3-way speaker with double waveguide and cardioid bass

[Scgorg]

I have been working on this project for a long time, and finally finished it. The concept is simple: a 3-way speaker with full-range directivity control.
Design and optimization of waveguides was done with Fusion 360 (modelling) and AKABAK (simulation).
I will not bore you with the details, but the simulation and measured results show excellent agreement, so clearly I did something right. Those wanting for more details can read my existing thread over at hifisentralen: https://www.hifisentralen.no/forume...med-dobbel-waveguide-og-kardioidebass.106808/ (google translate does a pretty good job of translation, so it should be readable). This is a loudspeaker made to be used with subwoofers, similar to something like the sigberg audio manta, so keep that in mind when looking at the bass extension.

Measurements were done with 16ms gating (62.5Hz resolution) and there is thus no data below that, so what you see is interpolated and almost certainly wrong. Furthermore 1/24 octave smoothing was applied, as the mid-high frequencies are a little too wiggly to easily read without it.

Horisontal radiation pattern, from 70Hz:

Vertical radiation pattern, from 70Hz:

Some off-axis angles normalized to on-axis response (1/6th octave smoothing to see the broader spectral trends):

The woofer is crossed at 375Hz, and you can see how strong the rear radiation rejection is. In this case it is strongest at about 160 degrees off-axis, which aligns quite well with the design goal of slightly higher directivity than a "textbook" cardioid.

The speaker looks like this:

Note that this is from a prior measurement session, and does not reflect how the loudspeaker was measured for these final results.
If I were to build this speaker from scratch, there are many things I would change, but within the limitations I had at the start of the project (more than 2 years ago) I am very happy with the end result. Feel free to ask questions.
Tagging @Blockader because I believe he wants to see these results.

 

[ppataki]

Really cool, congratulations!
Do you have some distortion measurements too?

 

[Scgorg]

Really cool, congratulations!
Do you have some distortion measurements too?

Due to the measurement distance (3 meters) there is substantial room influence, even for such a big room (REW cannot gate distortion measurements), because of this the distortion measurements are very corrupted below 200Hz (in fact you can see it in the measurement I will post here at ~220Hz). This is measured at a level of 95dB.

2nd order distortion is a little high from 300-600Hz, however this is perceptually benign, and still falls below the worst case masking threshold. All other orders are extremely low. The spike in distortion at 800Hz is a resonance in the midrange driver, I do not think it deteriorates sound quality meaningfully. Above 900Hz distortion is practically nonexistent.

 

[terryforsythe]

Very nice!

 

[terryforsythe]

FYI, Google Translate's language auto detect did not work properly for me. It worked when I specified Norwegian as the source language, though.

 

[Puddingbuks]

Nice!

How does the low end directivity / cardioid work as there seems to be no speaker in the back?

 

[Scgorg]

Nice!

How does the low end directivity / cardioid works as there seems to be no speaker in the back?

It's a passive cardioid solution with a so-called resistance enclosure. Sound is allowed to leak through the slots on the side of the speaker, and by adjusting the damping material and placement of these slots you can get very strong attenuation towards the rear, similar to having a second driver in anti-phase with a time/phase delay in the back. In my case the optimization of the slots was done in AKABAK, and it matches the real results well (in fact the real results are slightly better than the simulation now!)

Commercial examples of this include Dutch&Dutch 8C, ME Geithain speakers, and the Sigberg Audio Manta.

The drawback is that you get an acoustic short circuit, which drastically lowers output as you move low in frequency, similar to a dipole. For that reason this speaker requires a subwoofer for bass reproduction, even with a 12'' woofer.

 

[dcbingaman]

It's a passive cardioid solution with a so-called resistance enclosure. Sound is allowed to leak through the slots on the side of the speaker, and by adjusting the damping material and placement of these slots you can get very strong attenuation towards the rear, similar to having a second driver in anti-phase with a time/phase delay in the back. In my case the optimization of the slots was done in AKABAK, and it matches the real results well (in fact the real results are slightly better than the simulation now!)

Commercial examples of this include Dutch&Dutch 8C, ME Geithain speakers, and the Sigberg Audio Manta.

The drawback is that you get an acoustic short circuit, which drastically lowers output as you move low in frequency, similar to a dipole. For that reason this speaker requires a subwoofer for bass reproduction, even with a 12'' woofer.

To solve your deficient response below 100 hz but retain directivity, try a slot loaded dipole woofer such used as Nelson Pass’s Slot Loaded Open Baffle (SLOB). Works as a supercardioid below 100 hz because the slot provides 3 dB boost relative to the back wave. Requires equalization and 500W of Class D power.You might see one at AXPONA if you look hard enough !

 

[Scgorg]

I'll add one more thing, a comparison between the fully simulated loudspeaker and the measured example. Due to mesh requirements it has only been simulated up to 5kHz. The part above that is estimted by vituixcad:

There is a remarkable level of agreement between the simulation (left) and the measurements (right). The simulation was exclusively done for directivity purposes, and as such it has a flat frequency response. The biggest difference is really just that the simulated example assumes ~1dB higher SPDI/ERDI above 500Hz, otherwise the match between them is remarkably good. AKABAK truly is powerful software.

 

[Scgorg]

I made a mistake in the prior post. When that simulation was made back in fall 2022, I did not realize vituixCAD uses on-axis as the standard for calculating ERDI and SPDI, rather than listening window which CEA-2034 calls for. In other words the above graph has the simulation's ERDI and SPDI referenced to on-axis, while it is referenced against the listening window in the measured response. Here is the "fixed" graph:

Now the simulation and measurement is even more identical. Very nice!

 

[Ze Frog]

Nice, that's some great work.

 

[D!sco]

Very nice job! The 3D printed elements and how you arrived at their design and construction was really fun to see. I was looking at the original thread in google translate, and I was wondering if you could expand on the enclosure size. You had said that the volume didn't seem to matter much, which mirrors what I'm seeing in other passive cardioid designs. How small could the box be? Would it even have to be a tower? Some of the boxes designed by Kimmo seem to have a depth requirement for damping purposes. Did any of the simulations in AKABAK give insight here?

 

[Scgorg]

Very nice job! The 3D printed elements and how you arrived at their design and construction was really fun to see. I was looking at the original thread in google translate, and I was wondering if you could expand on the enclosure size. You had said that the volume didn't seem to matter much, which mirrors what I'm seeing in other passive cardioid designs. How small could the box be? Would it even have to be a tower? Some of the boxes designed by Kimmo seem to have a depth requirement for damping purposes. Did any of the simulations in AKABAK give insight here?

The box is already basically as small as it can be in the width and depth dimensions (350x230mm). If I made it much smaller I'd run into issues with actually fitting the woofer in there.
The height, as you have correctly identified, is unnecessarily large (1100mm, ~630mm internal height for the woofer, or rougly 33 liters). There are two reasons for this:

1. Even if I made the speaker pretty much the shortest possible height it could be, it would be ~800mm tall, which still puts it into the compact floorstander category, but now necessitates either angling it up towards the listening position (very scary in terms of stability when the speaker is not very deep) or using a separate low-profile loudspeaker stand (like those classic JBL stands). Simply making the speaker taller seemed the best compromise to me at the time, though it does make it look more visually dominating. It also made it even more of a pain to measure - measuring big speakers is absolutely terrible, particularly vertically!

2. the bottom 25-30cm of the woofer enclosure is filled with damping material, if you make the box as small as possible you run into the issue of where to even place the damping material, which is a necessity. In the current speaker there is no damping material on the side walls, for example. Instead there are 45mm open-cell foam plugs mounted flush with the cardioid slits from the outside, which means they build only 20mm into the inside, narrowly avoiding the basket of the woofer.

AKABAK is simple in the respect that you simply define the damping at a given boundary, it does not care about things like "thickness". For this reason you can simulate a seemingly good loudspeaker in AKABAK that is impossible to create in reality, because there exists no damping material that is thin enough with the required damping properties. AKABAK does not care about this, it just spits out graphs based on the numbers you give it, and whether or not those numbers are realistic is up to the designer to figure out.

 

[Blockader]

I have been working on this project for a long time, and finally finished it. The concept is simple: a 3-way speaker with full-range directivity control.
Design and optimization of waveguides was done with Fusion 360 (modelling) and AKABAK (simulation).
I will not bore you with the details, but the simulation and measured results show excellent agreement, so clearly I did something right. Those wanting for more details can read my existing thread over at hifisentralen: https://www.hifisentralen.no/forumet/threads/3-vegs-høgtalar-med-dobbel-waveguide-og-kardioidebass.106808/ (google translate does a pretty good job of translation, so it should be readable). This is a loudspeaker made to be used with subwoofers, similar to something like the sigberg audio manta, so keep that in mind when looking at the bass extension.

Measurements were done with 16ms gating (62.5Hz resolution) and there is thus no data below that, so what you see is interpolated and almost certainly wrong. Furthermore 1/24 octave smoothing was applied, as the mid-high frequencies are a little too wiggly to easily read without it.

Horisontal radiation pattern, from 70Hz:

Vertical radiation pattern, from 70Hz:

Some off-axis angles normalized to on-axis response (1/6th octave smoothing to see the broader spectral trends):

The woofer is crossed at 375Hz, and you can see how strong the rear radiation rejection is. In this case it is strongest at about 160 degrees off-axis, which aligns quite well with the design goal of slightly higher directivity than a "textbook" cardioid.

The speaker looks like this:

Note that this is from a prior measurement session, and does not reflect how the loudspeaker was measured for these final results.
If I were to build this speaker from scratch, there are many things I would change, but within the limitations I had at the start of the project (more than 2 years ago) I am very happy with the end result. Feel free to ask questions.
Tagging @Blockader because I believe he wants to see these results.

Thank you for tagging me. It's great to see the speakers performing as well in reality as they did in simulations.

What are your plans moving forward? Are you considering further modifications to the design, perhaps incorporating low bass drivers, or are you thinking of moving on with a completely new design project?

 

[Scgorg]

Thank you for tagging me. It's great to see the speakers performing as well in reality as they did in simulations.

What are your plans moving forward? Are you considering further modifications to the design, perhaps incorporating low bass drivers, or are you thinking of moving on with a completely new design project?

I am done with this project for now. I might revisit it in the future and make some changes for the better (I already have clear ideas on how to approach this), but for now I am happy, as any gains will be quite small.

My current main focus is a 2-way medium-sized speaker (10'' woofer and 12'' compression driver waveguide), In the same vein as the Genelec S360, Geddes' speakers, or the JBL 4349. Simulations indicate I should have a very good performer on my hands. The main barrier at the time of writing is cost. Building a bunch of speakers is not a cheap endeavor

 

[Scgorg]

MMM across a ~60cm width around the main listening position of the speaker in my room. No room correction/EQ applied. 1/12 octave smoothing. ~2m distance.

 

[AudioJester]

Very impressive build! Congrats

 

[neRok]

MMM across a ~60cm width around the main listening position of the speaker in my room. No room correction/EQ applied. 1/12 octave smoothing. ~2m distance.

Is that just the speaker, or speaker and sub?

 

[Blockader]

Is that just the speaker, or speaker and sub

probably just speakers with room gain. sealed speakers benefit more from room gain.

 

[Scgorg]

Is that just the speaker, or speaker and sub?

Speaker only. Almost flat on-axis and almost flat ERDI makes for a relatively bright sounding speaker, especially at the relatively short distances I listen at, so I've pulled down the 102 and 135Hz peaks, added 2 low-shelf filters (one for bass, one for lower mids) and added one high-shelf filter of -1.5dB to roll off the high end a bit. It sounds good!

probably just speakers with room gain. sealed speakers benefit more from room gain.

It's not a sealed speaker.