• 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!

ASR Directiva Open Source Speaker Review

Didn't say final. I said filter design. Its open source, so waiting to see what it is at all? What we see is the summed response of two drivers with DSP, but what EQs and Xover slopes have been used to get here?

I mean with a score of 6.0 i doubt a traditional 2-way speaker can go any higher than that.
Yep, the Purifi driver makes the speaker an expensive DIY project.
However, the performance in terms of harmonic distortion (HD), intermodulation distortion (IMD) and displacement volume (DV, i.e. possible achievable sound pressure level in the low bass range) of the Purifi driver are exceptionally good.

With another woofer from SBAcoustics, for example, one could achieve a similar result at a much lower price, with a few sacrifices in HD, IMD and DV.




When you look at the measurements like this, it's almost boring.

On the other hand, we have a tiny speaker that, according to Amir's measurement, achieves an f3 of around 40Hz ( without DSP expansion) and can also produce reasonably low bass sound pressure levels due to the Xmax of +-10mm.
View attachment 157094

A cheap 6 or 6.5'' driver has a typical Xmax of about +-5mm. The displacement volume of the Purifi driver therefore corresponds to that of about two inexpensive 6.5'' drivers.

The quality of the drivers used is also reflected in the DSP settings. Very few corrections have to be made to achieve a good result.
View attachment 157087





As Amir said, the increased harmonic distortion around 400Hz was caused by the measurement setup.
In this frequency range, the Purifi driver and the passive resonator show no abnormalities.

The slightly increased HD3 (see 1. in the graph) around 9kHz and HD2 (see 2.) around 13.5kHz is due to the violent (damned translation programs ;)) breakup of the tweeter around 27kHz.
View attachment 157096

In Amir's measurement the 27kHz break-up is just visible due to the lowpass filter of the measurement system. In reality, this is much more pronounced and the cause of the (inaudible) increased HD values.
Manufacturer measurement of the Seas-DXT tweeter with break-up resonance at 27kHz:
View attachment 157098


For those interested, I have attached the complete VCAD project that corresponds to the version of Directiva measured by Amir.

the 'directivity mismatch' doesn't appear as severe in your data set, is it just different axis limits on the graphs or 'because' of the uber high resolution of the NFS?
 
Really nice concept.
The combination of DXT -Tweeter with max. 6“-7“ midwoofer and the right choose of the baffle design has outstanding directivity behavior.
You should have a look at the speaker concept of Alexander Heissmann https://heissmann-acoustics.de/en/dxt-mon-182/ . He has no Klippel NFS, but in the past several years he optimized his speaker design and scaled his design concept from 5“ - 7“ midwoofer.
Somebody should build the speaker and send it to Amirm. The DXT-MON 182 passive build is a lot cheaper than the ASR Directiva. A comparison would be really interesting.
 
I mean with a score of 6.0 i doubt a traditional 2-way speaker can go any higher than that.


the 'directivity mismatch' doesn't appear as severe in your data set, is it just different axis limits on the graphs or 'because' of the uber high resolution of the NFS?
Ah yes, correct on that front. I was suprised to see the crossover is 2750.

I think most of the magic here is the DXT. Pretty much any woofer of similar size could be used. (DXTMon I'm lookin at you)
 
@amirm should there be something here that is missing for some reason?

1633332860951.png
 
The performance is there and the work is certainly admirable -- but the value just doesn't stack up to what "the studio monitor establishment" has come up recently, e.g. the $800/pair 3-way Kali Audio "2nd Wave" IN-8 is a complete solution, complete with built-in amplification and DSP.
 
Would love to see future iteration with 2 or 4 Purifi woofers with a linear response all the way to 20Hz with front ports / front passive radiator.
2 woofers and two of those PR's will make for one massive baffle, let alone 4 of them. it might work with reflex, but 20 Hz will need a massive port and will bring a lot of downsides on its own. Never mind the cost. If you want deep bass, I'm pretty sure there will be more cost-efficient offerings out there for such a cabinet.
 
I feel the JBL A130 is one example. Very good above 55hrz, quite low in distortion (and subjectively I though sound excellent) and on sale a pair of complete speakers costs less than 1/2 of one bare purifi woofer. I'd say if you are going to use a sub/subs then the purifi is really hard to value. Though compression is not measured here.
The A130 can be Eq'd to a 6+ harman score with 8+ using a sub as well.

(not saying I am not impressed with this speaker - I am, just not as sure about the woofers due to price)
I was thinking the same example, particularly with the uniquely low upper bass distortion and tunable response. The A130 still seems like one of the best values of any passive speaker on ASR.
Though, while the A130's FR is probably fairly comparable once EQed, the distortion and directivity would still remain a little weaker.
ASR Directiva Measurement Distortion Open Source Purifi Speaker.png

JBL Stage A130 Measurements Relative THD Distortion.png
ASR Directiva Measurement horizontal directiva Open Source Purifi Speaker.png
JBL Stage A130 Measurements horizontal directivity.png

Obviously, it's a bit apples to oranges with the price and size difference. But, once the spikes in the Directiva's distortion are gone, I think the difference would be more palatable. The spikes in the graph have no relation to anything and vanish at 86dB, a pattern typically only coinciding with active DSP speakers. I'm almost sure it's that darn MiniDSP 2x4 HD. MiniDSP really needs to update that device. It had subpar performance when it came out, and it's much more drastically lacking today.
 
Last edited:
2 woofers and two of those PR's will make for one massive baffle, let alone 4 of them. it might work with reflex, but 20 Hz will need a massive port and will bring a lot of downsides on its own. Never mind the cost. If you want deep bass, I'm pretty sure there will be more cost-efficient offerings out there for such a cabinet.

Well @Rick Sykora said they're interested in making a bigger directiva, so i thought that implied more drivers. and you need them in pairs if you want to maintain decent vertical directivity.
 
Not sure the design and tooling for that Purify driver could in any way be described as "minimal resources"
I think so, at least compared to the resources of the "big boys in loudspeakers", B&W, JBL, ...
Drivers are important, as are electronics, but there are other factors, cabinet design for instance.
 
Why would you say so? I can think of no plausible scenario in which the MiniDSP could cause resonances (unless clipping is involved which is a user error).
Well, speaker hiss is just high frequency distortion, and the 2x4 HD has a lot of that. Would explain the distortion peek around 15 kHz or so.
Amir literaly had the the minidsp physically sitting on top of the speaker for the measurements. It rattled. Shows up in measurements
Ohh, I didn't catch that part. That would explain the mid range resonance and why it shot up so much at 96dB.
 
I'm not sure what to make of it however. Has it been EQ'd in DSP within an inch of its life and if so, why the obvious hole in the response, at the crossover point, dead in the most important part of the midrange?
The dip in the sound power frequency response is mainly caused by the unavoidable vertical cancellation of the tweeter and woofer frequency responses.
The crossover frequency was deliberately chosen so that this dip lies precisely in the frequency range most sensitive to humans (ear channel resonance).

If the overall sound of the loudspeaker seems too "soft" in this frequency range, the dip in the axial frequency response of the tweeter caused by the edge diffraction, for example, can easily be filled by means of a PEQ, according to individual listening taste.
1633334007805.png

If you like to listen louder and jazz trumpet or metal guitars are present in the music repertoire, you might will already have problems with the long-term listening experience when listening to the speakers with this "optimized" setting.


I thought this was the final filter design?
Didn't say final. I said filter design. Its open source, so waiting to see what it is at all? What we see is the summed response of two drivers with DSP, but what EQs and Xover slopes have been used to get here?
I linked the entire VCAD project in post#38.

This will probably not be the absolute final version yet. Once @Rick Sykora has had a chance to listen to both speakers in stereo for a while, there will certainly be some fine tuning.

And since the project is open-source, in the course of time very likely "better sounding" crossover versions will be published by other users.


The combination of DXT -Tweeter with max. 6“-7“ midwoofer and the right choose of the baffle design has outstanding directivity behavior.
You should have a look at the speaker concept of Alexander Heissmann https://heissmann-acoustics.de/en/dxt-mon-182/ . He has no Klippel NFS, but in the past several years he optimized his speaker design and scaled his design concept from 5“ - 7“ midwoofer.
Somebody should build the speaker and send it to Amirm. The DXT-MON 182 passive build is a lot cheaper than the ASR Directiva. A comparison would be really interesting.
Yep, Alexander is an excellent speaker designer, with fair prices.

Due to the special baffle design, the directivity in the frequency range 4-10kHz should be slightly better with the DXT-Mon-182.

The challenge with the Directiva, was to achieve the best possible result with an off-the-shelf loudspeaker cabinet, in order to make it as easy as possible to rebuild.
As a result, however, slight sacrifices in the directivity of the speaker have to be accepted.


the 'directivity mismatch' doesn't appear as severe in your data set, is it just different axis limits on the graphs or 'because' of the uber high resolution of the NFS?
Since it was Amir's first time taking measurements for the development of a loudspeaker and those involved in the development were spread across three time zones, communication was not always easy.
The deviations are likely to be due to possibly less than optimal measurement data on which the simulation is based or small deviations in the miniDSP filter settings compared to VCAD.
But we all know, the first time is never perfect.


Ah yes, correct on that front. I was suprised to see the crossover is 2750.
The crossover frequency is around 2.5kHz, due to the edge diffraction the tweeter on axis has a dip in this frequency range and deviates from the target filter slope.
1633336200777.png
 
Well, speaker hiss is just high frequency distortion, and the 2x4 HD has a lot of that. Would explain the distortion peek around 15 kHz or so.
Incorrect. The miniDSP is not at fault here*. The 13khz distortion is the tweeters 26khz resonance peak folding down into the measurement range. It would not be audible.

*apart from where it rattles at 380hz.
 
Since it was Amir's first time taking measurements for the development of a loudspeaker and those involved in the development were spread across three time zones, communication was not always easy.
The deviations are likely to be due to possibly less than optimal measurement data on which the simulation is based or small deviations in the miniDSP filter settings compared to VCAD.
But we all know, the first time is never perfect.
Do you think that could be corrected in the future with the crossover design? or did that ship sail when the cabinet was built?
 
Do you think that could be corrected in the future with the crossover design? or did that ship sail when the cabinet was built?
One would have to systematically search for the reason for the deviation.
First, the two drivers would have to be completely re-measured under precisely defined conditions and if the deviations should persist afterwards, the implementation of the filters in VCAD versus miniDSP would have to be examined.

Since it fits perfectly in the low frequency range, anyone with measuring equipment can actually repeat the measurements - of course not with the resolution of an NFS.

But it is also possible to work quite well with the existing data.
 
Back
Top Bottom