Now if you don't sell those ER4s you could get yourself a pair of Radian LT2 waveguides (https://radianaudio.com/products/lt2-wave-guide there's an AudioXpress test of it here: https://audioxpress.com/article/test-bench-the-lt2-ribbon-transducer-and-waveguide-from-radian-audio-engineering) which should fit (the fixing bolt holes on LT2 are 92mm and on ER4 90mm apart so they are like a match, you should be able to use long bolts to fix the tweeter onto the WG) and build a test baffle, get a microphone, download VixxxxCad (that's f-word in Finnish, meaning something like a f**ked Up Cad, odd humor from the author) and start measuring from different angles (0-15-30-45-60). I don't really know those AMTs that well, they usually have quite a funky response. Anyway, you may need to use some 4" or so midrange between it and the woofer though Eton designs seem be able to go without (but they haven't published the polar patterns). Adam S3H/V has some 4" (?) dome (hybrid) midrange with smaller waveguide (compared to LT2) for the tweeter. Anyway, with the waveguide you could go as low as 2kHz, and the directivity would still match quite a large driver (maybe even a 6.5" one). There are many options for the midwoofers, and when using the WG designing the crossover may be quite an easy task, for an example a single capacitor can be enough if the 1st order crossover frequency is set close to 10 kHz using about 3.3-4.7uF cap (just a guess) - the waveguide has rising response towards the acoustic 2k crossover point. For the woofer(s) a 2nd order LP with the baffle step correction is needed, just two components (about 1.5 - 2.7mH coil with a cap like 15 uF, though the values cannot be known before measurements, they can be simulated in software).