I tried it both with the convex side facing the GG and the flat side facing the GG but didn't really notice a difference. Is there a correct way?
It would make sense that the plano-side of the PCX to face the ground glass as this is the side that would produce the parallel stream as the diverging light from the 35mm lens is converged by the convex side of the PCX, but if it also produced desirable results the other way then who can argue that?
It is my feeling that the problem lies AFTER the GG, so putting a PCX lens between the 35mm lens and the GG is the wrong approach.
I think this is where focal length of the lens matters. I would say the focal length should match (as close as you can) the distance of the gg to the 35mm lens....then my guess is it can be placed pretty close to the ground glass. Your focal length is pretty long, I would guess this is why that particular placement provided results. I do know that the closer you get to the camer lens and/or achromat all your doing is adding to more magnification and no longer correcting hotspotting...at least not in the same way (as more magnification can be thought of as a correction).