Hi Toymaker. I know you mention a swirl movement of air inside sphere, but have you thought about maximizing it?
Swirl effect separators are among the best in what concerns efficiency and reduced losses; and the trick is to lengthen as much as possible the travel of stream in spiral.
Personally I would see air admission in the spere in horizontal plane, tangent to the sphere's wall (not 90 degree), at the height where you have already fitted it; and exhaust vertical at the top (maybe continued inside sphere towards center) so not to affect swirl's movement and collect potential droplets.
Clean air has to make a steep turn at the bottom to go out upwards and you have (as much as possible) a general, coordinated movement of particles, which should be better.
View attachment 155246