ASR Directiva Open Source Speaker Review

[voodooless]

I'm not sure I quite follow, wouldn't larger (yet along multiple) woofers increase the Air volume requirement of the cabinet considerably making it already significantly larger?

What do you think the extra Purifies would do? But you don't really need a lot of volume for the Daytons, you can always DSP them.

In any case, did anyone per change revisit the Kef R3 review? Seems like distortion levels are really comparable. Let me borrow Erins plot:

That is below -45dB above 100Hz, which is below 0.18%

Sure it will have much better bass extension and low-end SPL, but midrange performance is on par with the Kef which is the same price fully finished and build, only one stereo amp is needed. But obviously, there is no fun in that

Edit: as @Sancus noted, I was about 10 dB off, so the Kef is not on par with the Purify, but it's not that far off either. Still quite impressive for what you get for the money though.

 

[abdo123]

What do you think the extra Purifies would do?

it would be much easier to add more Purifi woofers and make it a floor stander than to make it a 3-way though.

 

[voodooless]

it would be much easier to add more Purifi woofers and make it a floor stander than to make it a 3-way though.

Yes, you can make an MTM 2-way, but it really stops there. Add any more, and you'll need more "ways". And it won't go any lower either, just a bit louder. if that is your goal, it might suit you well.

 

[abdo123]

And it won't go any lower either,

Why not? wouldn't you get a 3dB boost with each sequential driver? Basically the F13 becomes the new F10 if you massage it into the EQ.

 

[voodooless]

Why not? wouldn't you get a 3dB boost with each sequential driver? Basically the F13 becomes the new F10 if you massage it into the EQ.

No, because everything becomes louder equally, F3 remains F3, F10 remains F10. MaxSPl is a different story. Above the tuning, you will gain 6dB in MaxSPL. Below the tuning, however, you will only gain a few Hz of extension because of the steepness in drop-off of the MaxSPL curve. It's really not worth the extra cost if you ask me. As long as your goal is to keep above 35 Hz, you should be fine with your very expensive MTM.

Btw, I noticed some people comparing speakers low end and only looking at the F3 number. That can be quite deceiving. F10 is a much more valuable number. Note that for something to sound half as loud, you need a 10 dB drop. And half as loud is usually still a very significant volume. So you can have a system that has a higher F3, but a lower F10 than another. Guess which one will appear to have more low-end extension?

 

[aac]

650 Hz actually is the crossover frequency but the biggest advantage is the lower IMD above at the more critical and audible mids.

You are right, I thought it was 800 for kh310 and 650 gor kh420 but its 650 and 570.
As for IMD there is a point.

 

[Koeitje]

Or add a 3rd driver to handle 20-300Hz?

Then you drop the small size requirement. Like he said: pick two.

 

[Sancus]

That is below -45dB above 100Hz, which is below 0.18%

-45dB is ~0.56%, -50dB is 0.32% so nope, this does still beat the R3, and by even more for 3rd+ harmonics which are the more important ones. But yes the R3's woofer does perform very well.

 

[ctrl]

Cool to hear you say you can't design by measurements alone. These already measure as some of the best 2 way speakers we've seen.

If Rick hears a change that he rates subjectively higher, but measures worse, how do you plan to balance that?

Then it would be Ricks decision.
But I don't expect Rick to suddenly want to put a 2dB bump in the axis frequency response at an arbitrary point when fine tuning.

In the end, however, the listening impression should always be decisive.

As long as it is only about the active version of the Directiva, the interaction of measurements and listening impression does not play such a decisive role, since the crossover can be changed and adjusted by anyone.

@ctrl @Rick Sykora how hard was it to get NFS measurements into VCad? Any major issues there?

No, not at all. @kimmosto, the developer of VCAD, has enabled direct import of exported NFS measurement files.

Would also like to thank @kimmosto here for giving us helpful tips regarding the use of VCAD.

By the way, there is also a crossover proposal from him in the "Introducing Directiva" thread (still needs to be adapted from miniDSP 48k to 96k, but no problem to do so), which shows the great advantage of an active Directiva version - you can compare virtually any number of configurations directly in a listening test and decide for yourself which crossover is best for you.

 

[voodooless]

-45dB is ~0.56%, -50dB is 0.32% so nope, this does still beat the R3, and by even more for 3rd+ harmonics which are the more important ones. But yes the R3's woofer does perform very well.

What have I been smoking then? Because when I put in the numbers something else came out. You are absolutely right

 

[mohragk]

Yes, but I guess it's not the point of such projects. It rather goes to show what can be done with minimal resources.

Minimal resources... I know that buying a Klippel system would be peanuts for the KEFs out there, but it's still quite pricey. Then again, you can rent one I imagine.

 

[restorer-john]

I know that buying a Klippel system would be peanuts for the KEFs out there

I don't believe a Klippel is reagrded as "peanuts" for any company, letalone a medium sized speaker manufacturer, or, for that matter, an individual.

Sure, in comparison to board machinery (CNCs, vacuum presses, laminating lines, paint lines) where a million dollars is par for the course, but consider these guys have already invested and have most likely amortised all their factory machinery decades ago. A Klippel would be a tough sell IMO. After all, the board will simply want a straight up ROI. What is the answer?

 

[voodooless]

What is the answer?

Well, they already seem to have an anechoic chamber, so the need for a Kippel analyser will really be a hard sell at this point. What they already have will get them already to up to 90% of Kippel performance. On the other hand, the thing is 50K. If you can save one or even a half staffer by having the system present, you are already cost-neutral after just a single year.

 

[abdo123]

I don't believe a Klippel is reagrded as "peanuts" for any company, letalone a medium sized speaker manufacturer, or, for that matter, an individual.

Sure, in comparison to board machinery (CNCs, vacuum presses, laminating lines, paint lines) where a million dollars is par for the course, but consider these guys have already invested and have most likely amortised all their factory machinery decades ago. A Klippel would be a tough sell IMO. After all, the board will simply want a straight up ROI. What is the answer?

I think the biggest sell-point of the Klippel is both automating and standardizing measurements, designers would not have to argue between themselves whether they're measuring 'correctly' or not, their designs would translate to every anechoic chamber and every measuring setup out there.

Ofcourse there is still a tiny human factor in it, mainly in setting up the 'on-axis' at the begining of a spin and stuff, which is why even with the Klippel we have a 'directivity error' because the way the drivers were measured intially deviates slightly from the way the entire speaker was measured for the review.

 

[voodooless]

While Klippel works great for bookshelf size speakers it has considerable limitations when measuring larger floor standing models, as shown with the Revel F328Be. Even if Klippel became a standard speaker manufacturer tool there would need to be recognized alternate setups for capturing data on larger floor standing speakers. When Klippel measurements show bass rolloff starting at 70Hz and the F328Be offers strong response down to 29Hz with REW measurements you know there's a glitch in the data capture.

You'd probably want to read the second section in this post of mine. F10 in those Revel measurements is something like 30 Hz. Add some room gain, and you'll have a speaker that is just about flat down to 30 Hz in-room. The measurement isn't wrong, your interpretation is.

 

[Tangband]

I see the JBL 530's are there as well.
How do these two sets compare, your DIY and the JBL's?

Yes, the JBL 530 passive are good speakers, but their performance can be raised slightly by making them active with a slightly higher crossover at 1,8 kHz .
My DIY loudspeaker HYBRID is sounding slightly better though, except in the bass where they drop off 12 dB/oct from about 80 Hz. So I need subwoofers with them.

The Genelec 8340 is slightly better still, with a little less distortion, more details in the sound, goes deeper in the bass ( flat to about 38 Hz in real life ) and can play louder. The resonanses in the Genelec cabinet is twice as high in frequency than HYBRID. That itself makes the sound a bit clearer, especially when listening to piano music.
The metal cabinet Genelec uses is a better choice in my opinion, than mdf - soundwise.

To be on the topic again, and a suggestion:

- with an extra purifi bass unit on a floorstander, one can do the crossover for the 2,5 way construction att the baffle step frequency, at about 500-700 Hz , using a simple passive aircoil in series with + for the extra purifi bassdriver. Thus one can use a simple minidsp , and two amplifiers when driven active, because one of the crossovers can be done with a passive coil. You dont need a threeway dsp crossover or three power amps in this example.

 

[beagleman]

Lol. No tangerine wave guide, instead we get fusilli screen. $240 each driver. Though I chuckled at the review by some dude who cannibalized the amp/dsp/etc from an ikea/Sonos speaker to use with it.

Right now The KEF 350 are listed fir $750 a pair. So the drivers alone would be 2/3 the cost of a finished pair. maybe one can DIY make a better measuring speaker that the 350 for the remaining $270 for hardware, finish, cabinets and crossover (if passive) and we don’t count labor). Those KEF engineers seem to earn their pay though.

All I am saying is that the Directiva is amazing and impressive and a feat at $1k and as others have noted bests many much more costly designs. The excellence speaks for itself.

No need to value engineer it. Indeed, not only is it good value, it makes me appreciate more the value presented by some manufactured speakers. The directiva demonstrates that DIY can now be about high-end quality and value.

It might surprise you to know, the drivers are only a tiny percent of the overall cost. The box, labor to construct it and so on, PROFIT and advertising Shipping and so on, are most of the cost really. The drivers are most likely well under $100.00 I would guess.

 

[Phorize]

I think this would be an excellent idea. While an excellent woofer, the price is quite exorbitant for a 6.5" monstrosity.

Monstrous, maybe, but what conveys audiophile 1337ness more than a woofer that looks like it was designed by H R Geiger?

 

[PeteL]

I think the biggest sell-point of the Klippel is both automating and standardizing measurements, designers would not have to argue between themselves whether they're measuring 'correctly' or not, their designs would translate to every anechoic chamber and every measuring setup out there.

Ofcourse there is still a tiny human factor in it, mainly in setting up the 'on-axis' at the begining of a spin and stuff, which is why even with the Klippel we have a 'directivity error' because the way the drivers were measured intially deviates slightly from the way the entire speaker was measured for the review.

I is still not clear tough from the review how Klippel system was used in the design of this speaker, Is there ways to extract Data and apply it to Mini DSP for exemple. In the end isn't directivity management mainly a product of component choice more than actual DSP?

 

[abdo123]

I is still not clear tough from the review how Klippel system was used in the design of this speaker, Is there ways to extract Data and apply it to Mini DSP for exemple. In the end isn't directivity management mainly a product of component choice more than actual DSP?

Directivity management is a product of both the components and crossover frequency, the later (from what i understood) has been influenced by small deviations in the first and final measurements parameters.

Basically at the very beginning the enclosure with the drivers inside was sent to Amir and the drivers were measured individually, put into VCAD and then the crossover was chosen based on that. The final product was measured (allegedly) slightly differently causing the 'directivity mismatch'.