2-way horn system based on the mk3b2

[aeroash]

We further experimented with a 3D-printed ring to narrow the exit angle of the CD to study the effects.

This is a test ring to check the fit.




The full adaptor including the ring.







The adaptor is wrapped with a POP bandage and then a fibreglass bandage. Ready for measurements.

 

[Vineethkumar01]

And then we went ahead and measured the heck out of it..

Here are the 0 degrees to 180 (not 0 to 90) degrees horizontal polar measurements of the HF146 on mk3b2 in all its gory details (the measurements beyond 100 degrees seem to be corrupted by reflections though)



Impedance measurement comparison of HF146 on horn, HF146 with ring adaptor without horn, and HF146 with ring adaptor on horn

Now what is possible in VituixCAD without EQ

With the following EQ

Normalized directivity comparison with & without ring insert into throat

Without ring insert

With ring insert

Here is a comparison of on-axis SPL between with & without ring insert

Spectral decay with flattened on-axis measurement

And it turns out we just beat the state of the art.
The ring insert enables the driver to extend its frequency response in the last octave.
That 1.4 inch throat compression driver which was earlier lazy enough to reach 14ish kHz now extends all the way to 18ish kHz (with appropriate EQ offcourse).
If only the driver had a better break up resonance behaviour beyond 10kHz, as seen in the spectral decay..
But that can be done later any day if at all needed..

 

[Vineethkumar01]

And I should mention again that @aeroash did all this hard work of capturing the measurements (very well) alone.
I just did the data processing for him..
Hats off to this guy...

From now on, no excuse for anyone to not capture polar measurements, in my book

 

[Hari Iyer]

Nice flat ln-axis response. But when I EQ for a flat on-axis response then my ears don't prefer such a sound. If it's without any EQ then its acceptable to my ears. YMMV. Maybe it has something to do with the system Q that needs to be adjusted for EQing. I have tried EQing with a constant Q equalizer - yet I don't like the sound from such a setup. Don't know why.

 

[Vineethkumar01]

Nice flat ln-axis response. But when I EQ for a flat on-axis response then my ears don't prefer such a sound. If it's without any EQ then its acceptable to my ears. YMMV. Maybe it has something to do with the system Q that needs to be adjusted for EQing. I have tried EQing with a constant Q equalizer - yet I don't like the sound from such a setup. Don't know why.

The moment you don't have constant directivity, the on-axis response won't follow the power response.
Then it is just going to depend upon a lot of factors about how an on axis flat EQ sounds in a particular room.
The important thing to note here is that we are playing around with very gently rising nearly constant directivity. Hence the on axis flat ish EQ makes sense. Also I don't care for the on axis flat EQ with this speaker. I am looking at a 15 degrees off axis flat EQ or even a 20ish degrees off axis flat EQ.

So in your case, measuring the polar response might help infer something about the problem that you have

 

[aeroash]

Mahogany planks that are nicely seasoned have been glued together and are now ready for CNC.

 

[aeroash]

Big Lip coming up !

Just like Jagger!

 

[aeroash]

 

[Hari Iyer]

The moment you don't have constant directivity, the on-axis response won't follow the power response.
Then it is just going to depend upon a lot of factors about how an on axis flat EQ sounds in a particular room.
The important thing to note here is that we are playing around with very gently rising nearly constant directivity. Hence the on axis flat ish EQ makes sense. Also I don't care for the on axis flat EQ with this speaker. I am looking at a 15 degrees off axis flat EQ or even a 20ish degrees off axis flat EQ.

So in your case, measuring the polar response might help infer something about the problem that you have

EQ is applied at one location of the mic and when the mic position changes then the EQ won't hold good. The listening window greatly reduces when you use DSP as as all listening position can't be realised through EQ. Naturally occurring flat response do not suffer from this issue as off-axis you may just get a little rolloff and not suffer from gross frequency response and phase anamolies

 

[Vineethkumar01]

EQ is applied at one location of the mic and when the mic position changes then the EQ won't hold good.

This may be applicable to common run-of-the-mill and not well-designed speakers. But we are not looking at or speaking about such speakers.
First of all, the extent to which the reference axis EQ (in this particular case of the mk3b2 horn, the reference axis is 20 degrees off-axis EQ) holds up to other off-axis angles depends on how uniform the speaker's directivity is. This is why I always recommend getting and looking at the full polar responses while making EQ/passband shaping filter decisions. The most important range of frequencies to look at is from 700Hz to about 7-8kHz.

For example, here is the normalized horizontal polars of the mk3b2 (measured the hard way by @aeroash ). Look at the beamwidth. It is almost constant from about 600Hz to greater tha 10KHz.

Here is the EQ applied on it


Here is how the spinorama looks

See how the power response, listening window response and predicted in room response all follow the reference axis response shape. That tells a lot about how well the EQ holds up off-axis.
Now, if one wants to look at the individual on and off axis responses with EQ, here is how it looks,

See how well the individual off axis responses closely follows the reference axis and are just reduced in level.

For a speaker that wouldn't hold up this well under EQ, here is the normalized polars of the Faital pro LTH 142 horn

Now compare this with the response of mk3b2 and one can understand the differences.

I would highly recommend anyone interested in all this to study this paper by Dr Earl Geddes: http://www.gedlee.com/Papers/directivity.pdf

The listening window greatly reduces when you use DSP as as all listening position can't be realised through EQ.

Just proved wrong above for this particular case.

Naturally occurring flat response do not suffer from this issue as off-axis you may just get a little rolloff and not suffer from gross frequency response and phase anamolies

This is one of the big misconceptions.
Naturally occurring flat response may be useful in a nearfield listening set up. But it is mostly useless in a listening set up 2-3m away from the speaker, as in this case. That "naturally occurring" flat response, in most cases, will result in a highly beaming horn which is unpleasant and unnatural to listen to (in a room at home) for most people (except in cases where people use very elaborate room treatment strategies). This is simply because the beaming causes an unnatural energy distribution across the spectrum and lack of a feeling of "spaciousness" depending upon where the frequency range of the beaming becomes significant.

I would highly recommend anyone interested in all these aspects to read another paper by Dr Earl Geddes: http://www.gedlee.com/Papers/Philosophy.pdf

Now, if some one had said that this horn/system is not designed for minimal 'HOMs' (higher order modes), I would probably have agreed.
But that is a story for another time

 

[prem]

I have owned the Summa speakers. I just realised it’s the same Geddes you are referring to.

From what I was told the origin of his designs dates back to Altec 19 and JBL 4430

I much preferred the Altec 19 to the Summa

In the current commercial speakers available, Audio Kinesis uses Geddes principle while designing their speakers. Duke LeJeune is the designer. He’s also teamed up with James Romeyn and has another line with him.

 

[Vineethkumar01]

I have owned the Summa speakers. I just realised it’s the same Geddes you are referring to.

Yes. It is the same Dr Geddes

From what I was told the origin of his designs dates back to Altec 19 and JBL 4430

I much preferred the Altec 19 to the Summa

That may be just that. A personal preference since I have heard about many others who prefer the Summa more

In this build, we are building a system more along the lines of the Altec 19 than the Summa. We are just employing some of the results from Dr Geddes' research. That is all..

In the current commercial speakers available, Audio Kinesis uses Geddes principle while designing their speakers. Duke LeJeune is the designer. He’s also teamed up with James Romeyn and has another line with him.

Thanks for this info. But I am aware of this.
I much prefer building my own speakers/designs than any other commercial designs though

 

[aeroash]

 

[aeroash]

 

[aeroash]

Woofer Cabinet: 25mm Marine ply



50mm Baffle

 

[aeroash]

 

[aeroash]

Installing the dual ports (104 mm x 140 mm), and finishing the skeleton for the bass horn. Next would be to assemble the horn skin, plane the edges, putty and sand down the box to a smoother finish.

 

[grindstone]

Get a cross-batt of fiberglass between the driver and the ports--touching all the walls. Usually, this lives on a cross brace (front-to-back and side-to-side, tied together) but in your case where driver access has been "frozen" to the bottom, something else will be required. Every standing wave in a 2-way cab is in the passband. Distribution of harmonics is already frozen-in by your choices of driver and port positioning. Those things were modelable (as well as the throat chamber resonance and port resonance) even in 1-P Akabak or HR (model as CH1OP). Too late for this project, but stated for the record. Knee-jerk by eye, it'll be ok, just not optimized (dirtier than it had to be). I'd line everything behind the driver and do the top too, personally. It's not just the standing waves, it's trying to eat some higher freqs from reflecting out through the cone, too. Applaud the commitment of resources and labor to make such a thing. Few would and you will be rewarded with a different window into sound than many will ever get a chance to hear. Just line 3 sides and get a cross-batt of something acoustically resistive in there to give yourself a fair chance to start.

You still have all the beeswax of pattern-matching at XO and staggering z-distances to tilt the central lobe up or down, etc as you figure-out where to cross. Such SPL curves (per angle) could also have been generated in the 1-P tools by angle and imported into vcad to add baffle ripple etc to work on XO. Power responses with filling are available for export (w/ impedance) from HR or Akabak, etc all upstream of this point. Any fault is my own for not being in the loop earlier--and I debated even posting now but chose to for posterity because your project will surely inspire others. Thank you for posting.

 

[Vineethkumar01]

For completeness, here are the technical details of the enclosure (copying from elsewhere).

We are using a bass reflex box with roughly the following specifications given below.
Since the box no longer has a rectangular cross-section along all dimensions, we are thinking the box mode distribution due to axial modes might see some attenuation along the breadth dimension. However, the axial modes along the height dimension might still be strong and some damping might have to be put in place to try and tame it. And as @grindstone pointed out, strategies are needed for taming midrange leakage through ports.

There are also the port pipe resonances shown by the pink curve starting around 900 Hz in below plot (though it is hopefully outside the passband of the bass driver). These might have come down a bit lower due to the longer effective length of the port now (due to those flares). We'll have to measure and see where they are now and what damages they do..

Native response curves with above tuning

With a bit of DSP assist and high pass filtering at the low end, the graphs look like this
Filter specs

Corresponding response curves

STATISTICS
f3 43 Hz
f6 37.8 Hz
f10 33.2 Hz
Zmin 5.2 Ohm @ 359.2 Hz
Zmax 63.4 Ohm @ 21.8 Hz
GDmax 20.8 ms @ 29.1 Hz
XmaxC 5.5 mm @ 26.3 Hz
VmaxR 9.9 m/s @ 38.3 Hz
Pmax 43.6 VA @ 42.4 Hz
-------------------------------------------------------------
DRIVER: FaitalPRO 15PR400, 1 pcs in series
n0 2.73 % Reference efficiency
SPL 96.5 dB/W Sensitivity
USPL 98.4 dB/2.83 Sensitivity
EBP 102.9 Efficiency bandwidth product
Dd 32.0 cm Effective diameter of driver
Vd 462.9 cm^3 Maximum linear volume of displacement
Cas 1.56E-6 m^5/N Acoustic equivalent of Cms
Mas 1.31E1 kg/m^4 Acoustic equivalent of Mms+Mme
Ras 4.8E2 Ns/m^5 Acoustic equivalent of Rms
Rae 8.44E3 Ns/m^5 Acoustic equivalent of Re+Rg
-------------------------------------------------------------
BOX REAR 1: Vb=110.0 l, Ql=15.0
Fb 45.0 Hz System resonance frequency
Cab 7.79E-7 m^5/N Acoustic compliance of air in enclosure
Rab 6.49E2 Ns/m^5 Acoustic resistance due to absorption
Ral 6.81E4 Ns/m^5 Acoustic resistance due to leakage
-------------------------------------------------------------
VENT REAR 1: Dv=10.4 cm, Lv=14.0 cm
Sp 84.9 cm^2 Effective area of port
Map 1.61E1 kg/m^4 Acoustic mass of air in port
Rap 4.54E1 Ns/m^5 Acoustic resistance of port losses



We probably need to take some impedance measurements with driver in the cabinet and see if there are other issues that can be solved with some damping material.

 

[Vineethkumar01]

Here are the details of the front design for the 15PR400 and the combined response with the bass reflex (ported) cabinet based on hornresp





Horn polar response




Now, for crossover and pattern-matching consideration with the CD-horn on top handling higher frequencies, here is the directivity index of the front horn


At about 600Hz, we have the DI it as 8.2dB.

Polar measurements (EQed) for the compression driver on mk3b2 horn


From the Power and DI graph, we see that the CD horn DI is also around 8dB at around/above 600Hz (will be slightly higher within +/- 0.5dB in practice since we haven't accounted for vertical polars in above plot). This is right in the ball park where we want the DI to be for matching with the bass driver front horn..
So, in theory, we have it all ready and waiting to be combined with an appropriate order low-pass and high pass filters
Let's complete the bass box build, measure and see what other surprises/devils are waiting for us..