3-way design based on Troels Gravesen's CNO-GRANDE

[Rodeodave]

I've been toying with the idea of building a 3-way speaker based on Troels Gravesen's CNO GRANDE design CNO-GRANDE for some time now. Why not build the CNO Grande you ask? Three reasons basically. Because I want to use some drivers that I already have on hands, because I'm not interested in overpaying for the kit from Jantzen audio, and lastly because the crossover values were not made public, so I'd have to reverse-engineer it anyway.

The CNO Grande is using the following Seas drivers: two W18NX001 woofers, a modified Seas W12CY0003 mid, and a T29CF002 tweeter.
E0042-08S W18NX001
E0044-08S W12CY003
E0040-06 T29CF002

Here is a link to the modifications to the mid driver: W12CY003

I'm guessing it's a few coats of surround glue applied with a brush, with the driver sitting on a low rpm turntable. I'll be doing these modifications myself, monitoring the results via measurement.


My design will be using the following drivers: two Seas U18RNX/P, a modified W12CY0003 mid, and a Morel ST728 tweeter. I'd get the enclosure professionally built, I simply don't have the time, space, tools, nerves or skills for that, sadly.

http://www.seas.no/index.php?option...e&id=102:h1571-08-u18rnxp&catid=44&Itemid=461
E0044-08S W12CY003
https://www.morelhifi.com/product/st-728/

Why this selection?

I currently have a TQWTL running the U18RNX/P and ST728 combo (factory matched tweeter pair), so I already have like half of the drivers.

And I'm convinced that the U18RNX/P is a better driver than the W18NX001 (W18 in black, U18 in red, matched sensitivity; look at the FR spikes in the W18 graph).

The ST728 looks better too. In my current 2-way design it has astonishing, near-constant directivity, resulting in a rock-solid stereo image. (T29 in black, ST728 in red, matched sensitivity)

I tried to model the crossover in VituixCAD using traced SPL and impedance curves, which are attached with this post (CNO_clone.zip). This is what I have come up with so far (XO points at around 1kHz and 4kHz, notice the 21mm setback of the tweeter):

I'm not sure I like the group delay and phase response.

I realize that the simulated response is not going to match the real-world, nor the in-box response, and much less the in-room response, but I have to start somewhere. The modifications to the W12 aren't taken into account either, obviously.

Now on to my questions:

How should I go about translating the simulated crossover into a real world build? I have a UMIK-1 and REW, so I can make basic measurements, although I have yet to spend more time with it.

Does the crossover topology make sense for my driver choice? Any criticism is welcome.

I really hope to get a discussion started, so any input is welcome.

 

[Rodeodave]

Any input?