Directiva R1 Sealed Build

[canukhd]

This is description of the build of a Directiva speaker system I started last spring. I used the drivers and enclosure from the published Directiva build but deviated on the design of the crossover.

The system was intended for both home cinema and two channel music playback and will be used with four sealed twelve inch subs. I am using a Denon 4400h receiver with Apple TV and an Nvidia Shield as the source for video. For music I am using Roon, Tidal and Spotify on an iPad with airplay (and spotify connect) to send the audio to the receiver.
I use the receiver pre-outs to drive left, right and center Directiva speakers using two MinidspHDs for the crossovers (one for L &R and one for the C).

Amplification for the left and right channels uses an Adcom GFA-535 60 watt amp for the tweeter and a home built Hypex UCD180 for the Purifi woofers. The center Directiva uses a single stereo Adcom GFA-535 to drive both the Tweeters and the Woofers. The subs are two 12” Shiva in a sealed enclosure with 300 watt plate amps and two 12” CSS SDX 12 woofers in two sealed subs driven by an external Behringer NX3000 amp.

The speakers are painted with black duretex so that they are not visible when watching a movie. The placement of the left and right is unfortunately very close to the corners and the center is on the floor which is also not optimal. I use Audyssey MultEQ-X on PC to equalize the left and right to 500hz. I equalize the center to the full 20K. The subs are connected to a minidsp and equalized to be flat to 25Hz in room.

Here are a few shots of the Directiva build:

I used the parts express knock down cabinet kit and covered the interior with Kilmatt pads to add some extra mass to the enclosure to dampen resonances. I then lined with a blue jean derived acoustic insulation.

I went with sealed cabinets with a linkwitz transform to equalize to 40Hz since they will be crossed over at 80Hz by the Denon receiver to the 4 subs. I looked at the excursion at 80 Hz sealed and it should generate 106db at 80hz without exceeding its excursion limits.

Here is a picture of the completed speaker under test:

I used a minidsp Umik-1 usb mic and REW for measurements.

Rather than use the Directiva crossover, I used the technique of equalizing each driver to be flat within an octave of the crossover frequency and then applied a 24 db per octave Linkwitz-Riley filter for the cross over at 1.6 Khz. I went with a lower cross over frequency to try to better match the directivity of the tweeter with the woofer.

Left tweeter no eq:

Left tweeter with eq:

left woofer no eq:

Left woofer with eq 1K to 3K.

Left speaker with X-over at 1.6 Khz 24 db/octave linkwitz riley. Plus baffle step and Linkwitz base transform:

Ground plane without baffle step or Linkwitz transform

For the baffle step and linkwitz transform filter design, I imported the ground plane measurements into vituixCAD and adjusted the filters until I got the curve I wanted.

Following is the ground plane measurement with the filters installed in the minidsp.

 

[canukhd]

I also tried simulating the baffle step but the ground plane measurements did not match the simulation that accurately. I ultimately went with the VituixCAD simulation based on the raw groundplane measurements.

Lastly, I built a turntable and did some horizontal off axis measurements. Mic distance is 1 meter and I set the acoustic center half way between the woofer and tweeter centers.

This is measured at 10 degree increments. I am very puzzled by the peak at around 11K that appears off axis but not on axis. Does anyone have any idea why this is happening? I thought it might be a measurement error and also measured the right speaker.

This is 1/24 octave smoothed but the same peak is there.



Final in room moving mic measurement:

you can see the effect of the off axis boost at 11K in the in room average.

Let me know if you see any problems with the methodology. I am somewhat concened about the off axis horizontal measurements as they do seem to effect the actual reponse at the listening position.

 

[puppet]

I'd be willing to bet that the 11khz peak has to do with the concentric rings of the "waveguide" causing off axis diffraction.
This shows up in a graph here:https://www.audiosciencereview.com/...uild-using-denovo-flat-pack.23535/post-989578
Also discussed here somewhat:https://www.speakerdesign.net/seas/tweeters/dxt/dxt_tweeter.html

Wonder if smoothing those down would help? ... but would also deviate from the guides design theory ... whatever that reasoning was exactly.

 

[canukhd]

Thanks, at least it is not a build or tweeter quality issue. I guess I will just have to toe them in to make sure they are directly facing the listening position.

 

[DualTriode]

Hello All,

I recommend that you do some two tone IM Distortion testing.

You may find that you may want to get rid of of the Linkwitz transform bass boost.

To transform (boost) the bass frequencies you will be pushing the x-max of the driver and adversely contributing to midrange IM Distortion.

Also note, if you look at the Purify distortion vs frequency plots you will see that the distortion levels have a marked increase below 125Hz. This is another reason to crossover to the sub-woofers at a higher frequency.

Use that sub-woofer that you were talking about.

Thanks DT

 

[ctrl]

The most important condition for a good radiation of the tweeter is a flush installation of the tweeter with the baffle (no gaps left from routing or protruding edges).

If this is ensured, then the hump around 11kHz of the non-axial measurements can be caused due to fluctuations in production quality or the linearization of the axial frequency response. A dip in the on-axis frequency response around 11kHz should be ignored.
A sleight dip around 11kHz can also be seen on the manufacturer's measurement of the SeasDXT.
The 30° frequency response almost touches the axis frequency response again around 11kHz, so the on-axis FR should not be equalized flat (around 11kHz).



If you wonder why the horizontal radiation of your version is less balanced in the frequency range around 2-4kHz, this is due to your chosen crossover frequency of 1.6kHz.
Your version versus orig Directiva hor FR 0-90deg

If you are tonally satisfied, you should of course not make any changes.

If the speaker sounds too aggressive with some music pieces, you could try it with the original crossover frequency around 2.3kHz (LR4).
The crossover frequency is chosen high for a 6.5'' woofer, but as long as early breakup resonances, distortion and the radiation behavior do not cause problems, often even advantageous.
"Rules" such as that the tweeter should always be used with the lowest possible crossover frequency are nonsense and very often lead to an uneven radiation pattern.

 

[Rick Sykora]

Sorry, have not contributed much here but was helping my wife get over Covid for the early part of December...

Will strongly second @ctrl's post notably about the crossover frequency. While the Directiva threads discuss some attempts at different crossover points, I experimented extensively with different frequencies and always came back to around 2400 Hz.

As for the 11 khz diffraction, hopefully you have adjusted. Unless you have a CNC, getting a circular hole that matches the tweeter edges and depth wise can be tricky even with a good router. I find making the cut a bit deeper and getting the diameter right is helpful. You can level the depth with gasketing tape and also sand down the sharp edge. If the diameter is slightly undersized, have found that a retracing it a few times with a small round file does the trick.

Hope this helps!

 

[puppet]

I have a stupid question re: the waveguide. It seems to me that producing a small radius on the outer edge might work out fine opposed to routing it down flush. How much difference can that really make?

 

[canukhd]

Thanks for the feedback. There is a small gap between the tweeter and the hole in the cabinet. (this is my first routing attemp) I have heard some recomend filling it bondo or drywall patch. I'll give that a try and see if it smooths out the off axis hump at 11 Khz.

The 1.6 Khz cross over was somewhat arbitrary; I was trying to go as low as practical to get closer to where the purifi started to beam and thought that a 24 db/octave LR filter would keep the distortion at bay. It is easy to change the xover in the mindsp and I will give that a try as well and report back.

Lastly, I do use the bass management on the receiver with the cross over at 80 hz. I used the linkwitz transform to linearize the low bass to 40 hz in attempt to better match the filter in the receiver. This may be mistaken, but I thought the filter in the receiver would attenuate the bass enough that it would be whithin its excursion limits below 80 hz.

 

[ctrl]

Thanks for the feedback. There is a small gap between the tweeter and the hole in the cabinet. (this is my first routing attemp) I have heard some recomend filling it bondo or drywall patch. I'll give that a try and see if it smooths out the off axis hump at 11 Khz.

You can simply tape over the transitions with (removable) tape and see if it's beneficial.

 

[julbo]

Hi @canukhd, thanks for sharing!
Very interesting angle on the Directiva especially in light of the fact that the original PR is underpowering the Purifi beast.
I'd be curious to see if you solved the hump by smoothening the gaps. Had similar issue, eventually I got convinced that the baffle might be more of a culprit and to improve I'd need to massively bevel at tweeter sides - as per Heissmann DXT-MON.
How the FR looks like at the xover frequency with inverted polarity of one of the drivers?
How does the distortion look like, especially in the 400Hz zone?

 

[canukhd]

Below is a picture of the tweeter with the gap. I added an additional gasket to bring the tweeter flush with the cabinet and then filled the gap with drywall compound. The new freqency response is below. I forgot to take a picture of the now flush tweeter and will add latter today. Flush mounting seems to have reduced the off axis peak; the off axis now does not exceed the on axis FR.

 

[canukhd]

I am now planning to experiment with a higher cross over frequency. I will run the horizontal off axis with a 2.4 Khz and 2.0 Khz to see what effect that has.

 

[canukhd]

Woofer distortion. The sweep is from 150Hz to 5Khz. I am using a 3.8 ms window.

 

[Tangband]

The most important condition for a good radiation of the tweeter is a flush installation of the tweeter with the baffle (no gaps left from routing or protruding edges).

If this is ensured, then the hump around 11kHz of the non-axial measurements can be caused due to fluctuations in production quality or the linearization of the axial frequency response. A dip in the on-axis frequency response around 11kHz should be ignored.
A sleight dip around 11kHz can also be seen on the manufacturer's measurement of the SeasDXT.
The 30° frequency response almost touches the axis frequency response again around 11kHz, so the on-axis FR should not be equalized flat (around 11kHz).
View attachment 251301


If you wonder why the horizontal radiation of your version is less balanced in the frequency range around 2-4kHz, this is due to your chosen crossover frequency of 1.6kHz.
Your version versus orig Directiva hor FR 0-90deg
View attachment 251302 View attachment 251305
If you are tonally satisfied, you should of course not make any changes.

If the speaker sounds too aggressive with some music pieces, you could try it with the original crossover frequency around 2.3kHz (LR4).
The crossover frequency is chosen high for a 6.5'' woofer, but as long as early breakup resonances, distortion and the radiation behavior do not cause problems, often even advantageous.
"Rules" such as that the tweeter should always be used with the lowest possible crossover frequency are nonsense and very often lead to an uneven radiation pattern.

Agree to eveything ctrl is writing .
On the work on the HYBRID, an active two way speaker with waveguide and a 6,5 inch midbass, all measurements on and off axis showed the best result with a crossover at 1,8 KHz, but the sound was much better with a crossover at 2,3 KHz .

A two way speaker with a 6,5 inch midbass is always a compromise regarding soundquality, beaming, distortion and perfect measured directivity . Many tweeters dont sound good at 1,6 KHz .

The method of linearising the drivers with a dsp one octave above and below the crossover frequency before applying 24 dB/oct linkwitz Riley filtering is a very good one , in my opinion.

 

[Chaos01]

Did anyone build a passive Crossover for these? Looks very nice, is it possible that they dig a little deeper then the original directiva with the passive woofer?
I saw the new Buchardt A10 and wondered how deep they play, flat to 28hz. Now im inspired to build something like these.

 

[boXem]

Did anyone build a passive Crossover for these? Looks very nice, is it possible that they dig a little deeper then the original directiva with the passive woofer?
I saw the new Buchardt A10 and wondered how deep they play, flat to 28hz. Now im inspired to build something like these.

Post Directiva r1 Passive Crossovers here

@TimVG beat us to the first one, so making a place for the passive r1 crossovers to be posted...

www.audiosciencereview.com

If I remember well, @ayane got something running.

 

[Rick Sykora]

Did anyone build a passive Crossover for these?

I built @ctrl’s design and tested it. Along with other proposed designs, it is posted here:

Post Directiva r1 Passive Crossovers here

@TimVG beat us to the first one, so making a place for the passive r1 crossovers to be posted...

www.audiosciencereview.com

Looks very nice, is it possible that they dig a little deeper then the original directiva with the passive woofer?

Guessing, you meant passive radiator (not woofer). While you might get a more shallow rolloff, not realistic to get a passive sealed design to dig deeper than an active crossover with a correctly tuned passive radiator.

I saw the new Buchardt A10 and wondered how deep they play, flat to 28hz. Now im inspired to build something like these.

The A10 is active but sealed. Thus it uses equalization to extend the bass rolloff. While a sealed design more readily supports equalization at lower frequencies, there is no getting past physics. If you push it too low with eq, it will hit its excursion limits sooner when pushed.

 

[Chaos01]

Post Directiva r1 Passive Crossovers here

@TimVG beat us to the first one, so making a place for the passive r1 crossovers to be posted...

www.audiosciencereview.com

If I remember well, @ayane got something running.

Thanks for your reply, but this Crossover was created for the original Version with the passive radiator. Im looking for a passive Crossover for this sealed Version here. I guess i could Not use the same.

@Rick Sykora yeah thanks. Think the A10 also got a Special Driver with 3 voicecoils.

Would be really nice If this Open source Projekt End in a diy Kit

 

[Rick Sykora]

Thanks for your reply, but this Crossover was created for the original Version with the passive radiator. Im looking for a passive Crossover for this sealed Version here. I guess i could Not use the same.

@Rick Sykora yeah thanks. Think the A10 also got a Special Driver with 3 voicecoils.

Would be really nice If this Open source Projekt End in a diy Kit

The crossover would work for a sealed design. Should be able to readily check this in VituixCAD.

I noticed the claim around a customized Purifi driver in the A10, but would have to have a major advantage for me to buy those speakers. If something happened to a woofer, means the the owner would have to turn to Buchardt rather than Purifi. As his volume is lower than Purifi, makes an expensive woofer potentially even more expensive. In any case, I generally like his designs, but not every specific choice.

By the way, It helps more when you state what your goals are for your speaker. Then it becomes easier to determine whether an existing design can be used/modified or need to create a new one. A speaker that is designed by picking the apparent best components from other speakers often yields a poor system design. A good speaker design is a matter of balancing driver choices with cabinet and crossover designs. For most of us, cost likely comes into play too.