Preamp designs using a pentode stage

[loudthud]

With the smallish cap (390p on the schem I saw) maybe the goal is to feed back frequencies from the primary that are higher than those present in the signal on the secondary side?

It's something like that. I think of it as early feedback, the OT is kind of a bottle neck for high frequencies and there is a phase shift to be dealt with when you are trying to squeeze every last bit of performance out of a given transformer. The Dynaco transformers had lots of interleaving and were quite remarkable for their day. They outperformed transformers costing twice as much. The ultralinear output with plenty of NFB resulted in low distortion and a high damping factor. A little feedback from the primary side keeps things stable at high frequencies.

[pdf64]

The Selmer PA100 (MkI / II) used an EF86 in its 3rd stage for the signal level recovery following the passive channel mixer
http://bmamps.com/Schematics/Selmer/selmer_pa100.pdf
It's dc conditions seem standard, but of note is that it's incorporated into the global feedback loop, which may make it one of the highest 'A /open loop gain' NFB loops I've clocked in the MI world (EF86 x 12AX7 LTP x EL34 p-p).
The 'B / NFB' value is correspondingly small, so the degree of NFB is probably in the normal range, which was good as I seem to remember that its OT seemed fairly low spec.

Here is an example... https://sites.google.com/site/string.../princeton-aot

Kudos - that's amazing