend a fortune on iron, and years tweaking their creations.
 
I have a Fender 135 PA amp that someone had modified into a
guitar amp, but it sounded rather sterile...the usual complaint with
Fender's ultra-linear designs.
I rewired the power amp section along the lines of the Vox
AC-100 power amp:
http://www1.korksoft.com/~schem/voxamps/ac100_poweramp.gif
I kept the ultra-linear connection with no NFB. It really has a nice
warm 2nd harmonic vibe. 
 
Notice the phase splitter is a floating paraphase. This type does not
inherently cancel 2nd harmonics as does the more common
long-tail. It is not as well balanced as the long-tail, causing less 2nd
order cancellation in the output stage. It also has phase shifts and
other weird stuff going on that looks bad in theory but sounds great
in practice.
A more detailed explanation of this can be found on Gabe Velez's
site http://members.tripod.com/~gabevee/pptube.html
 
I recently repaired a Leslie 147 power amp. It had a really warm,
rich sound for a 6550 push-pull amp. The splitter is the old-style
paraphase, commonly used in old cathode biased power amps.
http://home.image.dk/~isl45293/schems/147.GIF
Note this amp does not use a global NFB loop but does use local
NFB on the power tubes. This may help tame the pentode
harshness a bit. 
 
Anyway, I hope these ideas are useful.
 
Regards,
Mike D.