To get a delay of 1.5", or a fraction over a tenth of a millisecond (13.5"), just add a Paraline that's 3" tall by 1.5" wide.
The reason that you'll use one that's 3" tall is because the sound expands radially, so a 3" tall Paraline has a pathlength that's 1.5" long. (The radius of the disc that's folded inside of the Paraline.)
Having said that, I'd be careful about using the Paraline for pure delay, unless you have DSP handy. We don't have an easy way to model these things 100%, but from what I can see in Hornresp, they *seem* to be more sensitive to spacing than a regular Unity horn.
Here's my hypothesis on why this is:
In a regular Synergy or Unity horn, the compression driver does not 'see' the midrange horn, if you size it properly.
Basically there is a 'sweet spot' along the length of the conical horn where the horn is large enough that it's no longer providing much gain for the compression driver. So the sound of the compression driver is basically constrained into a cone, and it sails right past the midrange ports.
In a Paraline, we don't have that luxury. The volume of the horn is so small where the midranges tap in, they're *definitely* visible to the compression driver.
That gives us two options I think:
option a - get the drivers really really close together, so close that they're within 1/4 to 1/3 wavelength at the xover point. With an xover point of 1500hz, that's about 2.25" center to center. This is tricky, but do-able with 2" midrange like yours. The downside to option a is that it does not account for the delay that exists at the resonance of the two drivers.
Danley has hinted around about this, and it seems to be one of the reasons he likes the BMS 4550 over comparable options. (Check out the impedance curve of the 4550 and you'll see what I mean.)
option b - just stick the midranges outside of the Paraline. This is what all of the commercial offerings do. For instance, in the smallest VTC box the Paraline has a diameter of about 4", so the center-to-center spacing is about 8". That type of spacing would require an xover of about 422hz (if we ignore the delay imposed by the various resonances.) Obviously, 422hz is way too low an xover, so Yorkville and VTC use DSP delay on their boxes.
Obviously, the easiest way to get everything in phase is to simply build the darn thing, and then measure the impulse response. One very nice thing about the Paraline is that you can move the midranges around on the horn by simply replacing the bottom plate. You can literally build that plate in five minutes.
It's way way easier to fabricate than a conical horn, and it really makes it easy to experiment with the phase response without wasting aeons in the garage.
At the crossover point, moving the midranges an inch can spell the difference between a peak and a dip in the midrange, so having the ability to move the midranges so easily is really nice. If I had to hazard a guess, I believe the only reason that Yorkville and VTC don't put the midranges on the Paraline is that it severely limits your power handling. Due to the very small sizes involved, any midrange larger than 2-4" is basically impossible.
Unless you put the compression driver UNDER the midrange, like this: