Acoustic Horn Design – The Easy Way (Ath4)

You are not optimizing the shape for the situation. That is a very poor example.
That was to illustrate my point that you really have to start curving the profile to improve the results, with which you seemed to not agree. Then when I do optimize, for any situation, you don't like it...

The 19" optimized (a bit) -
 

Attachments

  • 19inch_optimized.png
    19inch_optimized.png
    20.2 KB · Views: 313
  • 19inch_optimized-profile.png
    19inch_optimized-profile.png
    7.7 KB · Views: 308
My point is that for the time being we mainly look at simulations. The response of real horns with real drivers can easily differ from the simulated plots, as the horn-driver interface is only approximately simulated using a hypothetical driver.
I would accept that objection in a situation where there were no measurements made yet. After several different practical examples that all confirmed a very good match I consider this to be a closed issue. Some waveguides were even measured with several different drivers - in the case of 1" throat maybe up to 15 kHz virtually no effect. You haven't noticed?

And please stop using the term "ATH4 waveguide" as if it was some type of horn. It is not. It's a tool implementing the OS-SE formula, allowing for a huge variety of very different profiles - you just can't call them all ATH4.
 
Last edited:
The 19" optimized a little further. Is this good enough? Probably yes.

attachment.php


A slight touch correcting the widening centered around 6 kHz and making the highest octave a bit more even. Is is worth it? I don't know but I would take this one.
(It has also a somewhat better loading ;) )

attachment.php
 

Attachments

  • 20inch_opt2.png
    20inch_opt2.png
    37.8 KB · Views: 754
  • 20inch_opt3.png
    20inch_opt3.png
    38.1 KB · Views: 482
Last edited:
I would accept that objection in a situation where there were no measurements made yet. After several different practical examples that all confirmed a very good match I consider this to be a closed issue. Some waveguides were even measured with several different drivers - in the case of 1" throat maybe up to 15 kHz virtually no effect. You haven't noticed?

And please stop using the term "ATH4 waveguide" as if it was some type of horn. It is not. It's a tool implementing the OS-SE formula, allowing for a huge variety of very different profiles - you just can't call them all ATH4.


We don't disagree and thank you for the sims of the OS waveguide that illustrate my point.

Of course, the horn is usually decisive, but small differences (sometimes larger than 0.5dB) between different drivers in an equivalent horn are reality, due to differences in the construction of the driver.
Such differences are visible in the "six packs" of the B&C DE502, DE360 and Sica CD60 respectively, which were posted earlier.

I used "similar ATH4 waveguide" to make the comparison with a specific OS waveguide.
A designation of a specific waveguide generated using ATH4 :)
 
Of course, the horn is usually decisive, but small differences (sometimes larger than 0.5dB) between different drivers in an equivalent horn are reality, due to differences in the construction of the driver.
Such differences are visible in the "six packs" of the B&C DE502, DE360 and Sica CD60 respectively, which were posted earlier.
Could you point out the differences in the attached picture? I'm really failing to see them. Maybe 15 kHz is too high, so lets say it is 12 kHz - that's still pretty good, IMHO.

- Or maybe there's a misconception here as I'm not sure what you meant by these differences. To clear that up - the simulations predict radiation patterns, i.e. what is on the attached picture. Absolute value of radiated power as a function of frequency is really not relevant for the analysis. Different drivers can differ a lot in that respect but the radiation pattern stays virtually the same, as you can see. That's what we are interested in here.
 

Attachments

  • 3wg.png
    3wg.png
    444 KB · Views: 97
The reason why the driver doesn't really matter for the radiation pattern is of course because it is too small to have an effect - the exit wavefronts are too simple to drive the waveguide in a more complicated manner. It will have an effect as the wavelengths get comparable with the throat diameter as the higher order modes come into play but below that it's basically all the same.
 
That was to illustrate my point that you really have to start curving the profile to improve the results, with which you seemed to not agree. Then when I do optimize, for any situation, you don't like it...

The 19" optimized (a bit) -

You seem to misunderstand me. I have always known that a more gradual transition to the mouth flare would reduce axial diffraction issues - it's clearly obvious. Your software has proven that to be the case, so I am in complete agreement on this point.

But you seem to be saying that a rising DI is an unavoidable consequence of doing the above and I don't see why that would be nor do I believe that you have proven this point. I also think that people discount the importance of a higher and flatter DI, which, if it is an inevitable tradeoff to mouth diffraction reduction, should be compromised for the flatter axial response. This is my point of contention NOT that smoother mouth flare curves are not effective.
 
But you seem to be saying that a rising DI is an unavoidable consequence of doing the above and I don't see why that would be nor do I believe that you have proven this point.
This is indeed what I say because this is what I find out every time I try - I have yet to see a device you speak of. I don't know if or how that could be proved but definitely it would be disproved by finding a counter-example. Do you know of any? If you drew a profile I could easily test it. So what would be your best bet? I'm out of luck.

I also think that people discount the importance of a higher and flatter DI, ...
People are asking for an evidence that a flat DI is really necessary. And frankly, when talking about a typical home listening situation, not PA audience coverage (where is seems obvious for other reasons), there seems to be no such evidence, other than "it sounds good to me".

"Reasonably flat" and smooth DI is what sounds perfectly fine to me as we typically don't want the power/room response to be flat anyway (bmc0 being the curious exception :) )
 
Last edited:
But you haven't shown the DI for the examples that you show, hence I am not sure that what you say is the case. Does the DI go up as you flatten the axial response in each of those cases?

As far as the DI goes I am just not up on what is available as regards "flatness", but smoothness everyone agrees to (although few abide by.) "Reasonably flat" is not well defined and we seem to be arguing degrees here. A few dB rise of DI is not going to be a big issue - is that "Reasonably flat"?

And yes we do want the power/room response to fall slightly, but this can be achieved independent of DI.
 
But you haven't shown the DI for the examples that you show, ...
Yeah, I'm aware of that - this is still something I want to implement, I just can't find the time I could sit for a while and finish it... I really want to do this and try some more systematic approach.

Does the DI go up as you flatten the axial response in each of those cases?
I'd say it's quite obvious from the measured data, but I think it's really the case in general - as the polars fall down (as they do), the DI will go up, although very gradually (if not in a straight line).
 
Last edited:
I'm not sure ARTA can do it directly but if you export the polars to text files you could import them e.g. to VituixCad - there you can normalize them to any angle you want (I think). The mic response is irrelevant - both the frequency response of the driver and mic response are put out of picture by normalization - what remains is really the "response" of the waveguide alone.

It looks pretty good, BTW! I don't think it is even necessary to compare it with the simulation. And it's obvious the DI will rise slightly.

With some DSP you could make virtually a flat response out of this. What a marvelous loudspeaker this would (will?) be.
 
Last edited:
I've just asked in the ARTA thread how to do this. Thanks for the tip about VituixCad. I'll have a go.

I'm nervous about chiming in about the DI question - but all the truly great domestic systems I've heard (so far!) have been with 'slow-ish opening' horns. Tractrix or lecleach etc. I'm not sure its just because of the narrowing sweetspot. There seems to be something about how the horn loads the air that I recognise from live music that seems absent from CD waveguides. But perhaps it depends on what you want your playback to do. Personally, I'm after the most dramatic sense of performance and thrill that can be found. This seems lacking on some constant Directivity WGs. So, I'm very interested to see how these horns sound - soon!