Acoustic Horn Design – The Easy Way (Ath4)

"You would need to insert an acoustic numerical solver into the loop to see also the actual performance. Otherwise the formulas are pretty much useless, I can tell you this right away. Typically, what looks good also performs well, but that's still far from what could be called an optimization. What you get e.g. in the Gen2 horns, are mainly the results of such optimization."

I was going to try to teach myself AKABAK for some simulations, but if I could do it all inherently in Grasshopper I'd do it. I must admit I have no idea what an "acoustic numerical solver" is but if it is geometry or formula based I'm sure I could get it to work native to the program, which would be the absolute end-game for utilizing the parametric design within the program.
 
but if I could do it all inherently in Grasshopper I'd do it.
No, I'm not aware of any readily available solution, unfortunately. Perhaps there are some open-source FEA options that could be used, but that's a completely different task.

BTW, to just play with the formulas, you can do it here already: https://www.desmos.com/calculator/lxpie8bbpa

These formulas are only convenience, in no way they guarantee good results by themselves. It's only a tool.
 
Last edited:
  • Like
Reactions: mjhara
"No, I'm not aware of any readily available solution, unfortunatelly. Perhaps there are some open-source FEA options that could be used, but that's a completely different task."

Grasshopper does have a plugin for acoustical simulation that I'm going to try out - it appears to be more directed at acoustical simulation in rooms rather than for drivers but I might be able to break it for what I am trying to do - we will see.
 
Last edited:
EQing? Not pretty is true. Compared with the usual room interaction and the necessary EQing (I use Acourate FIR for that) the corrections on the driver itself are a magnitude smaller. Not pretty but neither severe as long as there is no further influence.

In other words: if it is fully correctable without issues.....it does not matter.
 
BTW, to just play with the formulas, you can do it here already: https://www.desmos.com/calculator/lxpie8bbpa

These formulas are only convenience, in no way they guarantee good results by themselves. It's only a tool.
Thanks - yes, I was using that as my basis of design for integrating it into a Grasshopper script. I realize Grasshopper uses radians instead of degrees so it took a little fumbling around to get some of the math right but so far I've gotten much of it to have the same results.
 
I can't call it "favourite", my own experience with the big and the more expensive drivers is pretty limited, but if I was to choose one, probably the Faital HF1440 (beware, active EQ needed). The trouble with most large-format drivers is the top octave. Some will say you can't hear it anyway, or that a ragged top end can be even preferable (adding "sparkle") - I don't know, to me it's certainly not pleasant.

Perhaps I don't express it clearly enough but currently I'm really impressed by the performance of the BMS 4554 with a long throat. It allows to go down to ("only") 600 - 650 Hz, but everything higher up is virtually perfect. It kind of lessened my interest in bigger drivers at the moment.
 
Last edited:
  • Like
Reactions: ErnieM and mjhara
The other coaxial CDs (drivers with separate mids capable of large displacement to really push for 300Hz?) on the market are the ones from BMS: 4591 and 4592. The FR on the mids looks really smooth and the intended crossover is higher at 6kHz, which is cool. The distortion is noticably rising on the HF unit, I'm sure that has been discussed before. But these are 2" exit drivers, the FR for the 1.4" and 1.5" versions looks a bit more wiggly.
 
Hi

EQing of this small frequency response deviations IS possible. Before I do a XO I measure a driver; invert the measured frequency response (with some limitations) and use this as a pre-filter. This is common if using large FIR filters with Acourate. After this its ruler flat. Of course doing this with many IIR by hand is not really possible.

Now we have the discussion of the best compromise. You seem to prefer incredible HF and accept vertical lobing if matched with a woofer. I would like to have the best possible driver to driver match and I cannot hear above 13kHz anyway 😱🤷‍♂️
 
EQing of this small frequency response deviations IS possible.
It is not if it's spatially (off-axis angle) dependent. And that's exactly what happens with the HF wavefronts of those coax drivers - just see all the data. For example the BMS I measured, beside all the nasty resonances, was also beaming badly, simply because the exit wavefront is crippled in that kind of device - it's too big and complex. That's not what you can correct with an EQ. If you like it, go for it, I just say what I see and hear 🙂
 
Last edited:
If you do not find a dcx near you I would be OK sending you one from me. Just in the hope you get encouraged and maybe join tweaking it if necessary. 💪🤯🎉

BTW: I checked my 3D scan. Indeed the cross section surface is always expanding. But right after LF insection it has already nearly the surface of a 25mm throat so I cannot mod this in a realistic way.
 
  • Like
Reactions: sheeple