Danny, I read what you referenced. In your test, the tweeters were 6.5" (center to center?) apart. Have you done further testing where the centers were closer together?
Yes. the effect is less dramatic when the acoustic centers are closer.
My understanding (flawed as it may be) is that there is a correlation between the frequencies generated and the distance between the center of the drivers.
That is correct.
The higher the frequency, the closer they needed to be together. When proper distance was maintained, combing was avoided, and multiple drivers would create a single wave front.
Yes, for two drivers.
But when you have a line of four of them then you also have cancellation between driver 1 and 3, 2 and 4, and 1 and 4.
<speaking out my nose>
If the obliqueness of the angle were problematic because of time delay, why doesn't a large radiator like a Neo 8 or some of the larger 15 & 30 inch ribbon tweeters, sound bad off angle?
Vertically, yes. Anytime you have a line of drivers, or one long driver, you have limited upper frequency range limits above and below the line.
Wouldn't the sound from one end of the driver reach the ear at a different time than the other end of the driver?
</speaking out my nose>
But with a continues line of planar drivers (close together), or a single long line, when some drivers are creating comb filtering that drops out the response. Other drivers that you are still on axis with fills in the holes. So what you get is a frequency response up top that is +/-2db or so. And changing heights a little changes the placement of the +/-2db peaks and dips.
Again if I were closer I would run the center channel over for your free check up, and you could test it with your equipment. I am a rather empirical kind of guy, and I think you may be as well.
That would be good.
And running a short line is the same as a tweeter of the same height. So with four tweeters stacked you have a narrow range about the height of the tweeter middle tweeters when you have a reasonable response. Above or below that the upper ranges will really drop off. This creates an uneven room response and takes away imaging and spacial ques that are found in those top octaves.
Generally go with a single point source tweeter for more even coverage, or create a line long enough that keeps you in the near field.