Introducing Directiva - An ASR open source platform speaker project

[Ericglo]

Back when Rick asked for some suggestions for the tweeter, I offered the DXT as a possibility, but my #1 recommendation was SB26ADC-000-4 for its better distortion performance (posted earlier, my bad).

Rick had initially suggested the SB26STWGC-4. but the top end is rough and would need extra components to EQ (mandatory)

The SB26STWGC-4 is an SB26STC (well regarded) with a waveguide. The small waveguide it uses just doesn't have enough gain (5 dB) to keep up with the distortion performance of other, better choices:

• Revel uses a variant of the SB26ADC-000-4, which shows great distortion performance here and here. Great price. Troels makes it work OK (but not great) off axis here. The SB26ADC can be improved with a waveguide such as Augerpro`s free design here. The CAD file shows that the 4in waveguide would fit, and maybe even the 5ìn (see here and here).
• An oldie but still goodie is the SB26STAC, MarkK tests and here
• DA25TX00-08 is an option I want to explore some day based on this
• Seas DXT is a good option for its built in waveguide but it has much more distortion 1-2k than the SB26ADC. I was also surprised to read this diva complaint

I then looked into ring radiators, which would need a waveguide to keep up with the Purifi's power handling capabilities

• DX25TG09-04 and Monocor WG-300 waveguide, tests results here show very nice directivity and great distortion results. But the waveguide is big and the SB26 has better performance

• XT25TG-30 and Monocor WG-300 waveguide, tests results here also show very nice directivity but the distortion is much higher, the DX25 combo wins in the battle of the ring radiators

The SB26ADC-000-4 with a waveguide is the front runner for me.

CTRLs diffraction models then simulated options. For a much more old school approach to this simulation but one that includes the tweeter's native frequency response as well, see my old tutorial here hosted at Charlie's page. There are better free tools around now, but the design concepts will be the same.

I was thinking about Augerpro's waveguide with the SB tweeter as well when I first saw the topic.

 

[Rick Sykora]

They're exceptional quality and the ones I've heard (my current speakers use one too) sound excellent. Many well known speakers use same as well.

JSmith

You are right that Scan-Speak is known for high quality drivers. They also extract a price for that quality. Their most popular tweeter is more than 3-7 times the price of the tweeters we are considering here. They are not 3-7 times better. So as long as cost and value are considered, Scan-Speak often ends up out of contention.

 

[diddley]

Great effort from you guys, wish you luck!

 

[MZKM]

Have used the Vifa and its bigger brother for the same reasons you found. However, their distortion profiles are not as good as the SB26ADC and have more options to deal with directivity than have ways to get lower distortion. So as @DDF illustrates above, it is the reason we had it is an option. Since I own all of them, we can try some variations once this first one is done.

IME developing products for over 3 decades, there is rarely one best answer and usually more than one good one. Between the good ones, there are usually tradeoffs (performance, reliability, quality, cost, longevity, etc.). As such, you take your best shot and usually do not get to revisit. Here we can (as long as the budget allows).

No arguments; and yeah, the distortion was the caveat I mentioned (however, if you can cross over at 2500Hz-3000Hz, that may resolve that issue).
I have no experience, but I personally would like to see <1% (-40dB) at 100dB, and that SB26ADC achieves that while that Vifa (with a 1kHz 4th order filter) does not, the 3rd harmonics are especially high while with the SB26ADC is much lower than the 2nd harmonics.

And for sure, if the SB26ADC works well, then it works well.
It’s just, that Vifa has some seriously wide directivity, so would be fitting of a speaker named the Directiva .

 

[maxxevv]

Follow up to post#18

UPDATE: The simulated drivers are the 6.5'' Purifi woofer and the SB Acoustics SB26ADC tweeter.
Cabinet is width x height x depth = 216mm x 406mm x 260mm

The order of the graphs is always:
1. sketch of the loudspeaker
2. frequency responses of the tweeter without crossover.
3. frequency responses of the tweeter normalized to 15°

If you arrange the tweeter on the baffle so that the top edge is only half as far from the center of the tweeter as the side edges, you get a slight improvement - version v2
View attachment 112769 View attachment 112770 View attachment 112771


The next step could be to bevel the side edges. So we grab the virtual saw and make a few bevels.
Version v4 with 40mm/30° bevel.
View attachment 112772 View attachment 112773 View attachment 112774
This significantly improves our design.


If you immediately hurt yourself with the saw, you can also grab the sandpaper and start sanding curves (for hours and hours...).
Version v5 shows 18mm rounding of side edges.
View attachment 112775 View attachment 112776 View attachment 112777
This optimization is also a significant improvement over Version v1 and v2.


Now we'll get cocky and add a rounding to the top edge as well. We don't even need to think about that, do we?
Version v6 shows 18mm rounding of side edges and top edge.
View attachment 112778 View attachment 112780 View attachment 112781
The result regarding the directivity is mixed, in the range 4-5kHz there is an improvement, but in the range around 3kHz there is worse directivity.

Now it's up to the marketing department to decide what to do next

Perhaps you can try the top edge with a smaller radius rounding. Gut feel is 12~15mm, more to the lower side.
And perhaps round the top side edges as a test too ?

What is the model format you have there ? STEP ? Parasolid ?
If you need some fancy waveguide modeling, I can spare some time these 2 weeks.

 

[JSmith]

You are right that Scan-Speak is known for high quality drivers. They also extract a price for that quality. Their most popular tweeter is more than 3-7 times the price of the tweeters we are considering here. They are not 3-7 times better. So as long as cost and value are considered, Scan-Speak often ends up out of contention.

Fair call, however personally if I was building speakers I'd be using the best possible components, YMMV.



JSmith

 

[Ericglo]

You are right that Scan-Speak is known for high quality drivers. They also extract a price for that quality. Their most popular tweeter is more than 3-7 times the price of the tweeters we are considering here. They are not 3-7 times better. So as long as cost and value are considered, Scan-Speak often ends up out of contention.

Then why are you considering the Purifi if cost is a concern?

Fair call, however personally if I was building speakers I'd be using the best possible components, YMMV.



JSmith

You may or may not know this already. Most of the DIY crowd look at the individual drivers regardless of price and determine their value. The manufacturer can become secondary. Scan Speak obviously makes good drivers along with a variety of others like SB, Seas, Purifi, etc. I mean who would have thought that the Tectonic BMR would work so well as a midrange. The fact that it is inexpensive is a bonus.

 

[PaulD]

Then why are you considering the Purifi if cost is a concern?




You may or may not know this already. Most of the DIY crowd look at the individual drivers regardless of price and determine their value. The manufacturer can become secondary. Scan Speak obviously makes good drivers along with a variety of others like SB, Seas, Purifi, etc. I mean who would have thought that the Tectonic BMR would work so well as a midrange. The fact that it is inexpensive is a bonus.

Not to forget the Bliesma tweeters if they do not blow the budget http://www.bliesma.de/products.html

 

[ctrl]

Perhaps you can try the top edge with a smaller radius rounding. Gut feel is 12~15mm, more to the lower side.
And perhaps round the top side edges as a test too ?

For the simulated loudspeakers I use created parameterized scripts, therefore only such changes are possible that I have implemented in the scripts.
For special cases, either the scripts would have to be adapted or a CAD model with meshing would have to be created externally - which would extremely increase the time required for each simulation.

What is the model format you have there ? STEP ? Parasolid ?

For BEM and LEM modeling I use the no longer supported software "ABEC". The current version is called "AKABAK" and offers more graphical tools compared to "ABEC", the functionality is basically identical.

BEM = Boundary Element Method/Analysis calculates the acoustic field satisfying boundary conditions, i.e. including reflection and diffraction. The Boundary Element Analysis solves the Helmholtz Integral for the interior and exterior radiation problem
LEM = The Lumped Element Method/Analysis calculates flows and potentials of a network of components, such as resistors, capacitors, etc. In ABEC this analysis is extended to the mechanical and acoustical domain, where you have components, such as the Driver and Waveguide component.

 

[JSmith]

The fact that it is inexpensive is a bonus.

Sure is... yet one can only hope the parts, build quality and longevity are there too. It can't all be about measurements only.

I'm quite interested in proven DIY designs, so will keep an eye on this thread.



JSmith

 

[Rick Sykora]

Then why are you considering the Purifi if cost is a concern?
.

I have built speakers with many of the other drivers (Seas, SB, Dayton, etc.). The Purifi woofer offers considerably better distortion performance at a price premium I was willing to accept. As I pointed out a few times already, I know there is more than one good solution. If someones else wants to try another, then once we show to do one good design, they can follow the our pattern and make their own tradeoffs.

 

[maxxevv]

For the simulated loudspeakers I use created parameterized scripts, therefore only such changes are possible that I have implemented in the scripts.
For special cases, either the scripts would have to be adapted or a CAD model with meshing would have to be created externally - which would extremely increase the time required for each simulation.


For BEM and LEM modeling I use the no longer supported software "ABEC". The current version is called "AKABAK" and offers more graphical tools compared to "ABEC", the functionality is basically identical.

BEM = Boundary Element Method/Analysis calculates the acoustic field satisfying boundary conditions, i.e. including reflection and diffraction. The Boundary Element Analysis solves the Helmholtz Integral for the interior and exterior radiation problem
LEM = The Lumped Element Method/Analysis calculates flows and potentials of a network of components, such as resistors, capacitors, etc. In ABEC this analysis is extended to the mechanical and acoustical domain, where you have components, such as the Driver and Waveguide component.

That sounds like FEM rather than CAD.

Usually, its doing a CAD model and then its converted into a FEM model to do FDM/ FEM / CFD analysis.

The description of the product page reminds me of a very primitive ANSYS version I used briefly in the 90's while in university.
I remember I had to convert it from a 3D format into elemental meshes to do the load analysis on it. Much more primitive than the current day software of just directly accessing the CAD model and let the software do the meshing for you.

If you have a format that I can export to for your analysis, I'll be glad to help.

 

[Rick Sykora]

Not to forget the Bliesma tweeters if they do not blow the budget http://www.bliesma.de/products.html

It would blow the budget considerably and several others have already paired it with the Purifi woofer. There are lots of good answers. I am relying on my team’s experience to do something that is solid and fits the price target I have set.

 

[ctrl]

What I don’t get about the 6.5” Purifi is that while Hi-Fi Compass shows it to beam around 1500Hz:
...
The SPK5 measured here was nearly flat between 1000Hz-2000Hz even at 90° (crossover is 3kHz):

I don't see any contradiction there. The measurements of Hificompass, if done correctly, should have been made in a IEC standard baffle - i.e. under 2pi/half-space conditions, so to speak.
The measured directivity of the Purifi woofer looks pretty typical for the driver dimensions.

The measurements of SPK5 were made on a finished loudspeaker, whose baffle massively influences the directivity of the driver.

My issue with the Purifi is the low sensitivity. Sure “watts are cheap”, but the majority of people have <100W 8ohm.

It all depends on the tuning. The lower the f3 point is set in the design, the lower the sensitivity of the speaker.
If the tuning is moderate, [email protected]@1m sensitivity should be possible. If the baffle step is not completely compensated, a higher sensitivity is possible.

 

[Lbstyling]

Further improvement for your consideration:

Looking at the tests done by Troels, and Patrick Bateman et Al,

The round over on the sides of the cabinet needs to be 2inch to have effect towards the lower treble/mid.

A surprising side effect of doing so is that it reduces energy on axis, this is because it 'robs' the SPL from there to fill in the off axis a little.

The problem is how do you practically do this? Easy! Cut a broomstick handle into quarters, or you can buy quadrant mouldings like this:

https://somtim.co.za/site/products/pine-interior-mouldings/pine-quadrant/

Now you have a speaker that is flagship. I think you may struggle to attain a truly competitive design without it, considering the cost will be significant if using the Purifi.

P.S. I assume that you are using a waveguide with that tweeter? Belisma would really make a difference here, but cost is extreme.

You will need a PR for the bass driver, or a very long slot port running around the bottom and back face to do it. (See hifimcompass experiments)

P.PS There is already a ASR speaker under development that many of us came to similar design conclusions as this, is this a continuation of the same one? Have you asked the developers of that one what their thought are?

 

[headshake]

Instant edge radius....

Would you ever just add moulding to the box? I found this stuff and am using for my 1st speaker. It would give you a 44mm edge radius. My sims showed anything over 30mm gives a nice 2-5khz.

https://fypon.com/products/mld605-12?variant=30373559697501

It is just moulding that is ready to paint. I got it shipped to my local hardware store for under $50. The downside is that it is 12feet long and needs a miter saw to cut.

 

[McFly]

Scanspeak D2905/9900 is another waveguided tweeter option, albeit a rather classic one.

 

[Lbstyling]

Instant edge radius....

Would you ever just add moulding to the box? I found this stuff and am using for my 1st speaker. It would give you a 44mm edge radius. My sims showed anything over 30mm gives a nice 2-5khz.

https://fypon.com/products/mld605-12?variant=30373559697501

It is just moulding that is ready to paint. I got it shipped to my local hardware store for under $50. The downside is that it is 12feet long and needs a miter saw to cut.

This kinda thing is good. If I remember right, 50mm was where significant improvements start, but that radius is harder to find.

Someone on DIYAudio did find a cheap supplier at one time.

 

[Lbstyling]

Scanspeak D2905/9900 is another waveguided tweeter option, albeit a rather classic one.

The best measuring one was Brandon's one (Augerpro from HTGuide). It's all in trial and error of the right waveguide profile.

If someone is willing to put the trial and error time in, you can make the perfect waveguide for any tweeter.

If I remember right, general consensus is that the waveguide width should match the mid driver width for best response, but most tweeters struggle to get down to the cross for good pattern control with a 6.5 inch without distortion raising too much. 5 inch bass driver is superior in that regard, so it pushes the design between Belisma (as it plays low and can cross to 6.5 inch) or a 5inch bass (SB CAC good but very SPL and bass limited, so you end up 3 way or with a sub)

Physics limit the choices here.

Pick your poison.

I would expect this to be about as good as is currently possible for DIY without going to a M2 style clone. (WAY bigger, but lower distortion, higher SPL, more efficient, smoother control (potentially), and detail in the bass.

Potential of a coax is higher, but none exist of a high enough calibre with the possible exception of the Denovo 15, but distortion is still too high in the treble for that.

 

[ctrl]

The round over on the sides of the cabinet needs to be 2inch to have effect towards the lower treble/mid.

The radius of the "round over" is indeed dependent on the baffle width.
The studies on the influence of "round over" and "bevels" are partly more than twenty years old, as listed here. One should always consider the rounded/beveled edges in conjunction with the baffle. Therefore, appropriate simulation makes life easier.

With round over, edge diffraction in the 2-4kHz range can be reduced, but usually not completely suppressed.

Rick will show later which design was chosen to achieve good directivity and completely suppress the typical widening in radiation around 2-4kHz.
The examples shown so far are the way, not the goal of the final design.

P.S. I assume that you are using a waveguide with that tweeter? Belisma would really make a difference here, but cost is extreme.

For the first version of the speaker, the drivers are already specified. After the completion of this project, modifications are certainly conceivable - but there @Rick Sykora is the better contact.