Step 20: OH, Man...I wish I had...
--Used a heaver, possibly stock, metal chassis.
--Substituted a 10 inch speaker.
--Used a turret board, instead of point-to-point. The more I modified the project, the more a tangled web it became. After this project, layout for turret board makes perfect sense--tubes on the back, controls on the front and the component board between....
So, additionally, the chassis layout sucks, too. Initially my main concern was to keep the preamp tube far away from the power transformer. But the lead-dress is awful. The "star-ground" is in the wrong place, too.
It must be said: despite the "rat's nest," the amp is quiet.
--The filament tap is not separated from the HV secondary. Too bad I didn't just use a separate 3A 6.3v transformer for the filaments, and added the 6-7V from the main transformer to the HV tap (about 149V, vs 142V.) After all, I had to add the wallwart for the preamp anyway....
--Used tube sockets with some sort of retaining clips--the chassis is upside-down. It hasn't been a problem yet, but eventually... All new ceramic sockets might be better, too.
Oh, and I did add a negative feedback loop. Although it did lessen the punch and (audio) feedback, and tone down the bass, it added a bit of "fartyness," too. That flabby, farty tone was present whether the nfb was switched in or not.
So I ripped it out right away.
Sometimes this is just a solder-joint issue, sometimes it's a bad cap. But it might be a routing problem, which at this point is beyond fixing without a complete rebuild.