Baffle edge diffraction on a dipole baffle.

John Whittaker showed me a couple of frequency magnitude graphs with a series of dips and peaks "moving" in function of the frequency as distance to the dipole baffle increased. I suspected the triangulation of line source and baffle edge diffraction source (virtual line source) to be the leading cause of this cancellation pattern. I derived a simple model for this case. The plot below shows a calculated dipole baffle response with following assumptions:

The response of the driver is perfect
Diffraction is perfect and is equal for all frequencies. (Diffraction on a baffle is frequency dependent.)
Diffraction is calculated for a one sided baffle as was the case in John's measurements.
The baffle is rectangular

None of this applies in the real world but it makes the calculations easier.

Notice the large suck-out where the combination of baffle edge virtual line source and real line source coincides with the first dipole cancellation point. (680 Hz) Also interesting to note is the "enhanced" low frequency support of this situation. Without any diffraction the response of the above simulated ribbon/baffle combination would drop at a rate of 6dB/octave below 340 Hz.
Measurements have shown that keeping the diffraction on the baffle low will result in greatly reduced "moving dips" in function of distance to the dipole baffle driver combination. An immediate result thereof is the observation that the dipole cancellation dips are more consistent in magnitude. Since there is less diffracted energy before the rear and front waves cancel beyond the edges, the dipole dips will be slightly larger in magnitude.

To be continued.