Horn Speakers - Is it me or.......

[Bjorn]

Does "collapsing polar" just mean the directivity gradually increases with frequency?

It means the directivity collapses. When that happens in an import and sensitive area, the speaker has definitive weaknesses and is simply not a great design. What's the most senstive area to us? Basically the area approximately between 500 Hz and 5-6 kHz. Having a constant directivity lower and down to Schroeder is an advantage and same about above 5-6 Khz, but these areas are of less importance. A speaker with only constant direcivity down to 1 kHz area will lead to a big difference in the reflective energy and therefore generally will not sound right. Obviously this can be minimized with treatment, but always best to start with a uniform and broadband directivity.

 

[Tim Link]

It means the directivity collapses. When that happens in an import and sensitive area, the speaker has definitive weaknesses and is simply not a great design. What's the most senstive area to us? Basically the area approximately between 500 Hz and 5-6 kHz. Having a constant directivity lower and down to Schroeder is an advantage and same about above 5-6 Khz, but these areas are of less importance. A speaker with only constant direcivity down to 1 kHz area will lead to a big difference in the reflective energy and therefore generally will not sound right. Obviously this can be minimized with treatment, but always best to start with a uniform and broadband directivity

So it seems a worthy goal might be to devise a midrange horn that can run from 500 Hz to 6 kHz with constant directivity. How small can we make something like that?

 

[olegtern]

Collapse /kəˈlaps/ - suddenly fall down or give way.

It doesn't look like anything is collapsing here.

 

[olegtern]

So it seems a worthy goal might be to devise a midrange horn that can run from 500 Hz to 6 kHz with constant directivity. How small can we make something like that?

e.g. 30x30cm:

 

[Tim Link]

Collapse /kəˈlaps/ - suddenly fall down or give way.

It doesn't look like anything is collapsing here.

View attachment 401348

If I understand Bjorn correctly, it is "collapsing," as he describes it. Otherwise all those colored lines would be very close to parallel. Listening window looks pretty good, but estimated in-room response is sloping noticeably compared to on-axis. Still, I'll bet that sounds pretty darn good.

 

[Tim Link]

e.g. 30x30cm:

View attachment 401349
View attachment 401350

That looks like it should be crossed over by 1k Hz. So maybe 60cm x 60cm and a 2" exit compression driver could do it.

 

[Tim Link]

It means the directivity collapses. When that happens in an import and sensitive area, the speaker has definitive weaknesses and is simply not a great design. What's the most senstive area to us? Basically the area approximately between 500 Hz and 5-6 kHz. Having a constant directivity lower and down to Schroeder is an advantage and same about above 5-6 Khz, but these areas are of less importance. A speaker with only constant direcivity down to 1 kHz area will lead to a big difference in the reflective energy and therefore generally will not sound right. Obviously this can be minimized with treatment, but always best to start with a uniform and broadband directivity.

What does "collapse" mean in terms of directivity? It gets lower, or it gets higher? I assume it means it gets higher as the frequency goes up. Is that right?

 

[olegtern]

That looks like it should be crossed over by 1k Hz. So maybe 60cm x 60cm and a 2" exit compression driver could do it.

RCF 950 WG/drivers combos can be crossed at ~500 even in PA applications, at home SPLs it's no problem at all. The 4” driver works fine with such a low crossover (although somewhat less than ideal in the upper octave). And they are really good sounding waveguides, although they use diffraction to maintain constant directivity at high frequencies.

Of course, it's a modest size for a waveguide and can't be compared to something like Klipsch Jubilee, but it works very well.

Listening window looks pretty good, but estimated in-room response is sloping noticeably compared to on-axis. Still, I'll bet that sounds pretty darn good.

Sounds really good. I'm not saying that constant directivity (if other things being equal) wouldn't improve the sound, or that I wouldn't want to use a huge constant directivity waveguide (but I have a rented apartment and a wife))). But I prefer a more “dry” sound, with less room influence, this may not suit everyone.

 

[miero]

Are there any recording studios that are using constant directivity loudspeakers for monitoring?

And if so, they would also need to optimize sound for common loudspeakers...

More: http://seanolive.blogspot.com/2009/10/audios-circle-of-confusion.html

 

[Bjorn]

Are there any recording studios that are using constant directivity loudspeakers for monitoring?

And if so, they would also need to optimize sound for common loudspeakers...
View attachment 401426
More: http://seanolive.blogspot.com/2009/10/audios-circle-of-confusion.html

Constant directivity in what frequency area?

 

[miero]

ITU-R BS.1116-3 (Methods for the subjective assessment of small impairments in audio systems) requires the following:

7.2.2.2 Directivity index
The directivity index C, measured with one-third octave band noise, over the frequency range 500 Hz to 10 kHz, should be within the limit:
6 dB <= C <= 12 dB
The directivity index should increase smoothly with frequency.

Source: https://www.itu.int/dms_pubrec/itu-r/rec/bs/R-REC-BS.1116-3-201502-I!!PDF-E.pdf

This recommendation is the most strict I have seen, but I'm not sure how many existing loudspeakers fulfills that.

 

[egellings]

I suspect that honk is a severe amplitude peak in the mid frequency range. Is that something that could be EQed out?

If the honk is just caused by a frequency response error, then it could be EQ'ed out. If it's due to added distortion, then I think you would be stuck with it.

 

[Newman]

Constant directivity in what frequency area?

Constant in whatever area will stop you from referring to it as 'collapsing polar'.

 

[Bjorn]

Constant in whatever area will stop you from referring to it as 'collapsing polar'.

There are very few speakers with broadband horisontal constant directivity. And even less when we include the vertical. Thus we don't see them much in studios either.

Most common would be cardioide speakers, but they deviate more in the directivity and are commonly much wider in parts of frequency than what many believe.
Cardioide speakers don't really sound like broadband horn speakers. Big difference IMO.

 

[miero]

Mother, 3-way multi entry horn by additaudio:

-

View this content on Instagram

View this content on Instagram

 

[Scgorg]

Not a fan of a manufacturer putting down extremely thorough community efforts in horn/waveguide design because they don't fall in line with his particular view of what constitutes a good horn (and then not even posting examples of what constitutes "constant directivity" to him). To describe a waveguide which narrows from +-45 degrees at 1kHz to +-40 degrees as 15kHz as "collapsing" while also using that same term for another waveguide that is narrowing much more (by design) seems excessively strict. At that point it would seem prudent to use more precise language.

That aside, I have a constant directivity waveguide in the works right now that is looking very, very promising. It does narrow from +-46 degrees at 900Hz to +-44 degrees at 15kHz though, so perhaps it's actually collapsing?
The below are AKABAK simulations (circular symmetry mode, 28 points per octave, 85kHz mesh resolution), which from my (and other people's) experience has very high predictive power.

 

[Bjorn]

Not a fan of a manufacturer putting down extremely thorough community efforts in horn/waveguide design because they don't fall in line with his particular view of what constitutes a good horn (and then not even posting examples of what constitutes "constant directivity" to him). To describe a waveguide which narrows from +-45 degrees at 1kHz to +-40 degrees as 15kHz as "collapsing" while also using that same term for another waveguide that is narrowing much more (by design) seems excessively strict. At that point it would seem prudent to use more precise language.

That aside, I have a constant directivity waveguide in the works right now that is looking very, very promising. It does narrow from +-46 degrees at 900Hz to +-44 degrees at 15kHz though, so perhaps it's actually collapsing?
The below are AKABAK simulations (circular symmetry mode, 28 points per octave, 85kHz mesh resolution), which from my (and other people's) experience has very high predictive power.
View attachment 401743View attachment 401744View attachment 401742

Collapsing polar is a term that goes decades back and was used commonly for speakers that lost directivity control too high in frequency. It's not a term or definition I have made. It has nothing to do with the narrowing you are mentioning. That is a misunderstanding. I have explained earlier in the thread why a collapsing polar is detrimental, and I have also shown graphs of horns that are truly broadband constant.

An example of what can happens with a horn that loosed it's directivity high in frequency: The JBL 2384 collapses around 2500 Hz vertically, and the result is some serious frequency deviations smack in a sensitive area. That's something that's very important to avoid because of a wrong tonality.

 

[Scgorg]

Collapsing polar is a term that goes decades back and was used commonly for speakers that lost directivity control too high in frequency

I have never seen the term collapsing polar used in that way before, nor can I find examples of it by googling (most results are about various behaviours around the earth's magnetic poles). Perhaps it is a term that was once commonly used, if so, that's a misunderstanding by me and entirely my bad.

I have explained earlier in the thread why a collapsing polar is detrimental, and I have also shown graphs of horns that are truly broadband constant

As for your examples, I assume you mean this for example (found after digging through the thread a bit):

Which I wouldn't quite consider constant (though it's clearly a very good horn!). It's noticeably lacking in smoothness within the window of the beamwidth, and that's with the graph scale already flattering it quite a bit (due to the aspect ratio). It'd be easier to make a good judgemement if the normalized off-axis responses were shown, of course. The JBL 2360, which you mentioned earlier also has these issues to an even larger extent: at 30 degrees off-axis the 500-1kHz range is roughly 5dB down relative to on-axis, while the 5-10kHz range is still equally loud as the on-axis sound! Let me repeat that, a listener sitting 30 degrees off-axis is going to see a 5dB rise from 500Hz to 5kHz (barring room effects), which I'd consider completely unacceptable from a horn design point of view. These figures are taken from JBL's own datasheet of the horn.

I see your point, but talking about the benefits of a horn with constant DI when the individual off-axis frequency responses are messy seems suboptimal. Who cares if the frequency response shape is pretty much the same at 45 degrees off-axis as on-axis if the intermediate angles are messy? That certainly doesn't seem conducive to obtaining good sound at multiple points in the room. At that point even a beamy horn seems preferable, as it will at worst sound dull to listeners off-axis, rather than offensively bright (like the 2360 at 30 degrees off-axis).

JBL datasheet: https://jblpro.com/en/products/2360a-2365a-2366a-flat-front-bi-radial-constant-coverage-horns

 

[Head_Unit]

I get the impression that there's some technical issue with trying to make a CD horn/driver combo that is actually flat.

The high frequency power response of devices falls off at higher frequencies. So if you make the dispersion perfectly constant (whether wide or narrow), the top end will droop by comparison. Hence you could:
- Have a super sensitive midrange and EQ that flat in the crossover (for instance by a high crossover point with a slow rolloff as I noted or
- Possibly have a tweeter with a choked low end into the waveguide, achieving the same result, as per @Tim Link's interesting suggestion above.

That's the theoretical difficulty; the real world is then usually more difficult-like actually making a truly constant dispersion device.