As per Floyd Toole's book, loudspeaker frequency response flatness isn’t everything to aim for. Smooth off axis response is very important too.

Given I still use my Radian 950PB on JMLC 340 Hz horn, those JMLC horns profile do beam like crazy as the frequency increases.

Tractrix flare do much better in that regard.

​

Initially, I thought matching the narrow dispersion profile from the JMLC at 6500 Hz (it's around 40° at this point) to the Beyma CP22 40° conical horn, would be a match made in heaven.

And while it does work, the 340 Hz JMLC dispersion is around 120° at 500 Hz. Ouch! Quite the narrowing profile to a mere 40° at 6.5khz.

​

After reading Floyd’s book, I then decided to re-try the Celestion CDX1-1425 mounted on the FaitalPRO STH100 horn.

Yes, the Celestion has a slight downward response curve and requires equalization, but the benefit is a much wider dispersion.

​

Crossover point was arbitrary chosen at 3200 Hz, one octave lower than before.

Reasoning:

• The JMLC has around 70° dispersion at 3200 Hz

• The STH100 has around 70°  dispersion at 3200 Hz

• It would be a great directivity match.

• then have the CP22 kick in at 12khz as the CDX1-1425 reach its limit.

​

The sound system now goes from beaming more as frequencies increase, to much wider coverage and more constant dispersion. (Floyd would love me)

It does increase complexity as I'm moving from 3 ways to 4 ways, cost does go up as well.

Of course, the Celestion step response isn’t as good as the CP22, but it plays cleaner at lower frequency. In fact, it's able to drop to ~2.5 kHz with acceptable distortion profile and wide coverage. Step response of the 1.4" diaphragm Celestion is also much better than the big 4" diaphragm of Radian 950PB, diaphram mass matter.

 

Overall, the directivity of the new system will be much wider. Couldn't wait to test it out.

On the unproven subjective claim department, I’d have two Neodymium compression driver instead of one NEO (950PB Neo) and one Ferrite (CP22)

Once equalization has been adjusted for the CDX1-1425, I listened to a few tracks.

​

WOW! I never dreamed it would sound so much different.  In fact… incredibly better :)

OK, it’s now October 2018 and I'm still living in my temporary rental house in Alberta. The living room is far from optimal and the acoustics suck, but still, results are very promising. It engages the room much better and sounds more coherent with mids and bass.

This configuration will be a keeper for sure.

 

Looking at commercial Auto Tech’s horn speakers, they’re using a large 4" compression driver mounted onto a big horn, matching it with a small 1" CD on a smaller horn. There’s no crossover point disclosed, but I imagine they’re between 2.5 kHz to 4 kHz. They must have tried the same test I did.

​

The Cessaro flagship horn system employs a similar concept, but they go further and add a TAD ET-703 super tweeter on a custom horn.

This makes sense. They did their homework. 

​

I’ve got visit the Munich HiFi show to hear some serious horn-based systems. Why aren't they more popular in North America is a mystery to me.

​

----------------------------

​

Below is a second set of measurements, in same setting as the CP21 and DE35, for a better comparison between the 2.