• 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!

Introducing Directiva - An ASR open source platform speaker project

jtwrace

Major Contributor
Forum Donor
Joined
May 31, 2017
Messages
1,225
Likes
1,401
Location
Orlando, FL
OP
Rick Sykora

Rick Sykora

Major Contributor
Forum Donor
Joined
Jan 14, 2020
Messages
3,517
Likes
7,027
Location
Stow, Ohio USA
Thanks to you and the team for doing this. Does this include a well-documented write-up complete with lessons learned and how-to in one dedicated place? That would be great. If so folks can truly learn and modify (or just straight on build and set up miniDSP 2x4 to spec). One of the source of frustrations for anyone new (and not so new) to DIYing is the real lack of documentation or even accuracy about a given final design and build, followed by using idiosyncratic or obsolete software or spreadsheets that rely on Windows 7 or somesuch :facepalm:;), plus references to 30-page threads that are left dangling...

Agree, documentation is important. Right now, the main writer is me and am still learning as I go.

Once the initial prototype is built and sent for testing, I will build another and document it along the way. Maybe with some help (I know some writers). ;)
 

richard12511

Major Contributor
Forum Donor
Joined
Jan 23, 2020
Messages
4,335
Likes
6,700
Okay, this is not going to be easy.

The small waveguide of the Seas-DXT controls the sound radiation only at high frequencies. In the manufacturer's measurements (quasi half-space measurements) you can see that the waveguide controls the radiation well down to about 5kHz (constant directivity CD behavior when comparing on-axis and 60° measurement).
View attachment 112953

By "removing" the baffle from the tweeter as early as possible, it radiates wider than with a baffle (which acts like a simple waveguide).
Simulated Seas-DXT in cabinet v9 without XO (10° steps):
View attachment 112958
This allows us to significantly expand the CD behavior of the waveguide in conjunction with the baffle. The 60° frequency response (FR) now runs more or less parallel to the on-axis FR down to about 1.5-2kHz.

This can be better seen by looking at the graph normalized to the on-axis FR.
Simulated Seas-DXT in cabinet v9 without XO (10° steps) normalized to deg0:
View attachment 112960


The situation is somewhat different with the woofer. There, the cone acts as a waveguide at higher frequencies.
What we want here, however, is that the transition from the 2pi behavior (the baffle influences the radiation) to the 4pi behavior (the wavelength is so large that the baffle has no influence on the radiation) starts as late as possible and the transition takes place as evenly as possible.

Therefore, (in very simplified terms) the woofer should have as much baffle around it as possible. Because the baffle also acts here like a simple waveguide focuses the radiation at low frequencies.
View attachment 112965

What about a baffle design similar to the Grimm Audio LS1be?
 
OP
Rick Sykora

Rick Sykora

Major Contributor
Forum Donor
Joined
Jan 14, 2020
Messages
3,517
Likes
7,027
Location
Stow, Ohio USA
What about a baffle design similar to the Grimm Audio LS1be?

Most anything can be done with enough time or money, but since you seem rational, help with why you are asking?

Given the progression we have shown for the Directiva design, seems like you are asking whether we can fit a square peg in a round hole? ;)

Please elaborate. Thanks!
 

ctrl

Major Contributor
Forum Donor
Joined
Jan 24, 2020
Messages
1,616
Likes
6,086
Location
.de, DE, DEU
OFF-TOPIC

What about a baffle design similar to the Grimm Audio LS1be?
The Grimm LS1 shows that the combination of tweeter with small waveguide and wide baffle does not quite work out.
Please do not misunderstand the LS1 is certainly an excellent speaker, I am purely interested in the directivity of the LS.

1. Grimm LS1
2. Sketch LS1, in the simulation, the LS is upside down, but this does not matter.
3. horizontal FR deg0-90, normalized on-axis (15° steps)
4. Spectrogram +-180° normalized on-axis

1613504044338.png 1613503261045.png 1613503298649.png 1613503834388.png
Thus, starting from 5kHz, down to 1.8kHz, the 45° FR increases by almost 5dB and again reaches the sound level of the on-axis FR.

Measurements in the magazines confirm the simulation, for example here:
1613504651652.png
Source: hifitest.de

But the horizontal radiation pattern is not everything. For example, the (simulated) curve of the early reflections of the LS1 is very balanced:
1613504735402.png
 

richard12511

Major Contributor
Forum Donor
Joined
Jan 23, 2020
Messages
4,335
Likes
6,700
OFF-TOPIC


The Grimm LS1 shows that the combination of tweeter with small waveguide and wide baffle does not quite work out.
Please do not misunderstand the LS1 is certainly an excellent speaker, I am purely interested in the directivity of the LS.

1. Grimm LS1
2. Sketch LS1, in the simulation, the LS is upside down, but this does not matter.
3. horizontal FR deg0-90, normalized on-axis (15° steps)
4. Spectrogram +-180° normalized on-axis

View attachment 112980 View attachment 112973 View attachment 112974 View attachment 112977
Thus, starting from 5kHz, down to 1.8kHz, the 45° FR increases by almost 5dB and again reaches the sound level of the on-axis FR.

Measurements in the magazines confirm the simulation, for example here:
View attachment 112984
Source: hifitest.de

But the horizontal radiation pattern is not everything. For example, the (simulated) curve of the early reflections of the LS1 is very balanced:
View attachment 112985

Interesting. So you're saying the LS1 likely has those same off axis problems? For the $35k price, you would expect it to be perfect. They only show on axis and sound power on their page. On axis looks perfect, but sound power does have problems that you normally don't see with SOTA speakers. Disappointing if it also has off axis problems. Perfect on axis isn't enough when the asking price is that high(imo).

Edit: Sorry for the off topic. I just thought it was a relatively DIYable(simple cabinet) speaker that offers SOTA measurements. They say in their documents "Why isn't every speaker designed like this?"
 
Last edited:
OP
Rick Sykora

Rick Sykora

Major Contributor
Forum Donor
Joined
Jan 14, 2020
Messages
3,517
Likes
7,027
Location
Stow, Ohio USA
Back to our regularly scheduled program...

Here is what happens to the directivity when you do not recess the drivers:

1613506909584.png1613507013911.png

Not pretty is it?
 
OP
Rick Sykora

Rick Sykora

Major Contributor
Forum Donor
Joined
Jan 14, 2020
Messages
3,517
Likes
7,027
Location
Stow, Ohio USA
Is the narrowing toward the top of the speaker (option 8) accomplishing a similar goal as a wave guide that is sized to match the woofer? Is it narrowing the dispersion of the woofer to match the tweeter near the crossover? I can't quite tell from the graphs, and I don't have the experience to understand exactly what is going on.

This should help. A perfect version of this chart would look like so...
1613507427676.png

For more details, see here: Understanding Loudspeaker Review Measurements Part II | Audioholics
 

hex168

Senior Member
Joined
May 29, 2020
Messages
396
Likes
338

test1223

Addicted to Fun and Learning
Joined
Jan 10, 2020
Messages
503
Likes
511
Interesting project.

I would recommend to have a closer look at estimating the directivity index of the speaker. You will easily underestimate the frequency where the crossover takes place, if you focus to much at optimizing the horizontal radiation pattern alone.

All fancy edges at the front will also change the behavior when you alter the vertical angle a bit. Such edges make it harder to predict the outcome of the resulting di. But if you are willing to simulate a lot you may find a good design.

If you want to make a really good speaker you have to determine the listening distance for this design. Than you can find the important angles, where the first reflections take place and optimized the distance / phase between woofer and tweeter accordingly.
 

MarsianC#

Active Member
Joined
Jul 23, 2020
Messages
232
Likes
264
Location
Austria
Great thread!
I am still learning to use Akabak/Abec for my hunt after the perfect solution.... at least regarding directivity.
Step one is a project similar to Genelec's W371A. Not as tall, but using 14" and 12" too.
 

q3cpma

Major Contributor
Joined
May 22, 2019
Messages
3,060
Likes
4,416
Location
France
Great thread!
I am still learning to use Akabak/Abec for my hunt after the perfect solution.... at least regarding directivity.
Step one is a project similar to Genelec's W371A. Not as tall, but using 14" and 12" too.
I fear that, while the W371A is an interesting and certainly great hardware design, the DSP is where the "magic" resides.
 

DDF

Addicted to Fun and Learning
Joined
Dec 31, 2018
Messages
617
Likes
1,355
Interesting. So you're saying the LS1 likely has those same off axis problems?

You can also think of a wide baffle like a power response equalizer. For flat on axis, a wide baffle moves the power response dip to a lower frequency than for a narrow baffle. Here's a detailed LS-1 analysis I made comparing wide to narrow baffle, assuming equal woofer and tweeter native directivity.

Grimm LS-1 wide baffle normalized to on axis:
1613514216249.png


Standard parts express "narrow" box normalized to on axis:

1613514270741.png


To really perfect the wide baffle directivity concept, the baffle would be so wide that the off axis dips are pushed down to below Schroeder and suffer the slings and arrows of room modes, where they can be cleaned up using room eq without torching the perceived on axis. By that point, it would be fugly big.
 

Lbstyling

Addicted to Fun and Learning
Forum Donor
Joined
Jan 2, 2019
Messages
500
Likes
461
Interesting. So you're saying the LS1 likely has those same off axis problems? For the $35k price, you would expect it to be perfect. They only show on axis and sound power on their page. On axis looks perfect, but sound power does have problems that you normally don't see with SOTA speakers. Disappointing if it also has off axis problems. Perfect on axis isn't enough when the asking price is that high(imo).

Edit: Sorry for the off topic. I just thought it was a relatively DIYable(simple cabinet) speaker that offers SOTA measurements. They say in their documents "Why isn't every speaker designed like this?"

It's more of a 'stuff moved really fast once we all connected the dots about 6 years ago'. Some were there before others. I think Ziltch's Econowave really woke alot of people up. The Grimm was ahead for about 6 months/a year, but then we all started realising that certain shapes work and more importantly, why.

It's moved on again in several ways since.

Waveguides can be predicted accurately, and are understood to make the 2 way outperform most 3 ways for zero cost extra.

JBLs strange waveguide profile is now fully understood (for instance), transducer design has got much more competitive now that the field has converged on what it wants, consensus has converged away from passive crossovers (the old guard has died off frankly), low Le drivers are better understood, as off axis performance is preferenced by customers. Transducer design is prioritising cone geometry.

People forget, it was only 12 years ago that we were in the dark ages of flagship ring radiators that played up to 40khz and cost £300 each.

When the LS1 came out, some of this was being discussed, but there was almost no consensus.
 
Last edited:

Lbstyling

Addicted to Fun and Learning
Forum Donor
Joined
Jan 2, 2019
Messages
500
Likes
461
Interesting project.

I would recommend to have a closer look at estimating the directivity index of the speaker. You will easily underestimate the frequency where the crossover takes place, if you focus to much at optimizing the horizontal radiation pattern alone.

All fancy edges at the front will also change the behavior when you alter the vertical angle a bit. Such edges make it harder to predict the outcome of the resulting di. But if you are willing to simulate a lot you may find a good design.

If you want to make a really good speaker you have to determine the listening distance for this design. Than you can find the important angles, where the first reflections take place and optimized the distance / phase between woofer and tweeter accordingly.

Not the way I would choose to think about it these days. Active speaker, steep filters are available. This would have been a big deal in the old passive days yes. Active? Get the drivers as close as you can and the crossover as low as you can given the allowable distortion and that's the best you can do. You can time align to impulse correctly for a given height in software.

The DI is quite predictable with steep filters also.

I think they are spot on the money with the approach. Heavy on the modelling as mistakes are cheap at this stage. Lots of discussion, but some clear experience behind some of these choices, and none of the 'old guard' 'old software' red flags. If I see some proper trial and error with wood and ugly looking prototypes, I'm gunna be tipped into impressed. Quick dirty prototyping is where the real designers cut the mustard because its first principle driven.
 
Last edited:

Chromatischism

Major Contributor
Forum Donor
Joined
Jun 5, 2020
Messages
4,765
Likes
3,703
I think that might come down to the beam width. Current version is showing about 65deg. If that was significantly more, total energy in the top octave would be too high. Depends what the final version ends up at I would expect.
Even at 65 degrees, IMO, the 2500-4500 Hz range is too high. Great for older ears, though as that is a primary range affected by age-related hearing loss.

Off-axis the balance is even more skewed in that range.

Interesting discussion on waveguide design.
 
Last edited:

ctrl

Major Contributor
Forum Donor
Joined
Jan 24, 2020
Messages
1,616
Likes
6,086
Location
.de, DE, DEU
I would recommend to have a closer look at estimating the directivity index of the speaker. You will easily underestimate the frequency where the crossover takes place, if you focus to much at optimizing the horizontal radiation pattern alone.
I agree with you there. The vertical FRs are always simulated as well, but it would go beyond the scope of this thread to discuss them here as well.

In the final version there will be the complete information about the directivity of the LS.

If you want to make a really good speaker you have to determine the listening distance for this design. Than you can find the important angles, where the first reflections take place and optimized the distance / phase between woofer and tweeter accordingly.
The virtual microphone is located at a distance of 2m in all simulations. The phase shift due to the different signal propagation times of the drivers (driver distance is 0.145m) is only 12°, even at a high crossover frequency of 2.4kHz, it has hardly any influence on the axis frequency response in a crossover with even order.

In order to get a more complete overview of the simulated loudspeaker, for the possible final versions, all simulated FR (72 pieces for each driver) are exported and imported into VituixCAD to be able to view in advance the simulated loudspeaker according to CTA-2034-A (DI, PIR, LW...).

@Rick Sykora will surely show this in detail in future posts.
 

test1223

Addicted to Fun and Learning
Joined
Jan 10, 2020
Messages
503
Likes
511
Not the way I would choose to think about it these days. Active speaker, steep filters are available. This would have been a big deal in the old passive days yes. Active? Get the drivers as close as you can and the crossover as low as you can given the allowable distortion and that's the best you can do. You can time align to impulse correctly for a given height in software.
Yes digital active is the best way.
You can apply brick wall filter, but you have to be aware, that you don't get pre ringing issues and a higher time delay. So you will end up with some overlap in most cases. Nevertheless the potential di jump in the crossover region is always very important if you have an overlap you add another degree of freedom to change the di with the cost of dips and peaks at some angles.

Close drivers seems best. You get less source separation but everything is a compromise e.g. in an unlucky case the phase is not ideal and you didn't have the option for a passive crossover with fewer parts. There is also the problem with the surface of the woofer which interferes the most if the tweeter is placed nearby.

It is always beneficial to know the most of the room and placement.
E.g. there is the issue with none spherical waves in some frequencies so the distance will alter the frequency response. There are several other issues...
 
OP
Rick Sykora

Rick Sykora

Major Contributor
Forum Donor
Joined
Jan 14, 2020
Messages
3,517
Likes
7,027
Location
Stow, Ohio USA
PLEASE NOTE: Just to clarify some misunderstanding over the Directiva tweeter selection. Earlier I had posted that I was finalizing the tweeter selection. The context here was amongst the existing team and sharing some of where I had sought (past tense) help with tweeter selection. Some have misread my post as a though we were still looking for tweeters. I regret if this was unclear, but the Seas DXT is the tweeter we have selected. Later derivatives may consider other tweeters, but for the first implementation, I consider the tweeter selection to be final.

So while @ctrl was simulating directivity, I had been modeling and prototyping the cabinet tuning. If you had not previously noticed, I have posted the effort here: Can Purifi's SPK5 be improved with a passive radiator? | Audio Science Review (ASR) Forum

With the equalization and filtering available with an active design, we can address some of the challenges of getting low bass out of the Purifi woofer more readily than if the design were passive. As you have read, I have tried to ensure the Directiva is affordable. An important part of determining the final cabinet is the volume needed. My prototyping has been with the SPK5 cabinet, so had to consider how ports or passive radiators would fit into the Directiva cabinet. I think I have a handle on a vent size that does not have major misbehavior, but the passive radiator has more potential price/performance considerations.

The SB 5x8 passive radiator is inexpensive, but has a large footprint...
1613525745140.png

So, the other consideration is the passive radiator that Purifi designed for the woofer. Like the woofer, it is pricey. But as it is smaller, it is easier to incorporate on the Directiva back panel.
1613525665477.png

I had to order the Purifi PR and they just arrived today. As it is part of finalizing the overall design, will be taking some time away from posting to finalize the cabinet alignments. Before I do, will preview the proposed final cabinet. While I liked latest design, I had some reservation about its complexity, so I proposed a truncated pyramid. We had some back and forth over the dimensions, how the drivers would fit and angles that would be easier to cut. After more simulations, here is what we designed...

1613526337421.png

This has about the same directivity (maybe a bit better) than the prior cabinet shape. As I mentioned, my plan was to offer options for a choice of cabinet, but this is the one I plan to build and have Amir test. For comparison purposes, I think it would be valuable to build a version with the unmodified Denovo cabinet. But one step at a time for now.:)
 
Last edited:
Top Bottom