Acoustic Horn Design – The Easy Way (Ath4)

[Olombo]

Beta pleeeaaaseee 🫣

Just great to see another possibility in sim coming up! Since I use ATH I did not build even 1 speaker....but I tried 10 different approaches with it and the help of the guys hanging around here. Really encouraging!!

 

[mabat]

Not all is gold. This is what you can easily get when you're not careful enough.

 

[mabat]

This is the above without any ports, pretty typical for any poorly terminated device.

And below with the lower-midrange ports added. Probably not a good idea in any case.

 

[454Casull]

Boy, the more you post the more I feel as though we are in the golden age of audio.

When can we get the midrange responses?

 

[454Casull]

Would a small radius edge around the port openings inside the horn make any difference in the simulation?

Just put a bit of felt “upstream” of the midrange ports. That should reduce the diffraction effects from/on the HF, at the cost of a minor bit of efficiency.

 

[mabat]

Since I use ATH I did not build even 1 speaker...

Sounds familiar. I'd pin my hopes on this one, with a top-notch free standing horn above it:
https://www.diyaudio.com/community/...-design-the-easy-way-ath4.338806/post-7246372

There's still nothing I wouldn't like about that design. (Not necessarily with a 15", that was just an example. It could be much smaller.)

 

[mabat]

When can we get the midrange responses?

They are already available, it's just that without a LE model (TS parameters of a particular driver), it's probably not very meaningful.
This is a general, constant acceleration model:

 

[Olombo]

Yeah that dipole with horn top design could be a nice match. Would fit perfectly with a synergy horn on top 😜 for low xo to the woofer. Keep going! 🤯

 

[aragorus]

The response of the midrange drivers are dominated by the acoustic bandpass filter anyway

 

[mabat]

Without actual parameters you have no idea about the absolute sensitivity and excursion demands it brings, i.e. how big/small midrange driver can be used and what's the low frequency limit - that's what I'd be interested in.
Fortunately, it's not difficult to do.

 

[mabat]

This is for two Faitals 5FE100, at 1m from the ports, SPL and excursion at 2.83V.

 

[mabat]

Two vs four mid drivers - the HF response:

 

[kipman725]

With the 2 and 4 simulations I'm only seeing differences in the 5-10kHz range. The h results look like numerical artifacts as nothing has changed in this plane and they are at the highest frequencies in the simulation. Its interesting though that below 5kHz I am not seeing differences which suggests it's posible to 'hide' the ports from the HF.

 

[CinnamonRolls]

Do the directivity spikes between 7kHz to ~9kHz happen because those frequencies disappear into the port? If so, could you overlay the port with a material that stopped those frequencies from falling in the hole but allowed lower frequencies to pass freely?

 

[gedlee]

^ I've often thought that this is worth a try. Almost impossible to simulate a thin sheet of say silk (accurately,) but it could easily be done experimentally.

 

[mabat]

The h results look like numerical artifacts as nothing has changed in this plane and they are at the highest frequencies in the simulation.

Why do you think that the ports in the upper/lower wall shouldn't affect horizontal polars?

Here's a similar model again, with and without the ports shown.

Without and with the ports (horizontal polars):

Below the non-normalized data, smoothed with 1/6oct -
Horizontal polars:

Vertical polars:

 

[mabat]

BTW, closing the interior of the horn into a separate subdomain gives noticeably cleaner results -
Without and with the ports again (horizontal polars):

I'll try different port shapes and sizes.

 

[aragorus]

Great work expanding on the script Mabat!
Somewhere on the forum Tom Danley mentions different horn profiles in terms of loading provided to the midrange tabs.
Expansion rate vs distance from the throat vs smooth directivity is the 3 body problem

 

[mabat]

I yet have to find a way how to add the taps into a R-OSSE waveguide... Really curious about that.

 

[mabat]

^ I've often thought that this is worth a try. Almost impossible to simulate a thin sheet of say silk (accurately,) but it could easily be done experimentally.

What about a greater number of smaller holes? I even think I've already seen that somewhere. At what diameter of the individual holes would frictional losses become a real issue?