Acoustic Horn Design – The Easy Way (Ath4)

[aragorus]

Should we go back to multicell horns when using 2"?

Well we have learned, but it was of course already anticipated, that wider throats yield lower cut-ins of the diaphragm modes. For a 1" throat all the modes are high enough that we could likely ignore them. For a 2" they will pose a significant limitation on the HF response, most likely causing the need for another driver to handle this frequency region. To me, the take-away is that for home HF one should limit themselves to a 1" driver.

As to the effect that these modes will have on the response, I should make clear that many things can happen - it all depends on the phase of the HOM relative to the normal mode. One phase adds and the directivity will narrow and the opposite phase will substract, creating a possible hole, and anything in-between is possible. Hence, the driver specifics will determine the response once the HOM kick in and will differ from driver to driver. Below that, the HOM will have little to no effect on the drivers allowing only the planar mode to propagate.

These are all useful new knowledge, although neither surprising nor unanticipated, but still Mabat has shown that they are real and can have a significant effect on the results.

One could take these results and highly accurate polar results (not trivial to do at these very high frequencies) to calculate back to the throat and find out what the throat motion was. For 1" drivers this is likely a waste of time, but for 2" ones it would be feasible to do and likely very instructive.

 

[Patrick Bateman]

I found a bug in 4.7.1:

If the name of your config file includes a space, ATH won't produce a functional ABEC file.

If you use a config file name with no spaces, it works fine.

 

[mabat]

Looks bit more wilder this one. Can you test some other waveguide profiles as well? The first results you posted might be smoothest already?

Straight sided conical? how about small step/angle mismatch at the horn throat? some random device from the simulation runs you posted few days ago? Does mouth rollback have any effect?

Surprisingly, at least to me, nothing seems to have much of an effect. Off-axis variations change to some degree depending on a particular profile, but the frequency of the impedance peak is actually constant and only the peak value changes. I don't know whether this has any physical explanation or might be just an error (?).

The first quick&dirty batch for the 0,1 mode - some random mutations of a basic free standing waveguide (1" throat): ATH - Advanced Transition Horns - Gallery (the plots need improvement)

Anyone can easily try themselves. Attached are the required source definitions for the basic axisymmetric modes being discussed. This presents adding one line into an existing Ath script.

 

[mabat]

Has the Rot parameter been mentioned before? I don't remember having seen it described.

It rotates the whole profile (Rot=5 makes an originally 40° conical profile to be 50°), can be handy sometimes.

 

[TNT]

Maybe there is something in a horn/wg being only really good for 3oct after all...?

//

 

[mabat]

I made it at least four here, which I consider as a counterexample and so proven otherwise

 

[TNT]

So the above charts include both mode 0 and 1?

//

 

[mabat]

These are the actual measurements, which include all the modes that the exit wavefronts of the three drivers used are composed of. So, yes.

 

[TNT]

OK! The middle one look good - what is that?

//

 

[mabat]

Celestion CDX1-1425, the waveguide is the same in all three cases. It all has been here before.

- What are the green rectangles supposed to show?

 

[TNT]

OK! The sweet part of the horn.... 3 oct.

//

 

[mabat]

How is this "sweet part" defined? The horn is still the same, so why should it have three different sweet parts? And if it depends on a driver, then it's not a property of the horn, is it. I could have a driver extending the sweet part considerably then. Sorry, but this makes no sense. That the frequency response is not flat, is irrelevant for the quality of the horn.

 

[TNT]

Ultimately, and I'm talking extremes, I think that any FR variation is an anomaly. In any practical way, I agree with you.

//

 

[mabat]

No real acoustic source has a flat power response (i.e. an infinite bandwidth), so that's really a norm rather than anomaly. The rest are just various imperfections of a particular driver (although most of them still correctable), not the horn.

 

[TNT]

For me talking only horn is a bit too clinical. So boring with no music so my comments (3 oct) where for the playing combination. Local (not talking infinite!) variations are due to resonance or suck-out and that makes the impedance change. These things are unwanted. If they can be designed away or not I don't know - guess it should be possible... You seems to be a perfectionist so I'm sure your are glansing them with somewhat of an evil eye..

//

 

[mabat]

I'm only trying to understand what you mean when you say "Maybe there is something in a horn/wg being only really good for 3oct after all...?". But never mind, I don't have to understand everything

 

[TNT]

I did admit that I did not have an explanation for that statement. But as I said, it came from a, for me, trusted and knowledgeable source. I will check with it and see if I can dig out any more details.

//

 

[mabat]

I would like to see that CNC

Its time shall come
I have it disassembled again - need to change the way the whole z-axis moves. Still learning with each iteration...

Detail of the z-axis -

 

[mga2009]

Marcel,

Thanks again for this software. Sadly, I need help one more time.

I modelled a waveguide which works for me, size wise (165mm) and also having a wide directivity (i understand this is good in a small room) but when I try to export the surface to Fusion 360 I get an error.

Other designs (like the one signed with number 2) are properly imported in Fusion 360 without a problem.

 

[mabat]

I have seen this before but unfortunately I still have no clue what's wrong. Usually it helps to change the mesh grid somehow (perhaps generating fewer splines / points per spline, I don't know).

BTW, is this really what you want? The 'marto8' doesn't look like a proper waveguide at all, more like a mediocre direct-radiating dome tweeter. Number 2 is definitely closer to what I would choose.