2-way horn system based on the mk3b2

[aeroash]

My speaker quest started off with grand ambitions of finding a fancy 3 or 4-way active setup. Luckily, @Vineethkumar01 , the voice of reason, drilled some sense into me, urging us to go for something simple enough with better chances of success (or lesser chances of screwing up). The goal was to create a high-efficiency horn-based design, complete with active crossovers and DSP.

Enter @Vineethkumar01 , the bringer of audio enlightenment, who mentioned this brand spanking new redesign with improved directivity of Yuichi Arai's 290 called the mk3b2 Radial Fin Horn. It's a collaborative masterpiece from the minds of DonVK, docali, and fluid over at DIYaudio.com. We didn't waste any time, we quickly decided that this was the one we were going to attempt. And so, we begin this project with a dash of enthusiasm and a sprinkle of uncertainty!


Particulars:

• Drivers: HF: FaitalPRO HF146

LF: FaitalPRO 15PR400

• Bass Reflex cab with 120 Nett volume with F3 at 38Hz.

Action sequence:

1. Build the horn first. Which will include 3d printing and CNC.
2. Design and execute the adaptor for the CD.
3. Design and build the BR cab to suit the horn dimensions.
4. Extensively measure the horn and create a full spinorama.
5. Set the crossover and tune with DSP.

 

[aeroash]

1. HORN

We embraced the mantra of cost-cutting for this purely experimental endeavour. So we came up with a plan of printing only the inner wall of the horn and decking out the outer wall with good ol' plywood. Once that's complete we would probably go wild with a filling of POP (plaster of Paris) or expanding foam sandwiched in between. Because who doesn't love a speaker that's part 3D-printed, part carpentry?

and the printing begins....

 

[Vineethkumar01]

Enter @Vineethkumar01 , the bringer of audio enlightenment.

That is some high praise..

But honestly, I am very excited about this project.. This is a classic two-way high-efficiency speaker with some modern bits added to it.
There is the possibility to optimize the overall system very well due to DSP capability.
As already mentioned above. the horn is an evolution of the classic Yuichi A290 horn but now with some really nice polars and loading down to where it is needed for this project..
Also, this was a dream project I personally couldn't execute mainly due to manufacturing constraints and space at home..
So I appreciate @aeroash taking up this project for fun and learning. Somebody had to have the guts to take it up..
I am happy to have been a part of the project so far.
Wishing you all the very best. I hope to be of some help throughout..

Regards
Vineeth

 

[grindstone]

Good on you. That's exactly where I saw that box tuned when Vineeth was looking at it. Not a low frequency monster, but it'll be loafing and clean from barely moving.

 

[Vineethkumar01]

We were looking at the following bass alignment




Excursion limit at 20Hz is caused by about 16W of power. But by then we are talking about 105dBSPL at about 50Hz (without acounting for baffle step compensation). I think @aeroash' s usecase SPL requirement is lower than that.

In any case, since he has a DSP, we are also looking at trying out a DSP assisted bass reflex alignment.
WIth appropriate filters in place, it looks like below.



Better looking specs now especially f3 & f6.

Here is the overall of bass response without (dot-dash blue line) and with(solid black line) DSP assist.


It can be tweaked even more if required. For example, here is us being able to make the woofer handle more power and reach higher SPLs at the sacrifice of a little bit more low end extension (Definitely more chances of intermodulation distortion showing up especially with the driver handling both bass and mids but whether it is audible or not is a separate issue)



But the question is do we really even want all that..

 

[Vineethkumar01]

For anyone interested, this is a highlight feature of the mk3b2 horn

Possibility of getting constantish beamwidth (very minor narrowing) of around 100ish degrees from 700Hz to 7ish kHz.

Impedance profile

All that and the horn looks cool too..

 

[Decadent_Spectre]

Is the reflex cab part of the original design? Or is it just the horn and you have decided to add the reflex cab?

I feel a horn loaded cab would be best but reflex is easy and simple to do. The driver you have chosen is probably one of the best sounding 15s around. I actually recommended this to Vineeth I think. Faital is my favorite for good sounding LF drivers, not the loudest but they do sound good.

Good luck with your project but I ask this in earnest, why DIY? Lot of commercial choices. A known quantity is the K402 with the accompanying Jub bass bin, it will satisfy most users unless it is out of budget, you are very particular, don't have the space or like to tinker.

Again good luck with your project.

 

[aeroash]

Is the reflex cab part of the original design? Or is it just the horn and you have decided to add the reflex cab?

I feel a horn loaded cab would be best but reflex is easy and simple to do. The driver you have chosen is probably one of the best sounding 15s around. I actually recommended this to Vineeth I think. Faital is my favorite for good sounding LF drivers, not the loudest but they do sound good.

Good luck with your project but I ask this in earnest, why DIY? Lot of commercial choices. A known quantity is the K402 with the accompanying Jub bass bin, it will satisfy most users unless it is out of budget, you are very particular, don't have the space or like to tinker.

Again good luck with your project.

The horn is a standalone design and needs to be mated with whatever best suits your goals. In my case, my focus was on high efficiency and a mid-bass that would smoothly transition to the horn at around 650hz, in a simple enough cabinet, the obvious choice was BR. I would be happy if the BR touches 40Hz, anything lower will be a bonus.

Since I like woodwork, DIY was the only way I would go But, I would never build a folded horn, at least I have no desire.

 

[Decadent_Spectre]

If so then did you consider making it an MTM with 2 x 15s? Pretty standard to improve sensitivity and output. A folded horn is not usable upto 650Hz so it is a moot point. If you are very good you can design one that will measure that far but it won't sound good. A straight horn, now that's something else but to reach 40Hz the length/depth would be prohibitive in most user cases. I have heard a lot and I can tell you with certainty, at least for myself, nothing sounds like a straight horn and nothing can beat it. A no limits system will always use a straight horn in my book, whether open/ported rear or sealed or some other type might be debatable.

Interesting, good to know. Do you actually build all your projects yourself? Without carpenters? I also am curious why you do not wish to build a folded horn? By this I am assuming you are including a straight horn also.

 

[Vineethkumar01]

@Decadent_Spectre: Yes.. you had suggested the Faital drivers to me. Now I have a pair of these drivers at home wanting to be used in some project as well..

An experienced DIYer had an MLTQWT design for the 15PR400.
Here is the link to the original post:

15PR400 transmission line?

I read somewhere about a design using the 15PR400 in a transmission line enclosure? Right now I am just looking at a BR in a 150L enclosure tuned to about 40hz. But a TL may be interesting. Did GM have a design for one floating around? Thanks!!!

www.diyaudio.com

People have built things like this with the Yuichi


and these

Custom Yuichi A-290 Biradial 3D CAD Model --- Make Any Size!!

This is a unique product offered by Joseph Crowe where you can purchase the 3D CAD model for the Yuichi A-290 Biradial Horn or any combination of overall size including throat size (height or width). For the size chart click here. Simply specify your desired cutoff frequency and throat size...

josephcrowe.com

Then there are MTM's like these


And Finally, theese by Legis over at DIYaudio


Anything is possible I think. The mk3b2 is (hopefully) a nice (relatively) wide dispersion horn which can fit into many configurations.
Hence when @aeroash asked me about a high efficiency system, my plan was to start with a simple 2 way that forms the basis of a technically 'sound' system which can be scaled up if need arises. Also, more importantly, he mentioned that he liked my system subjectively, which is this (a simple configurable 2/2.5 way with 2x 9.5inch woofers crossed over to the horn at about 1200ish Hz)


Hence my suggestion to start with a project like this


The advantages are that once he learns the design process with such a simple (though it has its own set of big challenges ) yet reconfigurable system configuration, he can scale up to his heart's content if he feels like it in future (since a front horn loading can be added, more drivers can be added as in an MTM, etc later)..

 

[Hari Iyer]

Great project.
The simulated phase curve, the group delay and the vent velocity would have to be better tweaked imo. Otherwise the bass could sound as a one note bass. The system Q also needs some tweaking to allow a little higher Q. Using less damping of the box could be tried. Using DSP to manage Q can be done. But that should not be considered as design goal initially. DSP can be a band- aid later imo.

 

[Vineethkumar01]

The group delay target in this design is less than 2ms at 100 Hz, below that, there is not much point worrying about group delay in my opinion with a sensible bass alignment. The room just just dictates group delay aspects below that frequency region.
Vent velocity is important. But in a typical domestic use case (relatively high) SPL requirement (for music), the pictures look like this:

I don't see any vent velocity alarms here.

Box Q can just be varied according to one's taste and to adjust with the room (within reasonable amounts) with DSP.

The Idea is to have minimal damping. Just enough to handle any resonance issues

In my opinion, DSP should never be used as a band-aid. As long as one has reasonably good acoustic design, use DSP to the fullest possible extent. Otherwise one is just wasting available resources.

 

[Hari Iyer]

I persume the above simulated responses are based on driver TS Parameters data. Once you put the actual driver in the box and have the room factored in, everything will change. It will be interesting to note the simulated vs actual in-room responses.

 

[Vineethkumar01]

I persume the above simulated responses are based on driver TS Parameters data. Once you put the actual driver in the box and have the room factored in, everything will change. It will be interesting to note the simulated vs actual in-room responses.

Yes. All the above responses are based on T/S parameters only.
Once we put the driver in the box, unless the parameters vary drastically, we will see more or less the same frequency responses in the bass region with nearfield measurements (mic close to driver dust cap).
We are never actually in the farfield at the low bass frequencies anyway in a domestic environment. Hence, below 100Hz will be dominated by the room modes. The lowest frequencies being affected by the axial modes and then high order ones above.
Above 200Hz-800Hz will be influenced by the box dimensions and the resulting modes and how they are suppressed. This is the region where we want the driver to have minimal issues at least in this project since the woofer will be crossed over to the horn somewhere in the 650-800Hz range. With flat circular piston-based directivity, we will get about 90 degrees beamwidth with a 15-inch woofer around 800Hz.
Since the horn used in this build also has similar pattern widths around this region, it will help in directivity matching as well.
The relatively narrower (cardioidish) directivity due to the large diameter woofer-mid driver in this region compared to a typical 6-inch 2 way driver will help with less room influence too..
Anyway, we'll wait for the final frequency response measurements to see what all issues need to be addressed..

 

[Hari Iyer]

The reason I don't conclude with near field measurements are - I don't listen putting my ears near the dustcap. Even if they measure perfect over there, it won't matter much for me. I never measure at 1W/1M either for the same reason. As I won't listen at 1W/1M even though it's industry standard.
What I measure is at my listening position which is around 2 meter away from baffle between the speakers with only either the left or right speaker connected. So it's off axis by atleast 30 to 40 deg. Agreed they will not measure very well, but again I am considering the real time situation + the room for which I can apply further adjustments.

 

[Vineethkumar01]

The reason I don't conclude with near field measurements are - I don't listen putting my ears near the dustcap. Even if they measure perfect over there, it won't matter much for me. I never measure at 1W/1M either for the same reason. As I won't listen at 1W/1M even though it's industry standard.
What I measure is at my listening position which is around 2 meter away from baffle between the speakers with only either the left or right speaker connected. So it's off axis by atleast 30 to 40 deg. Agreed they will not measure very well, but again I am considering the real time situation + the room for which I can apply further adjustments.

First of all, I don't design speakers with one or two measurements taken at some random angles. I try to design a speaker in layers.
The first layer of design is to characterize/identify the anechoic response needed for an application and then evolve a concept and design the necessary acoustic constructs for it. This part of the design includes fixing bass alignment, power handling requirements etc w.r.t low frequencies mostly and deciding the rough full spectrum directivity goals.

The second layer includes doing all that is necessary to ensure that the speaker adheres to well-known scientific principles w.r.t anechoic measurements. Here is where the actual buying of drivers suited for the purpose, the build itself and its tuning tweaking, crossover design & verifying via extensive on and off-axis measurements happen. I try to see how the on-axis, listening window, early reflections, power response and directivity behave in the measured data and shape it to my requirements.

The third layer is the optimization I do with the speaker in the room based on its/theirs response at the listening position. I do the next stage of speaker passband shaping here. I mainly apply broad brushstrokes of EQ here to shape the speaker's sound according to my preferences.

So far I have had good success with the above methodology. Hence I don't plan to change it anytime soon. Also, the advantage of such an approach is that I get a good grasp of the design and can relate many things I hear while listening to the speakers to what I did in different stages in the overall design process. This also helps me avoid chasing ghosts and attributing the effect of acoustic phenomena to some components in the audio chain that make negligible effects when designed/selected/used properly.

Ok. Now coming to nearfield measurements w.r.t a monopole radiating low frequency set up, I use them not because I go and listen with my ears on the dust cap. I don't think anyone does that. I have value for it because of the following reasons:
1) It helps with driver passband shaping/linearization EQ especially with an active crossover
2) It helps with identifying low-frequency resonance issues together with impedance measurements
3) It helps one to get an idea about low-frequency bass alignment and separate out room-related issues from bass alignment issues & linear distortion issues that are inherent to driver itself.

In my opinion, anyone not using all/most of the available tools in speaker design (nearfield measurements being one) and still being able to make good speakers is either incredibly knowledgeable or incredibly lucky. Unfortunately, I have not found myself belonging to both the above groups. So I do things the hard way.

 

[soundnovice]

We were looking at the following bass alignment

View attachment 77825
View attachment 77826

Excursion limit at 20Hz is caused by about 16W of power. But by then we are talking about 105dBSPL at about 50Hz (without acounting for baffle step compensation). I think @aeroash' s usecase SPL requirement is lower than that.

In any case, since he has a DSP, we are also looking at trying out a DSP assisted bass reflex alignment.
WIth appropriate filters in place, it looks like below.
View attachment 77827

View attachment 77828
Better looking specs now especially f3 & f6.

Here is the overall of bass response without (dot-dash blue line) and with(solid black line) DSP assist.
View attachment 77830

It can be tweaked even more if required. For example, here is us being able to make the woofer handle more power and reach higher SPLs at the sacrifice of a little bit more low end extension (Definitely more chances of intermodulation distortion showing up especially with the driver handling both bass and mids but whether it is audible or not is a separate issue)
View attachment 77831
View attachment 77832

But the question is do we really even want all that..

Hello Vineet, curious to know which sw is this?

 

[Vineethkumar01]

Hello Vineet, curious to know which sw is this?

Everything was done using VituixCAD

 

[aeroash]

The thickness will be 2mm, and the fins will be printed separately and then glued. This also gives enough space to sand down to a smoother finish before the fins go on. These are printed on multiple machines simultaneously.

 

[aeroash]

The Fins