You've got all kinds of problems going on. When you decrease box size to get your full range to roll off earlier, you are increasing its qtc which clearly puts a hump In the driver's low to mid-low end to deal with. On the top end, if you truly want to blend a tweeter, use first order slopes and stagger your crossover points on both to achieve a 6db down at your chosen crossover point.
Transparency
I've been using the same full range drivers in a few different open baffles over the years, and after trying a few series or parallel notch filters, am trying a tweeter to take care of the upper treble, above 13000 hertz. The roughness is about 12000 hertz, so I'm lowpassing the full range driver about 11000 hertz, also first order, to put the roughness in the slightly deeper knee of the crossovers.
This works better than I anticipated; better than the deep or shallow wide or narrow notch filters I've tried.
I figured in for a penny, in for a pound, and so tried out a speaker level highpass first order about 120 hertz to reduce Doppler distortion.
Using a couple of decent quality paralleled metallized polypropylene caps, I'm losing a fair bit of detail in spite of the reputation of capacitors having much less distortion than inductors.
Comparing the same drivers with and without the plain Jane aircore inductors, the midrange frequencies sound unchanged.
This effect isn't apparent with some MarkAudio, Seas, or regular woofers.
Do I just have a set of drivers that are barely good, and can't withstand being highpassed with a 147uF capacitor?
I've tried bypass film and tinfoil caps up to 2.2uF to no avail.
And trading in added Doppler distortion and relying on an undersized pair of baffles to act as highpasses to meet the lowpassed sub.