• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Equidistant baffle for coaxial driver

edc

Member
Joined
Jan 1, 2025
Messages
12
Likes
25
8.png


Hi all, I'm new on the forum.

I have a pair of JBL coaxial midrange drivers (exact mids that are used in JBL Control 30), which I paired with compression tweeters (JBL 2413H). I'm using this configuration as mid/high only with high pass filter at 400Hz and active crossover, therefore not expecting any bass from this driver.

jbl-coax1.jpgjbl-coax2.jpgjbl-coax3.jpg

What I'm trying to do is attach the drivers to a cube shaped enclosure which I used in a previous project. Enclosure size is 33cm/13in cubed.

I'm playing around with the concept of an equidistant baffle. By equidistant, I'm referring to all edges that reach the sides of the enclosure to be equal from the center of the driver.

For example, all the lines below have equal length when unwrapped/stretched.
11.png
Since the driver is round, and the enclosure is square, the outside edges are folded in, but the inside edge remains a circle. The third item to the right is the baffle unfolded, which visualizes the equal distance. A sphere enclosure is also equidistant, but what I have is square, and have to adapt to it.
10.png

The yellow part is a bridge waveguide between the coaxial driver and the equidistant baffle. The cabinet is made from 18mm MDF, and the baffle will be 3D printed using PLA.
7.png

Question is: Is this a good idea, or I'm only wasting my filament and should stick with a regular waveguide instead? Is there any research done on these kinds of waveguides?

Thanks!
 

Attachments

  • 4.png
    4.png
    145.4 KB · Views: 86
  • 6.png
    6.png
    206 KB · Views: 72
Hi, typically one would make simple flat baffle, or, if you want to improve directivity than flat baffle provides make smooth curce and roundover to effectively remove edges and to approximate a sphere to reduce diffraction effects toward listening window. But since filament doesn't cost that much it will be nice opportunity to learn at least so go ahead :) Compare to flat baffle measurements to see a difference.

I predict the response is quite similar to a flat baffle, just with different diffraction related interference fingerprint making response vary to all directions bit differently than with a flat baffle. Perhaps I'm wrong as I'm just imagining how it turns out.
 


Eyeballing that shape, looks to me like you have a pronounced diffractive ridge in between the corners. That would not be my first choice.

I'd prefer to use round-overs with as large a radius as is practical.

If limited to a flat baffle with little or no round-over (which obviously you are not), I would off-center the driver so that it's a different distance from each edge. Imo it's desirable to have DIFFERENT lengths to the cabinet edges to smear the edge reflection in time, rather than having it be a coherent reflection (at least on-axis) which would maximize its effects.
 
Last edited:
Imo it's desirable to have DIFFERENT lengths to the cabinet edges to smear the edge reflection in time
This makes perfect sense.

Thanks for reminding me that it should actually be non-equal distances to the edges of the cabinet. Similarly to how a driver is mounted on an IEC baffle with an offset from the center. My mind always leans towards symmetry.

I'd prefer to use round-overs with as large a radius as is practical.
By round-overs are you referring to something like this?
12.png13.png

it will be nice opportunity to learn at least so go ahead
I've started printing the bridge waveguide, and done some initial measurements.

This is the sealed* (with lots of screw holes) enclosure I'm using. I've made an adapter ring to fit the coaxial inside the hole from a previous project. The raised front edges are a terrible idea, I know, but they were made to hold a grill.
18.jpg

Here is the bridge waveguide mounted on. The outer edge is protruding a few mm outside the front edges of the enclosure.
19.jpg

This is the grill mounted together with the bridge waveguide.
20.jpg

All measurements are done in room at 50cm distance with 5ms gate, and no smoothing. They are all on-axis.
Midrange starts at 200Hz, tweeter starts at 1kHz.
14.png15.png16.png

As you can see, there are major differences, especially on the tweeter side. The grill also makes things worse, as expected.
The midrange has a peak at 8.1kHz regardless of configuration. But that won't be a problem, since it will be crossed over long before that.

Here's a comparison with waveguide on and off.
21.png22.png

I've also done a repeat measurement after taking the waveguide off to see if there are major changes. Basically identical to initial measurements.
17.png

And last one for distortion.
23.png
 
Hi, yeah there is no off-axis data included in your post so it's hard to say anything about directivity, but a seen in the on-axis data it just shows diffraction related interference varies on all of these examples, and not any of them is particularly better than the other, just different.

The render you have for Duke quote is the way to go, "a waveguide baffle". A new waveguide covering the old baffle could be axisymmetric and round as in your rendering, or just something that blends with the square shape of the box, which ever is better to your eye and ease of manufacturing / attaching. What you are mainly interested is just the horizontal plane for normal hifi listening, but if it's an instrumentation amp you could have it round to have a little bit more uniform response no matter at which plane the listener is at.

The original baffle of the speaker dominates the performance in a way that nothing that fits behind the original grill doesn't really make the response any better, just different. But, removing the grille and turning the whole face of the box into proper waveguide with roundover removing the "edge" altogether would make a difference. Or should make a difference, the waveguide profile can vary quite a lot from bad to very good. The cone and tweeter assembly limits performance here, because they have shape you cannot optimize and set the profile basically, so you'll never reach textbook ideal response, which is response that shows no signs of secondary sounds toward listening window while having desired directivity. Bettering the original performance isn't that hard though, basically any continuous shape from tweeter to past the original baffle would make it technically better so ideal isn't achievable but not even needed, it'll sound fine with almost any nice continuous profile I predict. How much things get better perceptually is another thing, perhaps a lot, perhaps not too much, depending on your listening skill which is ability to notice a difference and ability to AB test perception to some particular feature of the playback system and/or auditory system. Also, prepare to optimize crossover for the waveguided version, the old one isn't likely ideal anymore, or perhaps it never was so a good task to do nevertheless.

You can optimize the waveguide profile with BEM simulation for example. See ATH thread in diyaudio. After measuring the new system, use VituixCAD to desing new crossover for it.
 
Last edited:
Question is: Is this a good idea, or I'm only wasting my filament and should stick with a regular waveguide instead?
I do not think it is a particularly good one.
Probably the worst shape of a baffle you can have is a circular one where all distances to the edges are equal.
But by folding the surface you will not get a "quasi"-circle anyways as the diffraction appears when and where the surface is curved (in the corresponding dimensional scale) . More curvature produces more diffraction. So "distance" is not the relevant quantity here.
My recommendation: Try to get as close as possible to a torus shape instead.
 
Probably the worst shape of a baffle you can have is a circular one where all distances to the edges are equal.
I'll expand a bit: circular shape is worst only when the baffle is significantly bigger (in relation to wavelength) than the transducer, in which case diffraction backwave bandwidth increases and delay increases, both of which most probably make the resulting diffraction related interference more audible. Standard round driver naked without any baffle extending it's rim is almost as good diffraction issue wise as any roundover because diffracting bandwidth is so narrow the frequency response is pretty much the same to any direction without undulation. Main diffraction hump is there though, which narrows response and is easily utilized for good, but on some systems this might not be good thing so adapt. This doesn't make frequency response vary (undulate) per direction, so there is not particularly distracting quality to it, just what directivity makes.

You could think minimal baffle, like a naked driver, has roundover as well, just a very tiny one, as big as fits to the edge of the driver ;) The bigger the structure is compared to the transducer on it, the bigger the roundover needs to be, because problematic wavelengths gets longer with the physical size of the baffle so the roundover needs to be bigger, basicaly as big as fits. By fitting I mean, if there is 10cm of baffle around the transducer, then biggest roundover that fits could be ~10cm in radius and would minimize diffraction effects. If the area around transducer is 1mm, then biggest roundover that fits to that space is 1mm radius, which is as good as no roundover at all, because 1mm is very tiny compared to size of the transducer. Everything depends on sound wavelength vs. size and shape of physical structure, they relate, wavelength is physical size of sound so this stuff is relatively easy to simulate in head, everything is relative so there is no need for math or numbers at all, simple visualization in mind is fine and easy.

In addition, there is likely bandwidth on which these issues are more audible than on some other, due to how auditory system works, which I'm not taking into account here because I do not know what this would be. Varying treble is very very easy to perceive when moving around in the listening window, so I speculate thats most audible effect and form of diffraction. So, it makes most sense just to minimize diffraction related interference on listening window and especially on the treble = as big roundover as fits, or at least as big as one can manufacture, doesn't have to be 100%, just big enough. Simplified rule of thumbo.

My recommendation: Try to get as close as possible to a torus shape instead.
Yeah torus might be as fine as any here. Diffraction backwave emits anywhere where curvature of the surface changes abruptly. If the coaxial woofer cone has conical shape there is no curvature, and if torus starts immediately after the curvature changes rapidly to that of the torus and diffraction backwave emits at the junction, so the flat portion needs to ramp up curvature smoothly to avoid diffraction here. Also, the curvature should ideally transition smoothly to the box side walls otherwise some diffraction happens there.

It might seem quite complicated stuff, but it is all based on very simple "rules" what diffraction is and how it happens. Only complication is just the shapes and locations of things, which can be quite easily simplified to basics to visualize most of the effect in mind. There is no magic in this just how sound interacts with physical objects.

And to recap I agree with you, any "smooth" shape like torus, would beat the original baffle in this regard because change in curvature is much less than with the jerks on the original baffle.
 
Last edited:
there is no off-axis data included in your post so it's hard to say anything about directivity
You're right. I wanted to do a quick measurement to check for changes before doing a spin measurement, as that takes a long time. By doing a 90deg spin, I found an interesting behavior of the tweeter. It looks bad on-axis, but once it reaches 20deg it's relatively smooth.

"distance" is not the relevant quantity here
That makes sense now. I don't know why I imagined that the sound would follow the curvature, but instead every curve adds to diffraction.

This simulator spelled it out for me: https://falstad.com/ripple/
30.png

This is the spin rig I've built a few months ago. It's not very pretty, but I tested speakers weighing 25kg/55lb on it, and it can handle more. I printed a 360deg dial with 10deg increments.
20.jpg21.jpg22.jpg

I've only done a 90deg spin for each configuration, 4 spins in total.

By waveguide, I'm referring to the bridge waveguide shown in post #4.

All measurement were done at 1m distance instead of 50cm with same output as in post #4.

Midrange waveguide off:
mid-wg-off.png

Midrange waveguide on:
mid-wg-on.png

For some reason, adding the waveguide makes things worse for the midrange around 2.5kHz.

Tweeter waveguide off:
hi-wg-off.png

Tweeter waveguide on:
hi-wg-on.png

Adding the waveguide improves the tweeter directivity once past 20deg. This tells me that this speaker should not be listened on-axis, instead towed out by 20deg.

For the crossover, it seems to be best around 3.0-3.2kHz.
mid+hi-x3.0k-wg-on.pngmid+hi-x3.2k-wg-on.pngmid+hi-x3.5k-wg-on.png

VituixCAD project (Google Drive)
 
Nice rotating jig!

Yeah diffraction related secondary sound source exhibition :D As you see response is different to any direction = only one listening axis can be adjusted for good direct sound balance = no possibility to change toe-in without changing system power response so in-room sound = cannot optimize spatial aspects of sound (effect of room) because toe-in is locked. Key is to fix the diffraction, which you likely cannot completely do with coaxial system. non-coaxial has it's own issues but woofer being tweeters immediate acoustic environment is solved so the structure can be optimized for both separately.

ps.
I'm on ranting mood today so few more paragraphs about the subject in general :D If you listen further away such response isn't too bad, because room early reflections sound dominates and they are all different and somewhat average out. If you listen at close distance, where direct sound dominates over early reflections, there is only one good listening axis. This feature makes sound of the system kinda uneasy, ever changing with slight head movement, a nervous feeling in a way, attention is taken away from the music. So, main advantage of good problem free coverage is that the direct sound is the same no matter toe-in, so the toe-in and positioning in general can be solely optimized for spatial aspects of sound, and it doesn't change much with movement making it relaxing and calming kinda sound and attention stays in the music.

You can use VituixCAD diffraction tool to get response of simplified ideal baffle with ideal flat pistonic transducer on it to compare with your results. Here you'd see what is achievable in ideal situation and with what roundover, and how far any particular real speaker at hand is from that and why. If your response varies db or few in reality thats really good response for a finished speaker in general. Playing with the diffraction tool demonstrates that even if the transducers were ideal but diffraction bad, the response can vary +-6db to any direction, which doesn't sound very good and really points out that speaker that measures really good is very good overall design structure included. So, main task for anyone thinking speaker design is to try and achieve the "ideal" technical performance with ideal drives and structures, can you do it? if not why, what would it take? and what compromises to take, does it matter in actual use that the response varies per observation angle? how to decide what compromises are more important than others? reality is always "worse" than simplified ideal system, so if the ideal system doesn't meet target performance surely real world example doesn't either.

Which is by the way simplest possible example to show that system design is more important than what drivers are used for example. Most hobbyists on forums tend to start from "best drivers" and then slap them on a box only to get technically worse performance than better structure with any set of drivers so that the sound isn't technically better, just different.

Structure of system has huge effect on measured performance as you've noticed and demonstrate here! So, in reality big part of speaker design is to make the structure good, remove secondary sounds, sounds that are not music at input of the system but from the structure, like resonances, midrange leak from port, and diffraction to name a few. If you weed these out, you are guaranteed to measure good performance from individual drivers which you can now equalize if necessary because it's spatially more uniform to all (important) directions, which enable you to have good crossover and end result. There is nothing one can do for bad measuring structure than rebuild it, doesn't get better by swapping parts. Perhaps perceptual difference between "good" and "bad" doesn't matter in some particular context too much, so what ever :)
 
Last edited:
  • Like
Reactions: edc
Hi all, I'm new on the forum.
Welcome!

Question is: Is this a good idea, or I'm only wasting my filament and should stick with a regular waveguide instead? Is there any research done on these kinds of waveguides?
I had no idea what the answer is but it looks like others here do so now I learned something from them and therefore: good question!
 
By round-overs are you referring to something like this?
View attachment 421182View attachment 421183

I would call that a "waveguide" or a maybe "shallow horn".

For a fixed diameter waveguide there is a tradeoff relationship between how low the waveguide maintains good pattern control, and how smooth the frequency response is. The more gradual the round-over of the "lip" portion around the perimeter the smoother the response, but the higher in frequency the waveguide loses pattern control because the diameter of the actual "waveguide" portion is correspondingly reduced.

You can certainly do that if you want to, but it's not going to maintain good pattern control down to 400 Hz (which may not matter; I don't know what you design goals are).

What I had in mind with my suggestion for maximizing the roundover radius was something more like this morphed into a square:

 
I'm not sure if any shape or size of baffle can flatten the on-axis response of the tweeter relative to off-axis, 20deg onwards. If I'm wrong, please tell me so.

The reason I'm saying that is because of the way the tweeter is built, using an aluminum faceplate that's glued to the diaphragm.
33.jpg

On-axis, the tweeter beams through its horn opening, and the faceplate is possibly having a direct effect.
34.jpg
0 to 10deg
40.png

Once the angle reaches 20-30deg the tweeter plate is out of sight and response starts to flatten out.
35.jpg
20 to 90deg
41.png

0 vs 30deg
42.png

Next up, I'm going to build a large squircle shaped waveguide to see if it improves the midrange response around 2.5kHz. Will upload a full 360 spin once done.
30.png

I don't know what you design goals are
My design goals are to make this driver configuration perform as best as possible by designing a waveguide suited to it. This is a one-off project for personal home listening at low-medium volume. I know these drivers are not easily sourced, and if I damage one of them it's probably game over and should look for something else.

The reason I like this midrange, is because it sounds "fast", it's agile, with fast decay, and the tweeter is easy to listen to, overall a noticeable improvement over my previous full-range drivers (Visaton B200) which I thought sounded "fast", but now I know better. Never listened to electrostatic speakers though, so I'm sure there are better ones out there.
 
My design goals are to make this driver configuration perform as best as possible by designing a waveguide suited to it. This is a one-off project for personal home listening at low-medium volume. I know these drivers are not easily sourced, and if I damage one of them it's probably game over and should look for something else.
Fascinating thread, welcome to the forum.

Waveguide is definitely the way to go... if you haven't dived into the ATH / waveguide threads over at DIYaudio you will find a lot of useful info along these lines.

Since you apparently have a 3D printer and good CAD skills, (honestly, I think the equidistant design was really cool even though it did not solve the right problem...) why not 3D print the whole enclosure with smooth shapes and stop worrying about diffraction entirely? If you print in a few parts and leave a hollow space within the wall, you can fill it with plaster or concrete and come away with equal or better density vs. MDF.
 
Waveguide complete, unintentionally matching the ASR panther color.

cx (5).jpg


I've built the waveguide from a board of high density extruded foam, which incidentally is pink. It's mainly used for home insulation and has high compression resistance.

Carving this material on the CNC leaves a nice texture that reflects light similarly to pink felt. It also works well with sanding paper and doesn't shift color too much.

cx (1).jpgcx (2).jpg

The sheet's dimensions are 1250 x 600 x 50 mm, which allows me to make 2 waveguides. The resulting waveguide has a dimension of 560 x 560 x 48 mm / 22 x 22 x 1.9 inches.

I've made a simple mounting system for the waveguide that uses wood screws and neodymium coin magnets. I've made it like this so that I can adjust the distance of the waveguide from the cabinet, and also for removing the waveguide when needed.

cx (8).jpgcx (9).jpg

The waveguide sits flush, or can be pushed out by unscrewing the screws on the cabinet. I can also rotate the waveguide by 45 degrees.

cx (4).jpgcx (6).jpgcx (3).jpgcx (7).jpg


Now for the measurements:

All measurements done on spin rig, at 1m distance, 5ms gate, no smoothing. By "sq" I'm referring to the squircle waveguide.

I'm not sure if any shape or size of baffle can flatten the on-axis response of the tweeter relative to off-axis, 20deg onwards. If I'm wrong, please tell me so.
My man, you are wrong again! By making a proper waveguide, you can flatten the on-axis response of the tweeter relative to off-axis. /s

Tweeter:
Tweeter response on-axis with and without sq waveguide:
hi-sq-on-vs-off-oa.png

On a serious note, I never imagined that a larger waveguide would actually improve the on-axis response of the coaxial + horn tweeter, but it definitely does.

Interestingly, the response of the tweeter is only affected on-axis, and once it reaches 20deg the waveguide has no major effect on response.

Tweeter response off-axis 20deg with and without sq-waveguide:
hi-sq-on-vs-off-20deg.png

Rotating the sq-waveguide by 45deg doesn't affect the tweeter response in a noticeable way.
hi-sq-on-r45deg.png

Pushing the sq-waveguide out by 1cm has a slight effect on the tweeter response.
hi-sq-on-1cm.png

Midrange:
Adding the sq-waveguide affects the on-axis midrange response considerably.
mid-sq-on-vs-off-oa.png

Rotating the sq-waveguide by 45deg smooths out the midrange response ranging 800Hz-2kHz, but nothing major.
mid-sq-on-r45deg.png

Pushing the sq-waveguide out by 1cm has similar effect as rotating by 45deg.
mid-sq-on-1cm.png

Spins:
Tweeter with and without sq-waveguide:
hi-sq-on-6p.pnghi-sq-off-6p.png
Nice!
The on-axis is better now, but the listening window response is similar to the one without the waveguide. Probably that's why the listening tests I've done didn't sound very different.

Midrange with and without sq-waveguide:
mid-sq-on-6p.pngmid-sqoff-6p.png
Not what I expected, but also nice. The issue at 2.5kHz is worse now, but the directivity is improved at 1.5kHz, and I can take advantage of this by moving the crossover point under 2kHz.

Distortion seems to allow for a lower crossover point, ranging 1.5-2.0kHz:
mid-dist.pnghi-dist.png

And lastly, 360deg spin with and without the sq-waveguide:
sq-on-6p.pngsq-off-6p.png

It seems that my efforts were not wasted, and with more crossover optimizations and few PEQ filters will give me a good performing system.
 
I've made more progress with the crossover and equalization part of the speakers.

I'm using a DSP amp from WONDOM, the JAB4 model that uses 2x TPA3116 IC's, and ADAU1701 DSP chip to power the speakers. This is a low-power board, but It's enough for these drivers since they are relatively efficient. One disadvantage of these types of boards is that you need a companion DSP programmer (sold separately) to take advantage of the DSP. I have the ICP5 DSP programmer.

cx-10.jpg

The DSP programmer is able to be used together with the free software SigmaStudio from Analog Devices to write to the chip on the amp. This combination unlocks many possibilities of crossovers, equalization, logic functions, and many more DSP effects available in SigmaStudio.

At first, I tried to eq the speaker using regular parametric EQ filters, but I found a better solution in the list of SigmaStudio functions, FIR filters.

First I generated an average response of the speaker ranging from 0 to 50deg. I then divided (A/B) the average response against an equalization curve from REW EQ (Generate measurement from target shape) which resulted in an inverted response of the speaker. With the inverted response, I generated a minimum phase (Generate Minimum phase...) on the Impulse tab of REW. I've also set the IR window to 5ms length, which results in 245 samples. Lastly, from the All SPL tab of REW, I selected the inverted response and exported IR as TXT with the settings below.

{16989772-12DE-4B9B-9F28-97E2CBD8235D}.png

The sample count is critical because the board gives errors above 330 samples per channel.

Once the inverted IR is exported from REW it can be imported into SigmaStudio. The window to the right is the crossover configuration.

{9BC8C614-7D05-4D40-A5C3-D2227EBCAA1A}.png

This is how I've programmed the amp currently. Stereo 2 channel input -> DC blocking ->high pass 200Hz -> inverted FIR filter -> crossover -> 4 channel out
The software allows for very intricate layouts and configurations. The only limit is the memory on the board.

{3603A5DE-C0F6-40D9-AE0E-82DBF5D66E4C}.png

I've also "upgraded" the spin rig to make use of the repeated measurements mode on REW for quick spins.

cx-11.jpg

These are the results after crossover optimizations and FIR filtering. Full 360 spins for left and right speakers.

L-XO+FIR 6p.pngR-XO+FIR 6p.png

It's not perfect, but a lot better than what I started with.

why not 3D print the whole enclosure with smooth shapes and stop worrying about diffraction entirely?
3D printing is still not cost-efficient for larger objects (translated: I'm cheap :D). I use the printer for complex items that cannot be easily built using other methods like CNC carving, welding, etc. One example is the printed dial on the spin rig which has a complex shape and it's relatively small. Since I already had the MDF enclosures, it was more efficient to adapt to them than to start from scratch.
 
Back
Top Bottom