Interesting new studio speaker design from present day production

[kemmler3D]

Related question and I know DIYAudio is a better place to ask this, but if you have a 3D printer and know your way around a CAD package, custom waveguides aren't hard or expensive to make at all. Probably less than $10 of plastic and maybe 10 hours of printing. They are however, still hard to DESIGN.

Anyone know the easiest way to simulate the output of an arbitrarily shaped waveguide? Is there BEM that can be done by non-engineers without spending $XXXX on software?

 

[Scgorg]

Related question and I know DIYAudio is a better place to ask this, but if you have a 3D printer and know your way around a CAD package, custom waveguides aren't hard or expensive to make at all. Probably less than $10 of plastic and maybe 10 hours of printing. They are however, still hard to DESIGN.

Anyone know the easiest way to simulate the output of an arbitrarily shaped waveguide? Is there BEM that can be done by non-engineers without spending $XXXX on software?

A lot of people (including me) use AKABAK/ABEC. For non-commerical purposes you can get a license for free (big props to Jörg Panzer), or you can use the demo version which has full functionality except allowing you to save results. The learning curve is still pretty vertical, but lots of people on DIYaudio are helpful in starting you out. I've attached an example of the kind of data it can give you (this particular case is from a midrange waveguide I am making, hence limiting the simulation to 250-5kHz). The normalized off-axis simulation is done in 5 degree increments.

You can also simulate a lot of other stuff in AKABAK/ABEC, I highly recommend learning it if you like DIY or are just curious about various acoustic principles.

 

[krabapple]

So presumptive and judgmental. You guys ain't Scientists, clearly.

I watched the complete video with an open mind. And guess what, it's a nice/clever teaser- YouTube style. it's just part one of two (or more)

Video 1 summary:
Premise: we want a surround sound Atmos monitoring system, but we don't went buy a whole ATC setup, so we're going to design our own.
We've chosen our amp/DSP, and it's the Hypex FA253.
[emd]

But that's boring. Let's create a video on it.

And where's the Science? Since you mentioned it.

 

[kemmler3D]

A lot of people (including me) use AKABAK/ABEC. For non-commerical purposes you can get a license for free (big props to Jörg Panzer), or you can use the demo version which has full functionality except allowing you to save results. The learning curve is still pretty vertical, but lots of people on DIYaudio are helpful in starting you out.

I looked at the product page for AKABAK and it looked fairly intimidating. My problem is I'm not an engineer and in fact I'm only really good with mesh modeling tools. Never got around to learning Solidworks or even Fusion 360 beyond really simple stuff. How do you set up a simulation in AKABAC/ABEC, do you need CSG files, can you import STL, or is it something even more hair-raising?

I'm OK with the general acoustic concepts, had some actual education on those, my main concern is getting 3D data into the simulator and back out again for 3D printing. I am also pretty bad with electronics (never had education on that beyond 9th grade) so hopefully that's not a big piece of it.

 

[Scgorg]

I looked at the product page for AKABAK and it looked fairly intimidating. My problem is I'm not an engineer and in fact I'm only really good with mesh modeling tools. Never got around to learning Solidworks or even Fusion 360 beyond really simple stuff. How do you set up a simulation in AKABAC/ABEC, do you need CSG files, can you import STL, or is it something even more hair-raising?

I'm OK with the general acoustic concepts, had some actual education on those, my main concern is getting 3D data into the simulator and back out again for 3D printing. I am also pretty bad with electronics (never had education on that beyond 9th grade) so hopefully that's not a big piece of it.

You can import mesh files. Personally I draw and iterate in fusion360, export as .step into gmsh, then use gmsh for meshing and import the mesh into AKABAK. Having a tool that gives you a lot of control over the meshing is helpful, as more mesh will increase computation time by quite a lot (each doubling of elements will require 8x the solving time in 3D). No electronics knowledge is required for waveguide design. As for not being an engineer: I'm not an engineer either

It's a big pain to get your first few simulations going, but after you've done it a couple of times it starts becoming intuitive, and you'll be able to look into more advanced stuff.

 

[dfuller]

Is directivity control necessary for 3-way designs? I though you could just match the surface areas of each driver to each other and you're more or less done?

The lack of round overs is concerning though.

Eh, sorta? There's a reason you see waveguides even on three ways.

 

[kemmler3D]

Cool, thanks! My ultimate goal is to match the Bliesma T34A to the Purifi NAA 6.5, which would require giving the tweeter more directivity in the ~2-3khz range before the midwoofer is fully crossed out. My plan is to do it active so I can have the crossover range be pretty short, but at 2khz there's already a pretty big directivity difference between the woofer and the tweeter, like 6db at 60deg, and it goes faster from there. However, from the look of it, the distortion will be minimized with a handoff around 2khz so that's the goal.

 

[Gringoaudio1]

I'm holding my horses-

What the baffle dimensions? It may or may not make a measurable or audible difference to go the whole 9 yards, particiarly in a design where costs need to be controlled.

For example, here's the baffle diffraction effect on a 1" dome tweeter on a typical 20cm x 40cm baffle, tweeter 10 cm from the top edge, and centered on the midline.

(1)

View attachment 233966

The green is what would happen on an infinitely large baffle ie. 6dB gain. The blue is what happens to the response instead-
Yikes! Lots of funkiness between about 550Hz up to about 15Khz, with almost up to +/- 3dB between 550Hz and 3Khz.
In a tweeter, where one would have to crossover around ~2.5Khz- this is what creates all kinds of challenges due to the funkiness that changes the on axis (shown) and the off-axis response (not shown, but equally complex).

Here's how it looks now with a 1" roundover:

(2)

View attachment 233967

Ah better, particularly above 6Khz, but still +2/-1dB effects between 550Hz and 3Khz.

(Now a waveguided tweeter, in effect, has it's own concave baffle, so the tweeter is not affected by the flat baffle it's on, thus avoiding the baffle diffraction issue of the lower treble ripple response as seen above. Related to this is the tweeter's output, which has a change in dispersion that closely matches the woofer's dispersion (that's why you want a tweeter on small 5" waveguide when matched with a small 5" mid-woofer and much larger waveguide when matched with a larger 15" midwoofer in a 2 way)


Moving on, here's the same tweeter 10cm from the top of a wider 11" baffle, which is typically what's needed for a 8-10" woofer in a 3 way.

View attachment 233970

Now you're going to say- well it's not as good as a rounded over cabinet! And I agree.
But certainly it's not as atrocious as the initial cabinet with no round-over.

But guess what? In a 3 way you're typically crossing the tweeter to the midrange cone/dome at or above 3KHz,
(3)

View attachment 233971


So the funkiness in the region of interest is +/-1 dB, not +/-3dB.

Here's the same, with just a 1/2" roundover:
(4)

View attachment 233975

So on a 3 way design, for a tweeter on a 11" baffle- the diffraction effects are reduced.

So here are the options:
A) do less baffle edge treatments (graph 4)
B) live with it (graph 3)


So let's be aware of our blind spots: optimal baffle shape and dimensions is but one part of loudspeaker design.
But it depends on what frequencies you're dealing with.


What happens when you put a 3" piston (approximation of a 3" cone or dome on the baffle. You can equalize out the peak around 1Khz.
Can you live with that 1dB dip around 2Khz? Or do you need a cu$tomi$ed waveguide?

View attachment 233976



It's not as "terrible" as you think.... It costs a LOT of money to go with a fully customized diffraction mitigation/adaptation/controlled baffle
ala. Vivid Audio...

Thanks. What software generated these FR curves?

 

[Galliardist]

I'm holding my horses-

What the baffle dimensions? It may or may not make a measurable or audible difference to go the whole 9 yards, particiarly in a design where costs need to be controlled.

For example, here's the baffle diffraction effect on a 1" dome tweeter on a typical 20cm x 40cm baffle, tweeter 10 cm from the top edge, and centered on the midline.

(1)

View attachment 233966

The green is what would happen on an infinitely large baffle ie. 6dB gain. The blue is what happens to the response instead-
Yikes! Lots of funkiness between about 550Hz up to about 15Khz, with almost up to +/- 3dB between 550Hz and 3Khz.
In a tweeter, where one would have to crossover around ~2.5Khz- this is what creates all kinds of challenges due to the funkiness that changes the on axis (shown) and the off-axis response (not shown, but equally complex).

Here's how it looks now with a 1" roundover:

(2)

View attachment 233967

Ah better, particularly above 6Khz, but still +2/-1dB effects between 550Hz and 3Khz.

(Now a waveguided tweeter, in effect, has it's own concave baffle, so the tweeter is not affected by the flat baffle it's on, thus avoiding the baffle diffraction issue of the lower treble ripple response as seen above. Related to this is the tweeter's output, which has a change in dispersion that closely matches the woofer's dispersion (that's why you want a tweeter on small 5" waveguide when matched with a small 5" mid-woofer and much larger waveguide when matched with a larger 15" midwoofer in a 2 way)


Moving on, here's the same tweeter 10cm from the top of a wider 11" baffle, which is typically what's needed for a 8-10" woofer in a 3 way.

View attachment 233970

Now you're going to say- well it's not as good as a rounded over cabinet! And I agree.
But certainly it's not as atrocious as the initial cabinet with no round-over.

But guess what? In a 3 way you're typically crossing the tweeter to the midrange cone/dome at or above 3KHz,
(3)

View attachment 233971


So the funkiness in the region of interest is +/-1 dB, not +/-3dB.

Here's the same, with just a 1/2" roundover:
(4)

View attachment 233975

So on a 3 way design, for a tweeter on a 11" baffle- the diffraction effects are reduced.

So here are the options:
A) do less baffle edge treatments (graph 4)
B) live with it (graph 3)


So let's be aware of our blind spots: optimal baffle shape and dimensions is but one part of loudspeaker design.
But it depends on what frequencies you're dealing with.


What happens when you put a 3" piston (approximation of a 3" cone or dome on the baffle. You can equalize out the peak around 1Khz.
Can you live with that 1dB dip around 2Khz? Or do you need a cu$tomi$ed waveguide?

View attachment 233976



It's not as "terrible" as you think.... It costs a LOT of money to go with a fully customized diffraction mitigation/adaptation/controlled baffle
ala. Vivid Audio...

I've been given this information verbally a few times but the measurements make it much clearer. Thanks.

I should probably read some better books on speaker design...

 

[tktran303]

Thanks. What software generated these FR curves?

The pre-eminent CAD software for loudspeaker design in C21, IMHO:

Features and users:

VituixCAD features

Apologies my last 2 graphs had different Y scale, which was unintentional.

I will go back and update later if possible,

 

[tktran303]

Sorry the video is boring and wastes time. Even more so when you need part two. Your short summary in video form would be highly preferred. What they think is a cute entertaining video isn't plus what they think is a clever bit of reasoning for a diy speaker also isn't.

Well that's the Youtube format. It's the captions, the graphics, the hook and then the content drawn out to past 8 mins.

We don't complain when Kal writes a few pages about equipment set-up, the ancilliary equipment, the listening impressions, the program material, comparisons, the features and the prices and the measurements, do we? Or do we?

Who'd want to read a 1 paragraph summary of a Stereophile review?

No, people should read the entire review to form an opinion. I'm quite surprised at Kal's comment that he didn't get past 1 min of "BS"

 

[tmuikku]

I looked at the product page for AKABAK and it looked fairly intimidating. My problem is I'm not an engineer and in fact I'm only really good with mesh modeling tools. Never got around to learning Solidworks or even Fusion 360 beyond really simple stuff. How do you set up a simulation in AKABAC/ABEC, do you need CSG files, can you import STL, or is it something even more hair-raising?

I'm OK with the general acoustic concepts, had some actual education on those, my main concern is getting 3D data into the simulator and back out again for 3D printing. I am also pretty bad with electronics (never had education on that beyond 9th grade) so hopefully that's not a big piece of it.

Acoustic Horn Design – The Easy Way (Ath4)

Acoustic Horn Design – The Easy Way (Ath tool) This is an associated thread for a waveguide generator called Ath (Advanced Transition Horn), version 4. Ath is a piece of software for designing waveguides and horns. At the same time it makes it possible to easily simulate their acoustic...

www.diyaudio.com

 

[DJNX]

Part 2.
Drivers.

 

[tktran303]

Bliesma Alu/Magnesium for hi-fi- the widest dispersion of the bunch.

 

[kemmler3D]

^^ indeed, and the purifis do pretty well up to about 2khz apparently.

 

[sam_adams]

All talk and no measurements to back up the 'extraordinary' claims. Also, why is there another discussion about this 'product'?

 

[fluid]

Cool, thanks! My ultimate goal is to match the Bliesma T34A to the Purifi NAA 6.5, which would require giving the tweeter more directivity in the ~2-3khz range before the midwoofer is fully crossed out. My plan is to do it active so I can have the crossover range be pretty short, but at 2khz there's already a pretty big directivity difference between the woofer and the tweeter, like 6db at 60deg, and it goes faster from there. However, from the look of it, the distortion will be minimized with a handoff around 2khz so that's the goal.

As I have the same drivers heading to my door currently I can help you out a little bit. The T34A is super wide with a tall dome, historically tall domes have done better with basic circular radius sections and the T34A is no different. A 6mm deep guide with a 90mm diameter circle made to be tangent with the baffle and meeting the edge of the surround work well to keep everything nice and smooth. Here is simulated AKABAK output in an infinite baffle. Ignore above 15k too much as mesh resolution starts to be an issue there.

An OS style throat in a deeper waveguide causes trouble at 10K plus as the tall dome shadows itself in a high sided throat. So creating more directivity lower down will translate to more trouble higher up,

You can see a lot of simulations from @ctrl and correlations to measurements in this great German thread. There the T34A was crossed at 1200Hz.
https://www.diy-hifi-forum.eu/forum/showthread.php?17431-Die-AB-Wave&p=247807&viewfull=1#post247807

 

[dfuller]

All talk and no measurements to back up the 'extraordinary' claims. Also, why is there another discussion about this 'product'?

The Bliesma M74S datasheet is pretty impressive, I'd say.

 

[kemmler3D]

As I have the same drivers heading to my door currently I can help you out a little bit. The T34A is super wide with a tall dome, historically tall domes have done better with basic circular radius sections and the T34A is no different. A 6mm deep guide with a 90mm diameter circle made to be tangent with the baffle and meeting the edge of the surround work well to keep everything nice and smooth. Here is simulated AKABAK output in an infinite baffle. Ignore above 15k too much as mesh resolution starts to be an issue there.

An OS style throat in a deeper waveguide causes trouble at 10K plus as the tall dome shadows itself in a high sided throat. So creating more directivity lower down will translate to more trouble higher up,

View attachment 234548

You can see a lot of simulations from @ctrl and correlations to measurements in this great German thread. There the T34A was crossed at 1200Hz.
https://www.diy-hifi-forum.eu/forum/showthread.php?17431-Die-AB-Wave&p=247807&viewfull=1#post247807

Really interesting stuff. And somewhat daunting that an increase in low directivity causes those problems higher up. Somehow I imagined a nice shallow WG might get something to happen at ~1-3Khz while leaving the rest of the spectrum alone... maybe the solution is to cross a bit lower and sacrifice a little on distortion. Hmm...

 

[fluid]

Really interesting stuff. And somewhat daunting that an increase in low directivity causes those problems higher up. Somehow I imagined a nice shallow WG might get something to happen at ~1-3Khz while leaving the rest of the spectrum alone... maybe the solution is to cross a bit lower and sacrifice a little on distortion. Hmm...

To get a nice balance of on and off axis responses with smooth power you might find that having the tweeter be less directive at the crossover actually helps.

https://www.diyaudio.com/community/threads/vituixcad.307910/post-6538511

1) Tweeter (wave guide) much less directive at XO than woofer.
2) c-c distance ca. 1.2 x wave length at XO.
3) Box shape to decrease directivity at XO and increase directivity octave above XO (smoothly with diffraction without sharp edges).
4) Phase match octave above XO and possibly clear mismatch octave below XO.

Items 1-3 are implemented in proper 2-way design. This is very rare in practice in my opinion.

If you find that you do need a bit more directivity at 3K, then you can put a vertical step in, with the profile on top, that helps without upsetting everything else too much.