Romy the Cat wrote:

Make an experiment, and I will simplify the case quite assertively. Take a typical compression driver, cross it at 800Hz, second order and load it into a proper contemporary horn (Tratrix or JMLC) of 300Hz. Listen that horn, you will get some sort of sound that let accept as OK Sound. Now begin to very slow lower the crossover point and keep listening the channel. While you lowering the crossover point, somewhere around 550Hz (I took this number purely arbitrarily as it would depends from VERY MANY circumstances) the horn will begin to demonstrate what I call “choked sound”. The “choked sound” is HOW HORNS SOUND IN 99% OF ALL HORN INSTALLATIONS OUT THERE – people just too damn to deal with it. The “choked sound” is the satiation when Sound produced by a driver can’t be “processed” by horn. In this “choked mode” a horn produces the “sonic boom” that was made by the horn’s bell and that “sonic boom” screw up the enter band-pass of the channel - the game is over.  Increasing of the crossover point for ¼ octave (for instance) will fix the situation - so we have approximately one octave between horn’s rate and mix crossover point…

sser2 wrote:

(in order to measure dynamic compression)… Measure frequency response at 2x power increments: 0.05, 0.1, 0.2, 0.4, 0.8, 1.6, 3.2, 6.4 and 12.6 W. Should get family of similar curves spaced at 3 dB. Spacing of less than 3 dB will indicate compression.

What is very interesting is that mentioned above “choked sound” or “sonic boom” is always accompanied with dynamic compression at the bottom knee… If so, then is it possible by tracing the level of parallelness of response curves at different volumes (not power as  ‘sser’ said) objectively find a distance at which a given horn would not be “choked” by a given crossover?