With some suggestions from Tom Danley himself, I put together a passive crossover circuit to approach linear phase response for the coSyne design. It works pretty well, not perfectly linear phase of course but a lot closer than nearly all speakers. Response isn't quite as smooth as with the last crossover, but I don't think the difference is enough to worry about. A plus is the crossover circuit is simpler and less lossy than the old one.
Here is a shot of the frequency and phase response, 1/6th octave smoothed:
And for those who think 1/6th octave smoothing is cheating, here is the unsmoothed (the delay plane is a bit misadusted on this shot):
The phase rise at low frequencies is because the bass design is ported, so will be approaching 180degrees phase at the port frequency (i.e., not linear phase in the bass, but with room reflections, nothing really is).
Here is the impulse response:
Not perfect. but pretty darned tight. I haven't checked to see what it does to square wave shapes yet.
And here is the crossover circuit:
Drivers are the same as before (buyout mids and woofers, Celestion CD driver.
I built it on a piece of 1/4" plywood, twisted component leads together and soldered them, then hot-glued each to the board. Very quick and easy if you are used to soldering. Here's a photo:
And here it is mounted into the cabinet (you can see all my messy wiring.. no magic cables for this guy!).
I sent the driver measurement files to Tom Danley, and he was kind enough to do a design (in the same day) for an active crossover for this speaker including EQ! The active crossover hardware I have (miniDSP) doesn't have build in sections for high order Bessel filters as Tom specified, though, so I will need to do some research on how to implement those using biquad sections in MiniDSP. Or else get a different DSP crossover system to try it with.
Hope this helps any builders out there.