Johnk, you're more than welcome to disagree with me. If we always agreed, one of us would be redundant!
In my experience (which is of course limited), it simply isn't feasible to use a minimalist high-pass filter topology with a constant-directivity horn because of the equalization requirement imposed by such horns. A tractrix is not a constant-directivity horn so a minimalist crossover is more likely to be feasible with it, but even then it probably calls for just the right compression driver.
To get more specific regarding driver integration with horn speakers, in my experience problems can arise when you have a high crossover (3 kHz or higher) between midhorn and tweethorn, and the midhorn is large enough that there's a significant distance between the throats of the two horns. The ear's ability to judge the height of a sound source is pretty good at short wavelengths, but poor at long wavelengths. So driver vertical integration is a function of not only vertical separation, but also the crossover frequency (and maybe crossover slope, but I'm not sure about that). In many cases a two-way horn system has a definite advantage here.
Unsound, regarding waveform fidelity, that goes to the ever-present issue of juggling tradeoffs, and my understanding of human auditory perception places other things higher in priority. Our nerves simply cannot fire fast enough to trace out a waveform, so the ear de-constructs the incoming sound based on energy distrubutions rather than waveforms. On the other hand, the ear can readily detect a broad hump in frequency response even if the height of the hump is only 1/2 decibel. So I prefer to juggle tradeoffs in favor of what I believe to matter most, namely frequency response, with particular emphasis on the off-axis frequency response because that's where the most opportunity for improvement exists (in my opinion, anyway). Now I will readily admit that the ideal would be waveform fidelity without tradeoffs in the frequency response domain, in which case the tradeoffs shift into the monetary domain.
In his landmark book "Sound Reproduction", Floyd Toole lists five measurements that have been experimentally shown to correlate well with subjective preference. Four of the five have to do with off-axis frequency response, and the fifth is the on-axis frequency response.
In my experience (which is of course limited), it simply isn't feasible to use a minimalist high-pass filter topology with a constant-directivity horn because of the equalization requirement imposed by such horns. A tractrix is not a constant-directivity horn so a minimalist crossover is more likely to be feasible with it, but even then it probably calls for just the right compression driver.
To get more specific regarding driver integration with horn speakers, in my experience problems can arise when you have a high crossover (3 kHz or higher) between midhorn and tweethorn, and the midhorn is large enough that there's a significant distance between the throats of the two horns. The ear's ability to judge the height of a sound source is pretty good at short wavelengths, but poor at long wavelengths. So driver vertical integration is a function of not only vertical separation, but also the crossover frequency (and maybe crossover slope, but I'm not sure about that). In many cases a two-way horn system has a definite advantage here.
Unsound, regarding waveform fidelity, that goes to the ever-present issue of juggling tradeoffs, and my understanding of human auditory perception places other things higher in priority. Our nerves simply cannot fire fast enough to trace out a waveform, so the ear de-constructs the incoming sound based on energy distrubutions rather than waveforms. On the other hand, the ear can readily detect a broad hump in frequency response even if the height of the hump is only 1/2 decibel. So I prefer to juggle tradeoffs in favor of what I believe to matter most, namely frequency response, with particular emphasis on the off-axis frequency response because that's where the most opportunity for improvement exists (in my opinion, anyway). Now I will readily admit that the ideal would be waveform fidelity without tradeoffs in the frequency response domain, in which case the tradeoffs shift into the monetary domain.
In his landmark book "Sound Reproduction", Floyd Toole lists five measurements that have been experimentally shown to correlate well with subjective preference. Four of the five have to do with off-axis frequency response, and the fifth is the on-axis frequency response.