Introducing Directiva - An ASR open source platform speaker project

[RickS]

I'm a DIY Audio newbie who's been toying with VituixCAD to slam drivers together like a child would attempt to mate transformers action figures. Do we have enough data on these drivers to provide usable FRD/ZMA files? I'd like to play with passive crossover values and see what's possible.

Welcome to ASR!

There are some earlier versions of the VCAD project files in the general thread, but would suggest you wait. In a week, plan to post a project based on Klippel measurements (rather than simulation).

 

[notabenem]

@Rick Sykora Is it possible to already tell if the distortion of the Directiva at loud volumes (>85dB) is going to be better, or comparable to that of the ELAC DBR 62 ?

 

[RickS]

@Rick Sykora Is it possible to already tell if the distortion of the Directiva at loud volumes (>85dB) is going to be better, or comparable to that of the ELAC DBR 62 ?

Have every reason to believe that Directiva r1 will have much lower distortion than the ELAC. Low distortion (notably IM) was the primary goal of the Purifi woofer design. To be fair, a pair of Purifi woofers cost more than a pair of finished ELACs though.

As for THD, how much lower Directiva is will be more accurately compared once Amir tests.

 

[ctrl]

@Rick Sykora Is it possible to already tell if the distortion of the Directiva at loud volumes (>85dB) is going to be better, or comparable to that of the ELAC DBR 62 ?

The harmonic distortions should be much better. The Directiva should only show two small humps in the harmonic distortions in two places (apart from the low bass), HD3@9kHz and [email protected]. Both small humps reflect the 27kHz resonance of the SeasDXT and are inaudible.

Up to 2kHz, the harmonic distortions of the Directiva should match with Purezza which @hardisj measured here.

 

[Selah Audio]

Now that's the thing, you make it sound as if the extra woofer output is coming out of nowhere, but in reality, you're just doing the same as padding the tweeter but with extra steps .

That's why me & @ctrl got confused.

The problem is increasing the gain that much on the woofer and then adding bass EQ. The single passive radiator will limit the output and you risk overdriving the woofer and/or clipping your amplifier.

 

[ctrl]

The problem is increasing the gain that much on the woofer and then adding bass EQ. The single passive radiator will limit the output and you risk overdriving the woofer and/or clipping your amplifier.

But the fascinating thing about the Purifi driver (with +-10mm lin. excursion) is that it reproduces low bass like some speakers twice the size and with an active concept or passive + DSP, this advantage can be fully exploited.

With linear/flat crossover tuning, f3 is below 40Hz. If a second order high pass at 30Hz protects the driver and the PR, f3 is still around amazing 41Hz.
In the range of linear excursion of driver and PR are sound pressure levels in the bass range down to 40Hz, are realistic up to 97dB.
Until the voice coil of the Purifi driver hits the back plate probably significantly higher sound pressure.

Simulated FR and excursion of the Directiva LS at 97dB (free field, 4pi - see brown curve) with HP-BU2@30Hz:

 

[tktran303]

@ctrl

using a subsonic filter may be useful if running this thing full range, at full steam. Now of course that applies to any midwoofer.

but one needs to know exactly how many volts you will dumping into system to set the limiters and subsonic filters optimally.

Will your design incorporate a known amplifier?
Or just the active equalisation part?

will the design have different amounts of baffle step compensation or settings for different placement options?

I think that is the most useful part of active EQ;
in getting the right tonal balance becasuse you never know where/how Amir or people will place it, then and judge it when listening in mono.

I mean it’s a certainly unfair to put the single speaker with 4” midwoofer in a large room and give it a headless panther. I don’t think the designer or buyer ever planned to do that.

In my own my passive design, I have in-room output to 32Hz. I designed it for my current listening room, which is “only” 30m^2 room; with speakers within 2 foot on front wall. I’d guess that most people place their speaker within 18” of the front wall, but then some people may put a speaker on their PC desk thanks to working from home arrangement, or right up against the wall due to aesthetic/space considerations, or even 5 feet away from the front wall in their dedicated multi seat home theatre, and then declare that active EQ for the low end is needed.

these are not philosophical arguments; these are practical considerations for equalisation/crossover tuning for the designer.

When people measure or audition speakers differently form the designer’s expected use case, the wheels can fall off…

I think this is more important than the use of limiters or subsonic filter to be able to push a driver to its limits- the limit I have is how much SPL I (or the family) can stand.

 

[abdo123]

I would love if the design was adapted so the hypex fusion plate amps are used as an all-in-one solution.

Would probably be one of the best speakers in the world for the price.

 

[RickS]

I would love if the design was adapted so the hypex fusion plate amps are used as an all-in-one solution.

Would probably be one of the best speakers in the world for the price.

Should be able to make this happen as @ctrl has experience with the Fusion amps.

Other than the time, the only other hurdle would be loaning or funding a purchase of an amp.

 

[ctrl]

but one needs to know exactly how many volts you will dumping into system to set the limiters and subsonic filters optimally.
Will your design incorporate a known amplifier?
Or just the active equalisation part?

Sure, it would be optimal to control the filter in real time via permanent voltage monitoring.

The not so good solution is to use a "fixed" subsonic filter. This is in any case much better than to use none, if you want to hear louder.

In Post#576 the simulation with a subsonic filter with BU2@30Hz at 97dB sound pressure level in the free field shows that the excursion of driver and PR just leaves the range of linear excursion - without really worsening the low frequency reproduction.

For comparison, let's see what theoretically happens if you set the sound pressure level in the low frequencies to only 93dB instead of 97dB without the subsonic filter.

Simulated FR and excursion of the Directiva LS at 93dB (free field, 4pi - see brown curve) without protection high pass filter:

Below 25Hz the woofer clearly leaves the linear excursion range and if the music signal still contains frequencies below 20Hz (which happens with electronic music and film), then there is already the danger that the voice coil hits the back plate.

The subsonic filter was chosen by me for the example in Post#576 in such a way that the low frequency potential of the Directiva is almost not limited, but the possible sound pressure level is significantly increased and the group delay is not increased too much.

Directiva FR with subsonic filter BU2@30Hz and without filter.


If you are willing to accept a little more group delay, you can protect the Purifi woofer much better, so that if the sound pressure level is too high, it is very likely that only the PR will have problems.

If a fourth-order high-pass filter is set, the sound pressure level in the range around 50Hz is even slightly increased compared to the "normal" version, with significantly less excursion of the woofer.


There are an "infinite" number of ways to customize it.

When @Rick Sykora has created a first final version, the whole VCAD project will be made available here in the forum and everybody can simulate for himself how and where one sets filters or not.

The VCAD project contains @amirm's NFS measurements of the individual drivers and probably a few examples of possible active crossover tuning.
This opens up the possibility of crossover tuning for everyone with a professional level dataset.

Later, there will also be impedance measurements of the individual drivers, which will then enable a passive version.

UPDATE: However, it should be noted that 93dB@1m sound pressure level at 50Hz when using two speakers and some reinforcement from the listening room's boundary surfaces, is already quite loud.

will the design have different amounts of baffle step compensation or settings for different placement options?

With the published VCAD project it is easily possible to make reliable individual adjustments.

I would love if the design was adapted so the hypex fusion plate amps are used as an all-in-one solution.

Should be able to make this happen as @ctrl has experience with the Fusion amps.

VCAD's simulations for the FusionAmp modules are pretty good. It should be no problem to transfer a crossover version for the miniDSP to the FusionAmp DSP.

 

[Selah Audio]

But the fascinating thing about the Purifi driver (with +-10mm lin. excursion) is that it reproduces low bass like some speakers twice the size and with an active concept or passive + DSP, this advantage can be fully exploited.

With linear/flat crossover tuning, f3 is below 40Hz. If a second order high pass at 30Hz protects the driver and the PR, f3 is still around amazing 41Hz.
In the range of linear excursion of driver and PR are sound pressure levels in the bass range down to 40Hz, are realistic up to 97dB.
Until the voice coil of the Purifi driver hits the back plate probably significantly higher sound pressure.

Simulated FR and excursion of the Directiva LS at 97dB (free field, 4pi - see brown curve) with HP-BU2@30Hz:
View attachment 138529 View attachment 138528

Well I think once there are a few out in the field for testing it will become more clear as to what settings may be unwise to use.

 

[RickS]

Well got a solid day in yesterday courtesy of our latest heatwave. Am testing at least 3 crossover candidates. The most recent ones are simpler and so I decided to cross check the correlation between the sim and my measurements. While not bad, was not even close to what we had managed earlier in the project.

Since our goal is to produce a VituixCAD project that would allow others to simulate the Directiva design without having the real speaker, we are syncing up to find where the correlation got broke. Versions of the software have changed, we added Amir’s measurements and I have made some hardware changes.

So, please be patient while we work to resolve this issue, but do not plan to post the project until we do.

Thanks!

Rick and @ctrl

 

[RickS]

fyi...

This was meant for an internal conversation, but accidentally posted to the thread too, so will use as a basis for a progress update.
team
First, I found this issue as I had zoomed in on a frequency response comparison between VituixCAD and my REW measurements. I was more comfortable with the deviations with the larger scaling, but now felt we need to do better. One change was the driver measurments and so we compared the original measurements for the speaker Amir has to the one I am using. The DXT tweeters were very well matched and the woofers were close. The woofer comparison is what is posted above.

So, looking for larger differences, here is what we have found so far…

1. The shelving filters did not match well. VCAD uses a 6 dB shelf filter by default. For it to match the comparable one in the minidsp, the VCAD dsp setting needs to be minidsp 48kHz and the corresponding minidsp filter needs to have a Q of 0.5.
2. Found some more reflections in my test rig that caused response rippling. This appears to be mainly due to the clothespin style mic clip and has been significantly improved by wrapping a microfiber towel around it. Have ordered another more streamlined clip to try.
3. We are currently trying to reconcile my measurements and Amir’s. some of this was the mic clip, but may not be all. As mentioned, there is some driver differences and he has a higher resolution setup. But before we send a crossover to him, we want to know we can expect very comparable results to the sim. This is likely to require validation testing at his end.

Have some appointments today, but stormy weather should allow me to spend some more time improving the sim correlation.

 

[RickS]

Decided to wait on the new mic clip before I remeasure. I thought I had ordered, but must have gotten distracted. The choices on Amazon lacked dimensional info so could confirm my UMIK-1 would fit. For that matter, this is plastered on every clip offering...

Consumer Alert

Most users do not need a license to operate this wireless microphone system. Nevertheless, operating this microphone system without a license is subject to certain restrictions: the system may not cause harmful interference; it must operate at a low power level; and it has no protection from interference received from any other device. Purchasers should also be aware that the FCC is currently evaluating use of wireless microphone systems, and these rules are subject to change. For more information, call the FCC at 1-888- CALL-FCC (TTY: 1-888-TELL-FCC) or visit the FCC’s wireless microphone website at https://www.fcc.gov/cgb/wirelessmicrophones.

I did not notice this Alert earlier in the week, but seems rather strange to post it with mic clips. Seems more apropos with electronics. For someone needing a clip for a wired setup not sure why I need to be alerted about wireless use of a mechanical part! Guessing they had some legal trouble...

 

[RickS]

New mic clip arrived, but was a little big for my mics, Fortunately, nothing a little tie wrap could not fix! This was not the major holdup on this project but is part of getting back to aligning the sim and measurements. The major holdup is outdoor projects and some paying work. Both are winding down and will get back to Directive soon.

 

[RickS]

Am planning to get back to Directiva today. I got another amp to rebuild over the weekend. It is a VTV Purifi EVAL-1. One benefit of doing this work is the chance to evaluate other products for use with Directiva. For example, had wondered whether the better SINAD of the Purifi EVAL-1 would reduce the low level hiss I hear when no music is playing. Today, I got that answer. It is NO.

It is much quieter here in the early morning hours, so gave me a chance to do a quick A/B compare between my NC252MP and the Purifi EVAL-1. As stated, no audible diff in hiss. As my QA401 audio analyzer does not measure any major difference between the two, am not surprised. This is not to say that the Purifi is not better, just that I have not encountered a condition in which it is audible. So any improvements in regard to hiss will have to come from different front-end electronics (or possibly gain manipulation).

 

[abdo123]

Today, I got that answer. It is NO.

Yeah the miniDSP has a SNR of ~107 dB so it's the limiting factor.

 

[RickS]

Ok, can now demonstrate high correlation between VituixCAD sims and my own measurements! @ctrl has been working on more crossovers so am evaluating them this week. We will use his for Amir’s testing, but as mine is not bad and simpler, plan to share both.

Hoping to wrap up and send to Amir by end of week!

 

[RickS]

First listening round went to @ctrl crossover. However as I compared the on-axis performance, I realized I had let mine droop much more over 14k Hz. One thing lead to another and was flattening it more through the midrange too. Still working out some kinks in the sim too. Ran out of time to listen critically so will resume in the morning and share some measurements.

My crossover is still a pretty simple and uses LR24 filters and his latest is a much more complex and uses 1st order Butterworth filters. As you might guess, on-axis frequency response is pretty comparable, but phase is a whole different story. I look forward to some more listening and sharing the results.

 

[RickS]

Hi Directiva fans!

My second round of listening was more subtle once I matched the high frequency responses of the two crossovers. Btw, @ctrl has already given me another 2 crossovers and am now using one with 2nd order LR for my listening tests. At this point, I decided to share some data with @ctrl. He reported back that I still had some waviness in my measurements. Both of us were getting frustrated by the distraction and so started ripping into the test rig again.

After trying different positions vertically and horizontally, went back and looked at the impulse response. I should have done this when I went to the new mic clip. The IR indicated a slight reflection around 2 ms. My typical gate is 3 ms, so this needed further investigation. After some more iterations, I decided to reorient the mic boom. As the primary space is our family room, do not have much actual floor space, so the boom was parallel to the front baffle. After removing a table and pushing a couch back, the boom is now perpendicular to the baffle. This smoothed the measurements more. As did pushing the mic about a centimeter more into the mic clip.

At this point, I needed to reestablish my baseline measurements. I need to do this before I can share pics of the results with you. Last night was a nice dinner outing at a winery, so back at it today. I will share that I left the music playing when I had to leave the room. While I was in our upstairs office, In The Air Tonight by Phil Collins started to play. Was pretty amazed by how loud this one small speaker was able to play! I look forward to more of this kind of excitement.