Hi Gut,
The diffraction graphs are interesting. Both the ones from Siegfried Linkwitz's site and those from the SOS article can be found, with more info, in Harry Olson's book "Elements of Acoustical Engineering" 1957. There are PDF scans of it floating around, like here:
http://www.scribd.com/akaliabetsos/d/51369856-olson-1957-ICheck out pages 21-23 (pages 41-43 in the pdf).
As you pointed out, the freqs at which these effects happen are dependent on the size of the shapes. You can see that the horizontal axis of the graph for spheres is labeled (D/lambda), where D is the diameter of the sphere and lambda is the wave length of sound.
For spheres you can see that the onset of the on axis freq response boost starts to happen at about D/lambda = 0.1, gets up to +3dB at D/lambda = 0.3, and reaches nearly +6dB at D/lambda = 1.
I'm not very familiar with thinking in terms of wavelength so I made an Excel sheet to help figure out at what frequencies these on axis boosts would occur depending on the diameter of the sphere. Here's some common sizes for the balls used on Omni mics and also the human head.
Sphere Diameter = 30mm, Onset of Lift = 1150 Hz, +3dB Boost @ 3450 Hz, +6dB Boost @ 11500Hz
Sphere Diameter = 40mm, Onset of Lift = 860 Hz, +3dB Boost @ 2560 Hz, +6dB Boost @ 8600Hz
Sphere Diameter = 50mm, Onset of Lift = 690 Hz, +3dB Boost @ 2070 Hz, +6dB Boost @ 6900Hz
Head Diameter =160mm, Onset of Lift = 215 Hz, +3dB Boost @ 650 Hz, +6dB Boost @ 2150Hz
Looking at the response measurements of the 4006 with these different sized Acoustic Pressure Equalizers (balls) shows similar trends, but they don't achieve the theoretical amount of boost (+6dB). Not sure why.
I get that reality doesn't always match theory but the responses DPA show fall back to 0dB at very high freqs. Wonder why?
Miq