Originally Posted by noah katz
So what do the 10" mid-horns better than a quartet of 2226's?
The only thing I can think of is CD, but is the difference that significant in that freq range?
A quartet of 2226's can't play up to ~1khz for the handoff to the SEOS horn. Well, they can, but they shouldn't because of null and comb filtering issues. You could take the quad 2226's and roll off the top end of each one so that only 1 driver plays up to 1khz, the driver next to it up to 700hz and the two bottom drivers up to about 450hz. This would solve that problem but add complexity.
The mid horn also provides a few interesting benefits. It provides a center to center distance similar to a 12" woofer instead of a 15" woofer so the vertical nulls are more easily managed. It also should control directivity lower and better than a single 15" woofer from 500hz to 1khz. Then there is the advantage of off loading some output duty allowing the quad 2226s to focus more on midbass and upper bass. This reduces IMD related to a driver covering the high excursion demands <80hz while also playing the 1khz range.
Now if the question is what frequency range should each section cover, that is a tough one to answer. According to my sims, the midhorn's excursion starts to rise a bit below 400hz. The quad 2226's shouldn't played above about 500hz. That would suggest the 400-450hz crossover range would be a good start. Also, these are not corner loaded like in the 7Pi speakers so I'm not sure how well they will play below 400-500hz.
The usual disclaimer applies though concerning the tangible benefit of these advantages. Will the speaker sound better than a SEOS-12 and single 2226? It should, but how much better is difficult to ascertain.
BTW, Scott, in looking at your FR chart, I can see a few issues that might be due to the rising on-axis response of the midhorn. I know you are just getting things squared away, but I think there can be some big gains with better crossover optimization. You might try raising the level of the midhorn 2-3db and placing a shallower slope filter around 700-800hz instead of what is likely closer to 1-1.2khz right now.