Please let's not confuse time alignment with phase, as so many still do. If you are using analogue active or passive crossovers time alignment is virtually impossible, a doddle with dsp and a programme such as ARTA and an impulse response measurement.
I am using horns and a horn super tweeter on an active analogue system, crossing around 8Khz. The mid horn has a passive 1st order LP and the tweeter an adapted passive 3rd order. I established putting a 10 (8-12) ohm resister in parallel with the first capacitor of the HP filter took the knee out of the 3rd order roll off and now created a gentler
- 6dB at the point the 1st order LP was -6dB, and voila, as they say...reverse tweeter polarity, absolutely flat response through the crossover point and still sufficient attenuation at lower frequencies to protect the tweeter. Xsim was invaluable at simulating the response from the circuit designs and proved wholly accurate when compared to the ARTA plots (calibrated mic) taken in real time. Both free programmes so no excuse for subjectivism and comment with no scientific imperative.
The impulse responses clearly showed that the time arrivals were different (as expected) by around .5ms, but the response was ruler flat and minimum phase plot showed the pair of drivers were in phase.
The wavelength at 8000Hz is around 40mm..at a listening distance of say 3m or 3000mm it represents a difference in distance of .013% . Given these frequencies are also easily reflected and absorbed and you are allowed to move your listening position a bit, the chance of you hearing the difference in arrival time of .0005 seconds of direct sound ( particularly where the ear is not so sensitive as say around 1kHz) is probably unlikely. The chance of you hearing the effects of a suck out of around 6 dB at crossover is entirely likely particularly as the response anomolies will vary on and off axis.
It's a choice you have to make because everything is a compromise. You may like the analogue route (I have gone back to it from 8 channel of dsp) and bear in mind once chosen everything in the chain, bar a dac to convert digital sources (RME), should stay in the analogue domain all the way through. I opted for Rane 23S active crossover which enabled me to use the "delay" function between bass and mid and mid and mid horn..it phase shifts, and is not physical time delay as in dsp, but you can tune for minimum phase and flat response through the crossover with the 24dB/octave LR filters. The mid horn to super tweeter is a passive network as outlined above.
Or you can go digital with a super clean 96KHz+ , 24 bit processing (and upwards) distortionless LMS and make all you adjustments, eq, crossover, and time alignment in a simple 1U controller. Should be a no brainer really, and for professional application its difficult to argue the benefits of any other system. But we are humans and personally I like the "sound" of analogue, class "A" valve EQ,s, etc, so having to tolerate an error of .0005 second time difference between super tweeter and mid horn at 8kHz is a compromise I can make. In truth, I lose little sleep over it!