2-way 10'' speaker with swappable waveguides for constant/linear directivity

[Scgorg]

This started as a conventional 2-way with a 10'' woofer and compression driver as tweeter (similar to Genelec S360 or certain JBL speakers), but along the way I stumbled my way into some interesting ideas. As such I thought I would share some design details before the project is finished. Anyone who has read the ATH4 thread over at DIYaudio (https://www.diyaudio.com/community/threads/acoustic-horn-design-the-easy-way-ath4.338806/) will know that well-designed freestanding waveguides offer unparalleled smoothness and directivity. While I don't use ATH4 myself, I am an avid follower of that thread, and do use AKABAK extensively for my designs. The drivers chosen are the BMS 5530ND, and the PHL 3411.

The concept: a loudspeaker in which the waveguide is a freestanding one, such that it can easily be swapped between a constant directivity and linear directivity one. By using something like a hypex plate amplifier, it'd be possible to store 3 different presets:
1. Constant directivity, with flat on-axis response
2. Linearly increasing directivity, with flat on-axis response
3. Constant directivity, with downsloping on-axis response
Furthermore, I've been considering making the cabinet for the woofer adjustable between a typical vented enclosure and a resistance enclosure. Running some simulations indicates that it should be possible, but it doesn't seem quite ideal (it appears the volume required for bass reflex is not ideal for a cardioid and vice versa), so let's shelve that idea for the time being. In any case, this speaker would easily allow one to test the sound of different design paradigms. It's worth noting that this is still a relatively small speaker, and as such the directivity control measures are only really effective in the ~1-20kHz range. Making it larger would enable directivity control lower in frequency.

Enough talk about ideas, let's talk simulations!
Firstly, the compression driver chosen is the BMS 5530ND. Chosen for its great performance at both high and low frequencies. It behaves much better at low frequencies than most 1'' exit compression drivers, and this is a benefit when we want to squeeze as much range as possible out of our compression drivers. Additionally, it has a very small phase plug exit, and as such it is possible to remove the bugscreen in front of the driver and place a throat piece specifically adapted to the horn into the conical exit of the driver. This allows finer control of the horn's expansion (and thus directivity and loading). The exit of the BMS' phase plug is roughly 16mm, that's roughly 0.63'' for you Americans, and means that constant directivity up to 20kHz is achievable.
Note that for the data I am going to present, all normalized graphs will be normalized to 10 degrees off-axis, as there are (very slight) interference effects directly on-axis (in the range of 0.5-1dB or so). The design axis for these designs will therefore be very slightly off-axis.

Constant directivity waveguide (black line in second graph is power response):

Linear directivity waveguide (black line in second graph is power response):

Both waveguides are somewhat narrow, and this is by design. The constant directivity waveguide would work well for a large variety of seating positions, while this is less true of the linear directivity one. Nonetheless I've tried to keep a roughly +-15 degree angle where the sound is pretty much the same, so that the listener doesn't change the sound simply by moving his or her head. The following are simulations of each of these waveguides combined with a 10'' woofer (also simulated in AKABAK), with what I believe are achievable crossovers. Note that the spins are *not* normalized to 10 degrees off-axis like the above graphs, as the ER and SP are referenced to the listening window, which effectively takes care of any aberrations.
Constant directivity waveguide (crossover 1075Hz):

Linear directivity waveguide (crossover 800Hz):

I have already 3D printed the constant directivity waveguide and measured it, while the directivity is as predicted, it unfortunately seems the the BMS 5530ND is very sensitive the the acoustic load it is presented with (in this case the acoustic impedance of the horn). I won't go into detail about it here, but it means that it is very hard to run this waveguide much further down than roughly 1.1kHz on this waveguide. The linear directivity waveguide can more easily be designed to provide good acoustic loading, and it is therefore assumed that an 800Hz crossover should be possible. I am currently 3D printing this waveguide, and eager to see the results.
Finally, just to give an idea of what the waveguides look like (first constant directivity, second linear directivity):

 

[norman bates]

People over on the multi-way area on diyaudio.com would probably like to see this...

 

[jhaider]

What an interesting project! Following for updates.

 

[kimmosto]

This started as a conventional 2-way with a 10'' woofer and compression driver as tweeter (similar to Genelec S360...

Slopes of directivity index, power response and predicted in-room look quite steep. You've probably noticed some changes in VituixCAD's target slopes in Preference rating window. These changes are related to preferred and practical directivity and power slope targets.

This (continuously preliminary) document https://kimmosaunisto.net/Misc/speaker_review_feedback.pdf was originally written for @amirm, @pierre and @hardisj, but it may contain some interesting and relevant items also for speaker designers and members who try to interprete measurements.

For example S360. In the next image LW is normalized to linear so that slope of on-axis is 0 dB/oct. Blue overlay is preferred slope of power response by Harman; -6 dB @100-10000 Hz. Dark orange overlay is slope of preferred in-room target by Harman; -0.5 dB/oct. The last one is quite much referred, but study was tiny with just 11 listeners. Anyway, measured slopes of S360 are steeper so sound balance would not be as flexible for different listening positions and acoustics as Harman's study expects.

Those target slopes are not necessarily very practical anymore with modern wider band power response measurements. In addiiton, preferred slopes were probably studied with classical/acoustical recorded at far field, and some hifi expo muzak by Jennifer Warnes etc. Not with 80s' hard rock, modern metal or whatever having at least some tolerance and challenge. So VituixCAD's default slopes for PIR and SP are higher to give some but still good chances in practice.
This is with the latest VCAD revision 2.0.119.3. PIR slope target is -0.7 dB/oct and SP slope target -1.2 dB/oct which is 0.2 dB/oct higher than maximum of some other old study. Exceeding those values is not a good idea imho in case target is to design something else that just near field monitor.

S360 is not so far anymore, but the slopes are still highish so the speaker is rather "near field monitor" than product for versatile music listening in a good acoustics. One possible cure is shallower wave guide producing lower directivity slope, but it could create S-curve to directivity at midrange. Another cure is to make 3-way with large 18"-21" woofer. Large cone and baffle increase directivity below midrange. 3-way with 21"+10"+1" works also with quite shallow wave guide, depth ~ 1.5". Slope of directivity index is so low that on-axis could have small boost at low frequencies. Listening area having balanced sound is much larger than for example with Neumann KH 150 having even more significant directivity issue than S360 (which is "saved" by larger baffle and cone).

 

[kimmosto]

The drivers chosen are the BMS 5530ND, and the PHL 3411.

One important thing with this concept is compression balance. 10" woofer does not easily keep up with compression driver. Your driver selection looks okay because compression driver is BMS with polyester. I've been using 4538 for years. The same concept with for example Ti cone could produce unlistenable sound quite soon when SPL increases.

 

[Scgorg]

Thank you very much for your comment, Kimmo! I love VituixCAD, and have learned a lot from your other comments around the web.

Slopes of directivity index, power response and predicted in-room look quite steep.

As mentioned in my original post, these are still relatively small speakers (oversized bookshelf category?), and so it is hard to achieve any meaningful directivity control under roughly 1kHz, which is why a resistance enclosure was considered (which I've done with great results in a prior project). I am also now seeing that I posted the wrong images for the "linear directivity" waveguide, it's slightly less beamy than shown here, and has a SP slope of roughly -0.8 to -0.9dB/oct above 800Hz, the PIR slope is likewise somewhat less extreme, at -0.5dB/oct above 800Hz. The problem, as you've accurately identified, is that the speaker is simply small (acoustically speaking) and as such obtaining higher directivity for a more correct slope at LF would require either making the woofer cardioid (by use of a resistance enclosure or extra woofer(s)), or making the speaker much bigger, to accomodate for example an 18'' woofer.

Unfortunately, I am limited by both space and funds in this case (I quit my full-time job to go back to school a couple of months ago, so my income at present is quite low), and so making the speaker larger is simply not an option. Making a resistance enclosure would be possible, but this speaker was originally intended to be ran without subwoofers (low bass reflex tuning + DSP to linearize LF response to ~40Hz), and if I decided to go for a resistance enclosure, that would no longer be an option. You're of course also right that the 10'' woofer isn't going to keep up with the compression driver in terms of output, but that's just another compromise of making a medium-sized speaker.

Here is the simulated spin of the correct linear directivity waveguide, this time with a slightly flattened LW:

 

[Scgorg]

People over on the multi-way area on diyaudio.com would probably like to see this...

Perhaps I'll copy the post over there at some point as well. I generally like to stick to having only one thread (if I make any at all) for my projects.

 

[kimmosto]

Neumann KH 150 having even more significant directivity issue than

Sorry about temporary hijacking, but graph I referred, and few extras

Next one is affordable in-wall design. Target line is 6 dB @200-12000 Hz. Sample is not perfect, but should reveal why this concept is acoustically so brilliant compared to too small speakers with a horn or wave guide producing too much directivity along with edge diffraction. Unfortunately these are mechanically quite annoying.

This is just for a reference - that previous slope targets are possible in practice with shallow wave guide

 

[Scgorg]

Slightly off-topic still, but as an example of how one can make a high-directivity speaker that still has a reasonable slope, I simulated this 15'' woofer in a resistance enclosure with a 1.4'' throat 500mm freestanding waveguide partially covering it:

Exporting the data and combining with the full simulation of the waveguide (not shown), and crossing at 525Hz gives the following:

The slope overlay for PIR is -0.59dB/octave, while the slope overlay for SP is -0.93dB/octave. There is a slight directivity jump at the crossover, this is a combination of the somewhat higher vertical directivity that obviously occurs, and the fact that the vertical directivity of the woofer actually widens a bit as frequency increases. See the contour plots:


The natural drawback is of course that this speaker would struggle with producing bass, if you want a truly loud speaker. Crossing it to a large woofer at 100-150Hz would be a good option. It might be a little challenging to figure out how to do it in practice, but looking at the results I think it should also be possible to place the waveguide such that it covers even more of the woofer without many bad effects, which could lessen the slight directivity error.

 

[Fledermaus]

Story of my life...

 

[kimmosto]

Slightly off-topic still, but as an example of how one can make a high-directivity speaker that still has a reasonable slope, I simulated this 15'' woofer in a resistance enclosure...

Still off-topic: Some of you may remember KS-1804 having 2x18N840 in resistance enclosure

and KS-1807 with 4x SLS-12 in WWMTMWW. That pair might be still located in some local mastering studio.

BUT directivity is just the second best method for producing higher acoustical resolution (though cardioid bass has advantages). "Extra" directivity (above bass) is not needed when acoustics is good enough, and degradation of sound quality - especially to off-axis - is common problem with directivity technologies. Playing with horns and resistance boxes looks kinda old-fashioned to me, and does not support well for example casual listening. It's interesting technical challenge, and helps in bad acoustics if that is unavoidable, but any amount of directivity can't totally eliminate bad acoustics.

 

[Vineeth Kumar V]

@Scgorg: I have recently built a relatively controlled directivity system (just for relatively lower SPL levels based home use and not PA).
Details are below.
The system uses the following drivers at the moment.. The horn + CD combo keeps changing frequently.
(Also, I have a bad, echo chamber of a living room where these speakers are placed.)
HF: SB audience Rosso 65CDN-T(with custom 1 inch adaptor)/BMS 4550 on (ATH generated) EXAR 400 horn ( driver has a series cap)
MF: Faital pro 15PR400 in sealed box
Sub: SB audience NERO SW800 in sealed box (faces backside walls)
Crossover implemented using minidsp flex-8
Some pics

Crossover looks like below (can be simplified further)
Please ignore the gain on the subwoofer branch

There is significant excess group delay increase around 100Hz with above crossover (4.4ms)
The LR4 phase linearization around 100Hz can get rid of it and make it like below. But my minidsp probably wont handle that kind of FIR filter

 

[kimmosto]

KS-1807...
View attachment 408941

This concept has vertical lobing issue so it's primarily for intensive hifi / mastering use. Directivity and timing performance could be like this at least in theory with linear phase FIR gear.

IRL, the project is semi-active with 2-way Hypex IIR plates. That speaker - with all it's faults and imperfections - has been one of the best in our house.

Anyway, my point is that horizontal dispersion can be really wide full range at the same time with significant and quite constant total directivity.

 

[Scgorg]

I have recently built a relatively controlled directivity system (just for relatively lower SPL levels based home use and not PA).

I've seen been keeping up with your thread on DIYaudio, it looks like a nice speaker, both visually and acoustically! My woodworking skill is unfortunately subpar at best, and it shows in my projects

 

[kimmosto]

Two different directivity concepts. What do you think, is the modern one better than classical or vice versa? These have been in the same room for some time, but the results of possible comparisons have been kept secret. I'm just curious if you DIY folks have any biases due to the brand or concept/acoustic design.

 

[D!sco]

I have had the Genelec explained to me, but not the classic, so my bias is clear. Though there are plenty of situations and examples where vintage speakers shine. The cost difference is clear too. Like it or not, that feeds into our preferences. And look at how many drivers there are on the genelec! It looks like it would play much more loud, or low, or clean. But I don't know anything about what it's being compared to while transparency in everything seems to be Genelec's unspoken motto.

 

[Fledermaus]

Of course we have biases, aren't we humans ?
As an example of mine, I know which one I'd rather see in my living space
I guess the Genelec monster, with its wide midrange driver distribution, is rather beamy in that region at least - and that the Aalto should have been taken care of, directivity wise, among other aspects of its development and tuning process.
Please show us the data !

 

[kimmosto]

...transparency in everything seems to be Genelec's unspoken motto.

when it is advantageous for marketing. Half of the owners of Aalto Speakers are Genelec dealers which may explain who gives the orders and who squeals like a pig.

 

[D!sco]

So you're not going to explain?

 

[Scgorg]

Two different directivity concepts. What do you think, is the modern one better than classical or vice versa? These have been in the same room for some time, but the results of possible comparisons have been kept secret. I'm just curious if you DIY folks have any biases due to the brand or concept/acoustic design.

I do like the Genelec concept. It's similar to a concept I simulated in quite a bit of detail some years ago, though in my case with a 1.4'' compression driver in the big waveguide rather than a coaxial driver. Part of that is more that is it is a technically interesting endeavour, not necessarily because it'll preferred sound-wise by most listeners (the concept necessarily has quite high directivity).


On the other hand for the classic speaker horisontally asymmetrical baffles are definitely not my thing. I find them aesthetically annoying, and I find the resulting contour plots to often be a bit of an eyesore. This is also more of a techno-fetish thing than this necessarily being psychoacoustically bad.

So to answer the question, my bias is definitely towards the genelec.