Vertical baffle spacing and null angles
In my view, minimizing edge diffraction of the waveguide is more important than squeezing the last available bit out of center distance.
See my pic @ #813, above, and also some of Jack's baffles. He recesses the waveguide flush.
I'm hoping Wayne will step forward with the answer; from his commercial designs, though, it's apparent he doesn't consider center distance that critical an issue:
http://www.pispeakers.com/catalog/product_info.php/products_id/97
Actually it is one of my highest priorities to space drivers vertically so that the null angles are outside the pattern slightly. The importance
how far outside isn't as important as the point that they be outside the pattern in the first place.
If the system design places nulls within the pattern, that's a deal breaker. If that happens, nothing else matters. You can have perfectly flat response and zero diffraction, rarefaction, reflection, distortion, no other anomalies whatsoever on the forward axis and if anti-phase nulls form at narrow vertical angles, it is a non-viable design, in my opinion.
Think about it. Anti-phase nulls form deep notches in the overlap band. The average amplitude through the overlap band at positions near the null angle is reduced only a few decibels but at specific frequencies and locations near the null angles, there are huge spiked dips. So if the angle is narrow, say less than 10 degrees, then you basically have no forward area where response is good.
Geddes Summa, for example, has a round horn that prohibits close vertical spacing between drivers. As a result, the anti-phase nulls start at 7.5 degrees. That's such a narrow angle that the null forms straight in front of the speaker for 12 feet back! It is absolutely imperitive that these speakers be mounted in such a way that they favor upward angles, because on a stand or wall mounting virtually ensures the listener will be treated to deep 10-15dB notches between 1.0kHz and 1.5kHz. No matter how good the rest of its performance is, if you're sitting in a null, response is bad.
I have a lot of respect for Earl Geddes, but until and unless this issue is addressed using an asymmetrical horn with shorter vertical dimension, the narrow null angles impose a serious performance penalty, in my opinion. That is why I do not use or recommend round horns for an application like this.
The best thing about 90x40 and 90x50 horns is you can setup the baffle spacing and crossover so the nulls fall above and below the horn pattern, just outside. It can be right on the edge or 5 to 10 degrees outside. That's a very nice arrangement, because then what you hear as you move up or down outside the pattern is a loss of HF, just what you would expect. You don't hear a weird phasey sounding anomaly like you hear when nulls are placed in the middle of the pattern. You also don't waste HF energy directed at the ceiling and the floor.
: ]
