I've been poking at the amp tonight, trying to compare it to simulations in LTSpice. I've noticed at least two places where I actually have different values than the original circuit (some DC blocking caps and the choke on the final), but putting those back into the simulation didn't seem to affect much. Definitely getting different readings, but haven't found the cause yet. Frustrating.
On the plus side, still making it really far on wspr :)