Any downsides to coaxial design?

[Matias]

Isn't one a consequence of the other?
- If the speaker has bad vertical dispersion, ceiling and floor affects significantly, and therefore must be treated.
- If the speaker (coax) has good vertical dispersion, ceiling and floor affects less, and therefore are not as important to be treated?

 

[Zvu]

Why wouldn't we use coax and treat the room ?

 

[aarons915]

I appreciate the sources! The first one in particular is very interesting and very thorough, had not seen it before. Lots of food for thought.

I found a thesis version of it online, and it's a doozy at 100+ pages. Not sure if you read the AES version and if there are any differences from the thesis, but that study includes both a test in a semi-anechoic chamber and a test made in a more normal room.

From what I could skim, it seems they found adding a ceiling reflection was beneficial in an anechoic chamber, but was pejorative in a room.

Experiment one is in the anechoic chamber, experiment two in the room:

"Experiment one shows that the median preference value for playback with a reflection is generally above the ‘no preference’ origin (5/6 comparisons). Interestingly, the opposite is observed in experiment two, showing that the median preference was generally below no preference, and only above for 12/48 comparisons. One possibility previously mentioned, is that a semi-anechoic chamber is not a natural environment and seldom used for listening to
audio for entertainment. The inclusion of a single reflection in experiment one may therefore, always be perceived as beneficial placing the listener within a more natural environment."

Interestingly, although they noted clear spatial effects when adding a vertical reflection, it seems preference (whether positive or negative) was mostly based on timbral attributes.

I'd like actually read through this one carefully though as there lots of great nuggets!

Toole refers to Olive and Toole (1989a) and Rakerd et al. (2000).

The former study was performed in an anechoic chamber with a simulated vertical reflection (so, a speaker basically playing the vertical reflection on axis), so a crossover dip isn't expected. It's also worth noting that just because the dominant effect was a timbre change doesn't necessarily means it sounded worse, it was just what was noticeable. Indeed based on the paper you referenced, in some cases it seems the timbral shift, at least in an anechoic chamber, is a good thing because it sounds natural?

So I do generally agree with you, to be clear. Personally I believe it is both true that:

* A good 'loud' vertical reflection (ala coaxial) is absolutely better than that of a typical speaker with a compromised vertical reflection.
* A speaker with wide horizontal but limited vertical dispersion is also better than a typical speaker with a compromised vertical reflection.

The second comparison, I supposed, might be somewhat akin to a speaker with bad horizontal directivity sounding better in a room with treated sidewalls. As opposed to a speaker with good horizontal directivity, in which case sidewall treatment appears to be largely a matter of preference.

So my question is still really whether, in a given room, it sounds better to have a coaxial with 'loud' ceiling reflections or to have an attenuated ceiling reflection via either the speaker's inherent directivity or ceiling treatment or high ceilings.

I didn't see the 100 page version, I skimmed the AES study again and it didn't mention the anechoic study at all, only that the tests were performed in a small room in the context of "listening for entertainment". Still, the fact that the listeners in that study were fairly evenly divided on preference with a vertical reflection with a huge null around 2500Hz is decent evidence to me that vertical reflections aren't automatically something we should be absorbing or minimizing. Also, odds are pretty good that a vertical reflection that more closely matches it's direct sound would be more preferred than a reflection that doesn't but it would be nice to see a definitive study.

To answer your last question, I know a lot of us here tend to debate technical details of speakers and have all sorts of rationale for doing so but I personally think it comes down to common sense and a little bit of preference. I want my system to sound like an acoustic set in a smallish space; vocals and most of the sounds coming from instruments are most closely replicated by a coaxial speaker and that is what I've found to sound the most natural to my ears and give me that experience. Some people seem to prefer more direct sound or may prefer music to sound more like a studio and they most likely prefer to absorb reflections.

 

[More Dynamics Please]

Never underestimate the power of the PP factor (Personal Preference).

 

[Beave]

Doppler distortion is seen as part of IMD.
You can find an article about this on Purifi's blog.

Thanks for the link, but this appears to be a discussion of doppler distortion for a single driver, which is not the same phenomenon as what occurs with a coax with a moving waveguide.

Unfortunately this is not correct, the Doppler distortion is described as frequency or phase modulation.

Source: Loudspeaker Nonlinearities – Causes, Parameters, Symptoms - Klippel GmbH

Again that applies to doppler distortion of a single driver. It's not the same thing as the oft-discussed issue with coaxial drivers.

If a coaxial driver is used in a 3-way loudspeaker, Doppler distortion certainly does not play a significant role.
With an appropriate crossover frequency around 300Hz, the excursion of the midrange driver remains well below +-0.5mm, even at high sound pressure levels. Thus, the influence on frequencies with small wavelength (of the coaxial tweeter) is very limited.
In addition, reflections generated by the coaxial midrange cone (e.g. Waveguide mouth reflections) are significantly more delayed in time.

On that I agree.

 

[ctrl]

Thanks for the link, but this appears to be a discussion of doppler distortion for a single driver, which is not the same phenomenon as what occurs with a coax with a moving waveguide.

As soon as the wavelength of the sound radiated by the coaxial tweeter is large enough to be controlled by the waveguide/cone, a Doppler effect should occur, the coaxial woofer frequency modulates the radiated sound of the coaxial tweeter - or not?

 

[Beave]

Yes, the moving coax midwoofer modulates the radiated sound of the tweeter.

But in a single driver doppler effect, the higher frequencies and lower frequencies come from the same driver, which is all moving. As the driver moves in and out relatively slowly to track the lower frequency, it's also moving in and out relatively quickly to track the higher frequency. You have high frequencies coming from a cone that is moving in and out - a doppler effect.

In a coax, it would be like the coax IF the tweeter moved along with the midwoofer. But it doesn't. It moves independently. The sourse of the highs is not moving in and out as the lows are played. But the *waveguide* of the highs is moving in and out as low frequencies are played.

So it's not quite the same phenomenon.

To measure what happens with the coax:

Use DC voltage to put the midwoofer coax at max positive excursion, and measure tweeter frequency response on and off axis.

Use no voltage to midwoofer, so it's at rest, and measure tweeter frequency response on and off axis.

Use DC voltage to put the midwoofer coax at max negative excursion, and measure the tweeter frequency response on and off axis.

 

[ctrl]

Yes, the moving coax midwoofer modulates the radiated sound of the tweeter.
But in a single driver doppler effect, the higher frequencies and lower frequencies come from the same driver, which is all moving.
...
The sourse of the highs is not moving in and out as the lows are played. But the *waveguide* of the highs is moving in and out as low frequencies are played.

The definition for the Doppler effect is not as narrow as you make it out to be here.
That it is not exactly the same is obvious. The question is whether the wavelengths of the coaxial tweeter, which are controlled by the woofer cone, are modulated in their frequency/phase, and if so, that they meet the definition of the Doppler effect.

From post#120 were a definition of the Doppler effect was given:

This effect can also be described by the varying distance between the radiating surface (cone) and a listening point in axis caused by the displacement of the diaphragm generated by a low frequency component.

Use DC voltage to put the midwoofer coax at max negative excursion, and measure the tweeter frequency response on and off axis.

This is yet another separate effect due to the moving cone of the woofer. This would be something like "directivity modulation" and if a driver operates with, for example +-6mm, the effects on the radiation of the tweeter should be significant - I agree with you.

 

[Beave]

The definition for the Doppler effect is not as narrow as you make it out to be here.

Certainly possible.

That it is not exactly the same is obvious.

To you, yes, and to others highly knowledgeable in speakers, but not to the run-of-the-mill audio enthusiasts who post on forums.

The question is whether the wavelengths of the coaxial tweeter, which are controlled by the woofer cone, are modulated in their frequency/phase, and if so, that they meet the definition of the Doppler effect.

From post#120 were a definition of the Doppler effect was given:

From that definition, wouldn't a coax driver have LESS doppler distortion, not more, compared to a single wideband driver? The "varying distance between the radiating surface (in a coax, the tweeter) and a listening point caused by the displacement of the diaphragm (in this case, the midwoofer but NOT the tweeter) generated by a low frequency component."

This is yet another separate effect due to the moving cone of the woofer. This would be something like "directivity modulation" and if a driver operates with, for example +-6mm, the effects on the radiation of the tweeter should be significant - I agree with you.

Right, and it appears to me (although I could be wrong) that in many forums, THIS is what people refer to as the IMD problem with coax drivers.

 

[Head_Unit]

Here is my question about an issue mentioned earlier: has anyone actually measured frequency response of tweeter-at-apex* coaxial, with the woofer all the way out, centered, all the way in? I would love to see such data, never seen such, I'm very curious. You'd think the changing woofer position would cause response changes-but do they really? Inquiring minds having had a few glasses of Malbec want to know.

*Not the autosound type where a tweeter is physically stuck out in front, that's not using the woofer as a horn. Those usually have cancellation problems and also phase problems not easily amenable to crossover design, at least all the dozens and dozens I ever measured did. (Blessings upon MLSSA and LEAP for making that part of life WAY easier)

P.S. I haven't digested all 8 pages yet, apologies if this actually came up.

 

[jhaider]

Here is my question about an issue mentioned earlier: has anyone actually measured frequency response of tweeter-at-apex* coaxial, with the woofer all the way out, centered, all the way in? I would love to see such data, never seen such, I'm very curious. You'd think the changing woofer position would cause response changes-but do they really? Inquiring minds having had a few glasses of Malbec want to know.

@hardisj did on his previous site.

 

[Beave]

^Thanks. I knew I had seen his tests, but I couldn't find them.

Now we need to convince him to add in a bunch of off-axis curves too! (Yeah right)

 

[Zvu]

And do it for 0,5mm/1mm/1,5mm/2mm/2,5mm ... and do the distortion THD and IMD to see what is acceptable excursion which will determine lowest crossover frequency to a woofer.

While we're dreaming

 

[ctrl]

From that definition, wouldn't a coax driver have LESS doppler distortion, not more, compared to a single wideband driver?

I would agree with you there. Is the opposite really being said?

Right, and it appears to me (although I could be wrong) that in many forums, THIS is what people refer to as the IMD problem with coax drivers.

Whether this is so, I cannot say, my interpretation of statements to coaxial systems about IMD, went (apart from the usual causes of IMD) always toward Doppler effect - but also I can be wrong there with my interpretation.

Whether the "directivity modulation" you describe can be subsumed as IMD is another matter - probably not.

And do it for 0,5mm/1mm/1,5mm/2mm/2,5mm ... and do the distortion THD and IMD at those excursions to see what is acceptable excursion which will determine lowest crossover frequency to a woofer.

If only a qualitative assessment of the impact is needed, a simulation can give you the answer.

As an example, consider a 1'' tweeter with a "simple" cone of 10cm diameter (roughly equivalent to a 6'' driver). The coaxial system is built into an infinite baffle.

First, we consider the radiation of the coaxial system at rest and then with only +-3mm excursion (cross-section with the tweeter cone and the woofer cone):

Don't mind the 3mm step around the tweeter, it has to be like that (to have a minimum of radiation control at high frequencies, without it the radiation looks bad - but can not be explained here).

Let's compare the frequency responses from 0°-90° in 10° steps:


To make the comparison easier, we normalize to the axis frequency response:

same as spectrogram:


It is easy to see that even 3mm of excursion leads to a considerable "modulation of directivity".

This should also make it clear that it is not a good idea to design a coaxial system as a full-range - at least if the sound pressure in the bass range could lead to greater excursion of the woofer.

 

[Plcamp]

I would agree with you there. Is the opposite really being said?


Whether this is so, I cannot say, my interpretation of statements to coaxial systems about IMD, went (apart from the usual causes of IMD) always toward Doppler effect - but also I can be wrong there with my interpretation.

Whether the "directivity modulation" you describe can be subsumed as IMD is another matter - probably not.



If only a qualitative assessment of the impact is needed, a simulation can give you the answer.

As an example, consider a 1'' tweeter with a "simple" cone of 10cm diameter (roughly equivalent to a 6'' driver). The coaxial system is built into an infinite baffle.

First, we consider the radiation of the coaxial system at rest and then with only +-3mm excursion (cross-section with the tweeter cone and the woofer cone):
View attachment 117549 View attachment 117550 View attachment 117562
Don't mind the 3mm step around the tweeter, it has to be like that (to have a minimum of radiation control at high frequencies, without it the radiation looks bad - but can not be explained here).

Let's compare the frequency responses from 0°-90° in 10° steps:
View attachment 117564 View attachment 117565 View attachment 117566

To make the comparison easier, we normalize to the axis frequency response:
View attachment 117555 View attachment 117556 View attachment 117568
same as spectrogram:
View attachment 117557 View attachment 117558 View attachment 117569

It is easy to see that even 3mm of excursion leads to a considerable "modulation of directivity".

This should also make it clear that it is not a good idea to design a coaxial system as a full-range - at least if the sound pressure in the bass range could lead to greater excursion of the woofer.

I would suppose some measure of the above must also occur when not using coax...ie in WTW alignments the tweeter’s output is modulated by bass drivers as close as possible to the tweeter, even though not concentric with it?

I also think in my open baffle case, the entire baffle is vibrating...becoming an unwanted modulation “cone” in itself.

 

[sigbergaudio]

I'd expect mayhem for coaxial 2 way at higher volumes

Less than 1% THD (mostly less than 0.5% THD) from 100-20,000hz at 96dB@1m with our 2.5 way coaxial.

 

[Mnyb]

Less than 1% THD (mostly less than 0.5% THD) from 100-20,000hz at 96dB@1m with our 2.5 way coaxial.

You could try to capture some of the IMD/Doppler theories touted here play a test tone clearly in the tweeters range of operation like 4-5kHz.

Then apply something to the woofer(s) like an A 440Hz or 100Hz ? Or find a the worst case a stress test for coaxials

 

[Plcamp]

Would the FFT of a low frequency square wave input to a coax reveal this effect? Record tweeter only then both and subtract one from the other. Anything remaining is a quantified modulation artifact?

 

[Zvu]

@sigbergaudio Here's how Kef R3 looks at 105dB/1m

It would be interesting to see imd multitone testing of Kef R3, Genelec coax and some coaxial 2/2.5 way design.

 

[Zvu]

.........It is easy to see that even 3mm of excursion leads to a considerable "modulation of directivity".

This should also make it clear that it is not a good idea to design a coaxial system as a full-range - at least if the sound pressure in the bass range could lead to greater excursion of the woofer.

That's why i see 2 way coaxial loudspeaker as flawed - but potential is there.