A well-flogged dead horse...
I finally got the trap cabinet configured with my arrangement of the Pyle/PE waveguide and the four 5.25" midwoofs (arranged 2M-T-2M) and made some measurements. Previously I'd tried this in a cardboard box test 'cabinet' and the results were murky because of the shaky box.
The results this time weren't so murky, just simply bad. The intention was that having smaller drivers close to the guide, and both above and below, would improve the vertical characteristics, and make the woofer pattern better match the tweeter's narrower vertical pattern. Measurements were made outdoors, over hard pavement (so there is a strong allison effect comb pattern)
First is a set of responses of the woofers by themselves, at 7.5 degree vertical steps, all normalized to the on-axis response. There is a notch forming below the intended ~1.5kHz crossover frequency range, and a response flying back up again (at 22degrees) right around the crossover point. Which is worse than a single larger woofer would give.
But the real fatal blow for this concept is in the tweeter curves. Something I hadn't paid enough attention to in other plots posted here by others -- the tweeter's vertical pattern control is nearly gone already above the crossover region. Vertical pattern control doesn't go as low as does horizontal control, it seems. So making the woofer's vertical match would be better met with smaller vertical dimensions, not larger!
I may still try to rescue the concept by the same driver set in a 2.5-way approach -- have only 1 set of woofers play into the midrange, which would provide a narrow center-center distance between midrange and tweeter to move out the vertical nulls. And let all four woofs play into the bass. I think that could be done by wiring woofers in series-parallel (horizontal pairs wired in parallel) and one of the parallel sets bypassed with a ~50uF capacitor. Should keep about the same response overall, but with the mids coming from just one row. Am I thinking that correctly?
Comments? Suggestions? (give up on the woof array and just wire in the 123A-1's?)