But what crossover can avoid all such colorations? Does this mean that multiple driver speakers are doomed to be colored unless high-resolution digital crossovers are used, with the given multiplication of DAC-to-speaker paths?
One interesting crossover I was looking at is this one:
http://www.geocities.com/kreskovs/Crossoverdoc.html
Other good discussion here:
http://www.silcom.com/~aludwig/Sysde...ove_Design.htm
I already have one capacitor in the signal path (DC blocking) in my DAC. This causes some phase non-flatness at the frequency extremes. However, I have read that a) when looking at it graphed as group delay, it's only important to be flat above 100 Hz or so, and b) continuous and smooth curve is important, not just the flatness. Discontinuities are very audible, which tends to fit with the article gaboo found.
I have heard hifizen's (from diyaudio) tube speaker amp, which is not even OTL, and it has an even less flat phase, yet it sounds quite nice. That either means the specific phase non-linearity is not very audible, or that it's the kind of distortion that may be called, er, euphonic coloring. However, I did not get a chance to evaluate it for imaging, where phase is very important (though again, not in the low end). Of course, most recordings already screw up imaging; in the end only binaural recordings with headphones, or complex crosstalk cancellation processing in specific speaker setups can really reproduce it properly, so the playback system's faults will not be apparent.