After all it seems to me that making the exit of the phase plug as small as possible is still the best solution for an even response. Also, the smaller it is, the less important the actual curvature of the wavefront will be as it more effectively averages out. At least these are my impressions so far. The principle I have presented may look interesting on paper at first but it seems to be only magnifying the problems in the end. This modal behaviour can be quite nasty and the remedy overwhelmingly complex.