We’ll ya’ll knew the mad scientist would have to chime in on this one
First, the reason for all ML centers having a tweeter is, as some have noted already, dispersion characteristics.
A small (12” high or so) panel element has next to no vertical coverage above 2Khz, so they have to cross over to a transducer that has broader dispersion, which is a regular dynamic tweeter.
The reason for the woofers in an ML center is likewise, the small ESL panels can’t really go below 500hz with any volume ability, so they cross over about that point to a single or usually a pair of 6” dynamic drivers.
The fundamental challenge is that the basic ESL design is really a line-source system, where you have several linear feet of reproduction surface to cover the primary frequency range (~300Hz – 20Khz) as well as handle vertical coverage (= the height of the panel) and because of the curve applied, it can provide the horizontal dispersion as well.
But since few people have acoustically transparent screens and front projection in a darkened room (to place the center behind the screen), the usual form-factor for centers is a small rectangle that goes below or above the screen.
This size and locations are less than ideal regardless of transducer topologies, so even a box full of dynamic speakers is still a big compromise (IMHO). So getting a decent ESL-based result is an even bigger ask.
But ML’s centers are good designs (considering the challenges) and hard to do any better in the form-factor they need to adopt.
So while having a tweeter in the middle of ESL panel seems like a bit of a strange match, it’s absolutely the only way to make it deliver the kind of spatial coverage required of a center.
Now, I felt strongly enough about these compromises that I went my own way on this topic, but then I seem to be in a unique minority at the bleeding edge (of tech or insanity, who knows)