I've seen some designs that look like they have 2 conical sections within the horn's horizontal expansion. Examples are the Peavey MB-2 and Rog Mogale's MT122. The first section, attached to the throat, appears to be about 1/2 the coverage angle of the second section terminating at the mouth. Is this still a conical horn, or is it a crude approximation of another horn type? When the wavelength becomes shorter than the first section's depth, does it beam at that coverage angle?

• Room corner characteristics

As for the wider angle near the mouth of a CD horn, that's done to try and maintain constant directivity as a horn begins to transition to a diffraction slot at the low frequency end of its range, just before it loses pattern control. Looking back at the patterns made by single-slit diffraction, notice the ripple in energy distribution out to the sides of the main pattern. As frequency drops, just before a horn loses pattern control and the pattern widens up, it becomes slightly narrower through a band of frequencies. Spacing down the horn from the throat, if you widen the flare at that point, it will tend to widen the pattern through that transition region. The idea is to widen it up just enough to match the wall angle from the throat forward to the transition point.

I've bookmarked your link to room corner characteristics. It's valuable information to non-techies like myself! I've done as such with several of your posts, Wayne!

The reason why I haven't yet shared this design is because I want to make sure that the forum isn't designing it for me. I'll post an outline of it for reviews and guffaws when I'm done tweaking on paper!

Actually, I simulated my horn (as is) in HornResp about a month ago. I modeled the Delta 10A, since you already showed that it worked well in a front-loaded horn, plus it was an inexpensive choice at Parts Express. The throat, being a slot, wouldn't model just using the total area, so I pretended that it was a square of the smallest dimension (the width). The horn was modeled in 4Pi space as the worst-case scenario for a live SR subsystem. One box was a little disappointing, but 4 boxes stacked as a line array --usually the minimum box requirement for most pro audio line arrays-- flattened right out and rolled off close to the lower cutoff target. The upper cutoff was a little higher than in your Midhorn (which I used as a reference point, thank-you) though it was inconclusive whether the 2kHz goal was met.