Introducing Directiva - An ASR open source platform speaker project

[muad]

@617 -

I thought it might be helpful to describe what is special about these directiva projects, what they offer, what they do not offer.

Traditional DIY designs tend to target all-around performance, with the same compromises that commercial speakers have. In other words, you set a budget, you set a size, and you make the best speaker you can within those specifications.

The directiva projects offer something a little different - they seek to achieve performance characteristics which are rarely available in DIY or commercial designs. They give you a taste of state of the art performance, but this comes with certain costs.

In the first speaker, the cost was complexity and expensive, esoteric drivers. What you got in return was probably the best performing small speaker on the planet, with limited bass performance.

This second speaker I offers better value, but it is still a complex design. What you get in return is a very special radiation pattern which is similar to a large waveguide speaker like a Gedlee Abbey or something. The bass and overall output is dependent on how much you are willing to spend, and although I hope the team develops a few bass modules, in reality any DIYer can make an active bass speaker.

In both cases, you get a design validated by careful measurements and a degree of customization - want 3db less treble? You can dial it in according to your room and your taste.

There are outstanding speakers on the market these days, and I believe the value proposition for DIY has weakened. Why DIY when Elac, KEF, Harman and Arendal make such good passive speakers? Why DIY when you can get a whole active system from Kali, Neuman or Genelec? Except in the case of very large speakers, DIY doesn't offer a ton of value, unless it does something special. That's where these speakers excel.

@617 - I completely agree with you. The limitations of DIY are the availability of good affordable waveguides or concentric drivers and the R&D to get the dispersion characteristics right, followed by the lack of actual anechoic/NFS measurements. I recently bought a LS50meta and feel that traditional DIY could not achieve that level of performance (dispersion and response) short of the DXT-mon or the Directivas. It's why I'm so interested in this project. I feel that directiva2 is likely end game material.

@TimVG
Directiva2's dispersion pattern looks incredible!

Really appreciate the hard work you guys are putting in. For some of us, this project is the closest an average joe with ever get to hearing true state of the art!

 

[617]

@617 -


@617 - I completely agree with you. The limitations of DIY are the availability of good affordable waveguides or concentric drivers and the R&D to get the dispersion characteristics right, followed by the lack of actual anechoic/NFS measurements. I recently bought a LS50meta and feel that traditional DIY could not achieve that level of performance (dispersion and response) short of the DXT-mon or the Directivas. It's why I'm so interested in this project. I feel that directiva2 is likely end game material.

@TimVG
Directiva2's dispersion pattern looks incredible!

Really appreciate the hard work you guys are putting in. For some of us, this project is the closest an average joe with ever get to hearing true state of the art!

The LS50 meta is indeed an incredibly well performing speaker, provided you don't want more controlled dispersion and can augment the bass appropriately. I'm tempted to get a pair, but I'm too addicted to DIY.

 

[muad]

That is an interesting question, for sure.
I would think that maintaining cardioid directivity down through frequencies that are traditionally problematic could only be a good thing (say, ~100Hz).

But engineering seldom admits "a good thing" without also introducing something "less good"

By reducing the rear wave in the bass, wouldn't that in theory result in more uneven bass?

What I mean is that the bass response at the sitting position is a result of the interaction of all bass sources and the distances to reflective surfaces. Would losing the low frequency rear wave reduce the amount of "averaging out (summing/cancellations etc)" that occurs in room?

 

[617]

By reducing the rear wave in the bass, wouldn't that in theory result in more uneven bass?

The modal region of a room is so chaotic that, in my opinion, it is very optimistic to assume that differing radiation patterns will help bass linearity. IIRC ported speakers are monopoles at some frequencies and dipoles at others - presumably they go through cardioid transitional region as well.

To me the clear benefit of cardioids are the greatly diminished side wall reflections, which are most important above above 1000hz or so. I can say that in my small room, no speaker has sounded more clear or present than a large (12") OS waveguide speaker.

 

[TimVG]

Directiva2's dispersion pattern looks incredible!

It's quite a little speaker isn't it?

I think we can mitigate diffraction a bit more by using a larger roundover (I only had an R6,5 roundover bit around, R13 would be better still) according to the sims, and by using more appropriate screws on the DXT (had to use what was around, again) and covering the ones in the waveguide with bits of tape, maybe @Rick Sykora will try one or more of these 'tweaks'.

 

[617]

Will just add that @617’s ”limited bass performance” is a matter of taste. Directiva r1 measures down to low 40s by Amir’s spin data. For many listening to music in apartments, condos and smaller rooms that is plenty. For low bass or HT special effects, few speakers are capable and you will need a subwoofer anyway.

Rick, that's a good point. What I meant was limited bass performance compared to a floorstanding speaker.

I think somewhat overlooked in the discussion of the purifi's outlandish displacement is the incredibly smooth and resonance free midrange.

 

[617]

It's quite a little speaker isn't it?

I think we can mitigate diffraction a bit more by using a larger roundover (I only had an R6,5 roundover bit around, R13 would be better still) according to the sims, and by using more appropriate screws on the DXT (had to use what was around, again) and covering the ones in the waveguide with bits of tape, maybe @Rick Sykora will try one or more of these 'tweaks'.

Tim I didn't mention it but please never mount a driver with drywall screws again...I will literally send you a bag of black oxide pan head screws if it means I don't have to see that again.

 

[abdo123]

It's quite a little speaker isn't it?

I think we can mitigate diffraction a bit more by using a larger roundover (I only had an R6,5 roundover bit around, R13 would be better still) according to the sims, and by using more appropriate screws on the DXT (had to use what was around, again) and covering the ones in the waveguide with bits of tape, maybe @Rick Sykora will try one or more of these 'tweaks'.

Would the crossover signficantly change if these tweaks are adopted?

In other words are you planning to surgically design the crossover?

 

[TimVG]

Would the crossover signficantly change if these tweaks are adopted?

In other words are you planning to surgically design the crossover?

I don't believe there will be significant changes to either the results or the crossover. There's also sample to sample variation to consider when doing these things, and sometimes these differences only show up on-axis. Meaning if you decide to compensate you'll end up with different sound power responses.

Tim I didn't mention it but please never mount a driver with drywall screws again...I will literally send you a bag of black oxide pan head screws if it means I don't have to see that again.

lol - I'm honestly surprised no one said anything until now.

 

[TimVG]

I can say that in my small room, no speaker has sounded more clear or present than a large (12") OS waveguide speaker.

Same with my build (avatar). Very curious to compare it to R2 (I can just take the horn off and use it as a bass module down to where the subs take over)

Very different on paper.

 

[Rick Sykora]

I don't believe there will be significant changes to either the results or the crossover. There's also sample to sample variation to consider when doing these things, and sometimes these differences only show up on-axis. Meaning if you decide to compensate, you'll end up with different sound power responses.


lol - I'm honestly surprised no one said anything until now.

It yours and a prototype. As long as the gasketing is good, no complaints here!

 

[617]

Same with my build (avatar). Very curious to compare it to R2 (I can just take the horn off and use it as a bass module down to where the subs take over)

Very different on paper.

View attachment 180000

View attachment 180001

It's actually interesting to see this because although they are somewhat different from each other, they have more in common with each other than they do with almost any other speaker. No trace of a crossover point, and both are mostly free of the tell-tale curvature of direct radiation drop off.

I think as I develop the presentation for this speaker I will make the distinction you made earlier in the thread - the goal is not cardioid response, but controlled directivity. Waveguides, cardioid and dipole radiation, driver beaming, wide and narrow baffles are all mechanisms to control directivity. This speaker uses more than one mechanism but blends them all together.

The main differences between different classes of directivity control mechanisms is whether they use cancellation (dipole, cardioid, cone beaming) or restriction (horn, waveguide, baffle) to narrow dispersion. Achieving CD through cancellation tends to burn a lot of driver xmax, but restriction makes efficiency better. It's sort of interesting that some speakers like the Gedlees rely heavily both on a restrictive mechanism (horn) and a cancellation (cone beaming).

 

[Rick Sykora]

It's quite a little speaker isn't it?

I think we can mitigate diffraction a bit more by using a larger roundover (I only had an R6,5 roundover bit around, R13 would be better still) according to the sims, and by using more appropriate screws on the DXT (had to use what was around, again) and covering the ones in the waveguide with bits of tape, maybe @Rick Sykora will try one or more of these 'tweaks'.

For those not on the team, Rx.y refers to versions of the r2 cabinet...

The roundover for simpler cabinet is spec'd at 13mm. Due to how close the tweeter is to the upper baffle edge, pretty sure that is all we can do. this was a tradeoff the I made rather to keep the woodworking simpler than a more rectangular cabinet with deeper beveling/roundovers.

For those who might be surprised by the number of versions, along with different midwoofers, gave waveguides a shot too. In the end, the optimal cabinet came down to 2 primary choices. The simpler one I will build and another more rectangular one with beveling. Both had comparable directivity, but the beveled one would require a much thicker front baffle. I did not like that option but left it as an option. Tim just got a CNC, so his is a unique variation of the simpler (house-shaped) cabinet.

 

[Lbstyling]

Yes, I had some understanding of this from a previous project. After looking into patents and AES papers, modifying and existing (cheap) speaker, ctrl was able to work on a simulation model, and measured results correlate very well.

love how you put this....like it was nothing....
even though you just laid out the hard work on a design with the potential to perform in the top 0.1% directivity of any speaker in history.

you should be very proud!

 

[TimVG]

Tim just got a CNC, so his in a unique variation of the simpler (house-shaped) cabinet.

..Of which I'll be happy to share the SVG files after the project finishes.

For those interested: you can have templates CNC or laser cut locally and simply use a jigsaw and a template (copy) bit on a standard table router. If you have an edge router, but not a table, it's super easy to mount it in one. Youtube will help in that regard.

love how you put this....like it was nothing....
even though you just laid out the hard work on a design with the potential to perform in the top 0.1% directivity of any speaker in history.

you should be very proud!

Thanks for your kind words! But it's really a team effort. @ctrl did a ton of work with the simulations which really improved the workflow and the rest of the team is working hard on the other aspects of R2

 

[HammerSandwich]

By reducing the rear wave in the bass, wouldn't that in theory result in more uneven bass?

This is difficult to compare, at best. You could look at Stereophile, as @John Atkinson has measured numerous speakers, including Kii & 8c.

Meanwhile, I pulled the 8361a & Kii from soundandrecording.de. The Kii is smoother in the ~80-500Hz range. It's not night & day, but it's there. Interestingly, Goertz uses less correction down to 35Hz - a full octave below the cardioid range.

Blue is raw response, green the correction, and red the EQed result.

 

[muad]

I'm just about to pick up a 2x4HD for this project, and just to have general crossover/dsp capabilities. I already have a stereo hypex amp and was going to buy another for the bass module. Is there any benefit to getting the hypex plate amps instead for the directiva?

 

[617]

A quick visualization of bass module sizes, assuming a desired 43" tweeter height (110 cm)

From left to right: 4x 15cm" drivers, 2x 17cm drivers, 1 x 23cm, 2 x 23 cm, 1 x 27cm, 1 x 12".

 

[D!sco]

Has a smaller diameter 3-way or 2.5-way been considered as an oversized bookshelf option, like some of the Hi-Vi/Swan or Edifier boxes? I also find the form factor difficult, like most multi-box speaker systems. It seems like there's plenty of space behind the midbass' cardioid chamber. In a lot of ways, the monitor's thin form factor is really appealing until the lack of bass is taken into account. Maybe when there's more information I'll try coming up with a deeper monolithic enclosure in sketchup.

Are these bass modules going to be sealed, ported, dipole, combination? Definitely a fan of the first two front baffles.

 

[ctrl]

It seems like there's plenty of space behind the midbass' cardioid chamber. In a lot of ways, the monitor's thin form factor is really appealing until the lack of bass is taken into account. Maybe when there's more information I'll try coming up with a deeper monolithic enclosure in sketchup.

The depth of the cabinet does not play a decisive role in the Directiva r2, but still contributes to the radiation pattern.

If you really plan to build a bookshelf speaker with a significantly greater depth, the radiation pattern will change somewhat.
When the time comes you can contact me, then I can simulate for you how big the effects will be compared to the original r2.

Update: Of course only if your design is open source again