Introducing Directiva - An ASR open source platform speaker project

[boXem]

How can good directivity be optained with a front like Kii3? When Heissmann - you link too - clearly states that the DXT needs a very narrow - almost non existing - front baffle, to get a minimum of disturbance and widening in the response near the X-over......

Are you sure you do not mix the Kii3 with the Grimm?

 

[Digital_Thor]

Are you sure you do not mix the Kii3 with the Grimm?

Oh.... actually no.... I just remembered wrongly. Kii3 actually has a nice slope around the tweeter to compensate for this - instead of a bevel. Just googled a picture and acknowledged that my memory played tricks with me - sorry
I was purely thinking of performance vs ease of design.

 

[boXem]

Oh.... actually no.... I just remembered wrongly. Kii3 actually has a nice slope around the tweeter to compensate for this - instead of a bevel. Just googled a picture and acknowledged that my memory played tricks with me - sorry
I was purely thinking of performance vs ease of design.

The Grimm LS1 also use the Seas DXT with a baffle that is everything but narrow

 

[dwkdnvr]

Yeah, I find the Grimm LS1 a rather interesting design. Was headed in the direction of mocking up a DIY version to play around with last year before deciding move which has put DIY efforts on hold for a while. The idea of going so wide with the baffle that you push both the diffraction reinforcement for the tweeter and the baffle step on the woofer into different regions than you typically see. Moving the baffle step down to ~250Hz means you can handle it as part of a blend to a sub/woofer, and it gives you pretty uniform directivity down lower compared to smaller stand mounts. Polars probably aren't quite as impressive as the Heissman, but still very good.

 

[Digital_Thor]

The Grimm LS1 also use the Seas DXT with a baffle that is everything but narrow

Indeed.... I red the white papers:
https://www.grimmaudio.com/publications/speakers-white-paper/
But what I meant, was that I remembered the Kii3 and the Grimm as having a more flat front baffle. If it should be easier to make, I would do the bevels, since they can be cut with a bit of patience... and maybe they measure and also sound the best.
The rounded front on both the Kii3 and Grimm, seems tricky to make nice for a DIY.
But maybe if I did it a little more "edgy"...... hmm... still like the bevel more
http://groundsound.com/GalleryHarryEng_2.php

 

[tktran303]

With this decision in hand, I ran some Bassbox and VCAD sims and adjusted cabinet volume for isolating the tweeter. This resulted in a cabinet volume of 11.5 liters and a f3 of around 42 Hz.

Smaller box, more bass!

What’s the F10?

 

[McFly]

PR alignments fall off a cliff after F3, so F10 is probably something like 35-38.

 

[Rick Sykora]

Smaller box, more bass!

What’s the F10?

f10 is low 30s. Depends on whether you trust Bassbox or VCAD more for projection. At this stage, damping can move things a bit one way or another...

 

[Rick Sykora]

Spent most of the day woodworking with a bit of redesign along the way. Bassbox has good tool for designing enclosure along with compensations for internal parts. The default cut list has a recessed front and back baffle. I usually adjust the default to have a seamless front baffle. Dry fit looks good. Will supply pic after dinner!

Still have some chamfers to complete before I start to glue.

 

[Rick Sykora]

Here is pics of dry fit of cabinet internal...

and a view with front baffle in place..

 

[Rick Sykora]

Chamfers and t-nuts are done...

Last bit of woodworking will be front baffle roundover. We are going with 22 mm roundover. I did not have, so I ordered. For those still on standard measures, that is 7/8 inch roundover. Gluing will commence later today!

 

[Rick Sykora]

Since the Directiva custom cabinet has non-right angles, standard pipe clamps may not work, but gravity can be leveraged along with a solid 90 degree fixture (in my case, a table saw fence).

First I sanded the top inside edge of the bottom, cleaned and applied wood glue. This mates with the bottom edge of the back. I angled the back slightly as I lowered it and pushed it against the fence. This forces the glue to the inner seam and not on the side resting against the fence...

 

[Rick Sykora]

And laying each side down, glue the 2 seams. This was done a side at a time, but only took the one pic...

 

[Rick Sykora]

Approaching the finish line, I glued the 3 inside seams for the top and placed the cabinet on it upside down. I also used the brace that will isolate the tweeter to spread the cabinet sides. Added some weights to hold it in place like so...

In a square cabinet, to get a good glue joint, would normally have glued the brace in sooner. With the angled edges, this is not an issue. Also, while the dry fit seemed ok, the brace is a bit to long now that the cabinet is tighter. As I need it in place for initial measurements, will trim and glue it next.

 

[Rick Sykora]

While I wait a day or two for that roundover bit, thought I would share the latest directivity sim from @ctrl.

He tells me that we can expect to meet or exceed this. Once I have some mounted driver measurements, will create an initial crossover, measure and simulate again. Expect to iterate at least once more before sending off to Amir for a Klippel analysis.

 

[ctrl]

He tells me that we can expect to meet or exceed this.

I don't even know what "exceed" means, I couldn't possibly have claimed that

The simulation depends very much on how well the physical dimensions of the drivers involved have been measured.
Any non-ideal behavior of a driver in reality (cone or surround resonances,...) will not be captured by the simulation.

If it turns out well, the simulation of the frequency responses (hor and ver +-180°) is in the range 200-9000Hz at +-1dB (compared to reality).

This means that the simulation shown by Rick should be very close to reality.
Here as large images (normalized to on-axis response):


But that reality outperforms the simulation will rather not be the case. But if the final result comes close to the simulation, then it is better than the majority of Amir's tested speakers in terms of directivity.

 

[Biblob]

Great job @Rick Sykora and @ctrl!

 

[Rick Sykora]

I don't even know what "exceed" means, I couldn't possibly have claimed that

The simulation depends very much on how well the physical dimensions of the drivers involved have been measured.
Any non-ideal behavior of a driver in reality (cone or surround resonances,...) will not be captured by the simulation.

If it turns out well, the simulation of the frequency responses (hor and ver +-180°) is in the range 200-9000Hz at +-1dB (compared to reality).

This means that the simulation shown by Rick should be very close to reality.
Here as large images (normalized to on-axis response):
View attachment 118035 View attachment 118039

But that reality outperforms the simulation will rather not be the case. But if the final result comes close to the simulation, then it is better than the majority of Amir's tested speakers in terms of directivity.

Well your english is way better than my german! I recall you said something a while ago about maybe seeing improvement above 10 kHz?

If not, just blame the marketing guy.

 

[tktran303]

Yeah lost in translation

“It's not the length that matters, it’s the girth”

 

[Rick Sykora]

its all about the need to überschreitet!