"Unitized" Image Control Waveguide

True.

A low will come down to "what polar response can I live with?"

Because the midbasses that are on the outside will have a response that isn't 100% consistent with the ones on the face of the enclosure.

A lot of this will be mitigated by the low crossover frequency, which should be somewhere between 350 and 700Hz.
 
69TchNN.png


h9UEonW.png


It's starting to look like a speaker now

UGemnJu.png

This gives an idea of the size, compared to the Cosynes

4QlTryo.png

I laser cut the trim rings for the midbasses. Basically the wood turns the foam into a constrained layer damping composite. I can upload the SVG file for the trim ring if anyone wants it.

kPXXyVQ.png

Our cat thought I made her a fort
 
More progress

First coat of paint after priming. This is pretty damn close to the finish one gets from a DIYSG SEOS waveguide.
 

Attachments

  • F70E64CD-8665-443C-A7DB-E0320F051CF7.jpg
    F70E64CD-8665-443C-A7DB-E0320F051CF7.jpg
    991.3 KB · Views: 1,029
  • EF55DB19-E8A7-48BE-B39F-76308B9B6FE4.jpg
    EF55DB19-E8A7-48BE-B39F-76308B9B6FE4.jpg
    636.6 KB · Views: 219
Last edited:
I've been wanting to build a Synergy or Unity project for quite some time, but the complexity combined with lack of sufficient time kept me from making reasonable headway. When the .stl of the March 8th waveguide was made available, I finally had a path forward.

Woofers are the Dayton TCP115-4 4-inchers, chosen for their small diameter and ability to belt out surprising amounts of bass for their size. They are inefficient, but four together seem a good match for the unity. The Daytons are in a QB3 reflex enclosure tuned to 48Hz and lightly filled with Acousta-stuff. The Unity is in it's own subenclosure and uses a 4-ohm version of the 2" Gento midrange, along with a SB19 tweeter. The tweeter uses a .4mH inductor and 6.2uF cap for a second-order crossover at around 3200Hz. The crossover between the woofers and Unity is at 650Hz, third order, via MiniDSP.

I finished these yesterday evening and haven't had much time to listen and get the kinks worked out. The most anomalous thing I've seen so far is a broad ~10dB peak at around 8kHz that I have treated with DSP (REW-generated biquad). I have some work to do on the response (using a MiniDSP HD), as the midbass also seems to be lacking (maybe too much Acousta-stuff?). I've probably made a mistake somewhere, as I was trying to get them finished quickly and get that first listen in before having to go back to work. So far, I can say they image precisely and have a lively sound, as horn-loaded drivers often do.

The waveguides were printed in blue PLA and took about 26 hours each (60mm/s IIRC). I didn't have any separation or other problems.

I wanted to thank "Patrick Bateman" for all of his work and for posting the .stl file for his unity waveguides, thus enabling me to finally build a unity project in the small amount of time I have available.
 

Attachments

  • Unity Gento SB19 TCP115-4 (1 of 1).jpg
    Unity Gento SB19 TCP115-4 (1 of 1).jpg
    625.9 KB · Views: 1,816
Good Afternoon,

Here's a fairly important update to the project, so if you have bought one of my spare waveguides, or printed your own, please read this post.

RNEnrxy.png


This is the current crossover.

Af2fkIS.png


When used with a MiniDSP with the EQ settings above, this crossover will yield polars similar to this.

There's ONE thing that I mentioned in my original post about the crossover, and I probably should have highlighted it:

There is polyester fiberfill in the midrange taps.

Not a lot of fiberfill; about half the size of a cottonball, per midrange.

mbws4Ut.png


Here's why the polyester fiberfill is a very very important part of the speaker. The midranges in the speaker have a nice and well behaved rolloff... Until you hit 7khz. Above 7khz, these midranges really start to "sing" and that creates a whopper of a peak out-of-band. And that's *exactly* why @metzb1 is seeing a peak in his measurement. The polyfill needs to be there.

Note that the midrange crossover has literally zero components in series; the only thing that's taming that high frequency peak is the polyfill.

Now, the obvious solution to the peak would be an electrical or electronic lowpass filter. I personally found that an acoustic filter (the polyfill) worked much better. It didn't just tame the peak better, it also improved the response of the tweeter and it lowered distortion.

So it's a small piece of the design, and easy to overlook, but VERY important.

The reason that the polyfill works so well is because electronic and electrical crossovers don't reduce harmonic distortion. IE, if your midrange is playing a fundamental note at 2000Hz, there will be 2nd harmonic distortion at 4000Hz and 4th order at 8000Hz. The crossover can't reduce it, because it's a product of the woofer, not the original signal.

But the polyfill CAN reduce it, because the polyfill is an *acoustic* filter.

It's a neat little trick that can be used in Unity horns.

axMd9CW.png


Here is the compact Unity waveguide WITHOUT polyfill

h9UEonW.png


Here is the compact Unity waveguide WITH polyfill

Also, the passive crossover is not a standard topology. It is loosely based on the LeCleach xover, but I used Bill Waslo's XSIM software to come up with an optimal design.

To give you an idea of the progress, this post shows the difference between the first xover and the latest, which evolved over a month : "Unitized" Image Control Waveguide

This isn't to discourage anyone from trying their own crossover, by all means please do. But keep in mind that the crossover is arguably the most important piece of the entire design. A big part of the reason that I never finish projects is because getting the crossover on a Unity horn to work it's magic is a fairly arduous process.

In a nutshell, it is REALLY easy to get a Unity waveguide to play flat on axis, but getting it to produce those beautiful polars and good phase response, THAT is the hard part. I believe that nearly everything you need to know is included in "Crossover, a Step Further" by LeCleach, but a conventional xover won't give you 100% of that Unity magic. Basically if you really want the speakers to disappear you need a crossover topology that compensates for the pathlength differences. For instance, the midrange taps on this Unity waveguide are about three inches ahead of the tweeter. If you don't compensate for that somehow, your polars and your phase response won't be correct.

It IS very possible to make a Unity horn that measures flat on axis, but has sub-standard polar and phase response. The juggling act is getting all three correct.

(Hint : read the LeCleach paper.)

At some point I need to make some kind of Wiki or user's manual for this speaker, because I know the info is scattered all over this thread.

It's on my "todo" list :)

In summary:

1) if going passive, please try the passive that illustrated above

2) if using my crossover, add about half a cottonball of polyester fiberfill to the midrange taps of each midrange

3) Copy the MiniDSP EQ settings shown above
 
Member
Joined 2007
Paid Member
cool idea on the polyfill I also have the same out of band peaks on my mids, they are 20dB below the in band response whereas yours are only 14dB above the peak in band response (I'm presuming the image you posted is before the fibre fill?). Do you have an after fibre fill response plot you can share? also have you any measurements at the horn mouth?

752052d1556146750-bm-d446-ph-4220-a-fillers_comarison-png


mbws4Ut.png
 
zRoA6jM.jpg


u0Y9YlV.jpg


This is the individual response of my ribbon unity horn project, and the polars. From an older project (1).

I just noticed that they ALSO exhibit the same issue that @metzb1 discovered, but that the peak from the midrange manifests itself as a DIP in the overall response, due to the difference in arrival times.

Just thought that was interesting; it illustrates how it's not sufficient to simply roll off the midrange, you also have to insure that it's well-behaved out of the passband, or the combined response will suffer.

(1) from : Ribbon Unity Horn
 
Here's some pics of the box coming together:

4geeVaN.jpg


aeLIUi8.jpg


GshusV1.jpg


Pics of the loudspeaker with a couple of panels missing

jCAPfaz.jpg


5z5Gq5v.jpg


Cc3s2Mu.jpg


Pics of the speaker with all panels mounted. As usual, my finishing skills are top-notch and my craftsmanship is the envy of all. /S

I think that once the entire loudspeaker is wrapped in fiberglass, I add some bondo, and paint the thing, it should look better.

On the upside, it is much MUCH smaller than the Cosynes, about one third of their volume.

I'll post measurements soon...

traveling-while-audiophile-fig14.jpg


For comparison's sake, the Donald North Audio "Sequence"
 
Last edited:
Hey patrick, just following this loosley, so does this last design with the woofers in the DNA style array not actually use the mid taps? or is this a 3 way?

I dragged the speaker outside to get some better pictures. It's still ugly, but at least the pictures are sharp!

OJkiOMK.jpg


JgKsWe6.jpg


1vVyXLy.jpg


yPMeYFQ.jpg


This is a three way. The arrangement of the midbasses and the shape of the enclosure is to accomplish a couple of goals:

1) I want to cram this speaker into a corner, similar to what Chris is doing with his Jubilees and what I am currently doing with my Cosynes. You *can* listen to this speaker away from a boundary, but it's shaped to be crammed all the way into a corner. I could have used a prism shape to accomplish the same thing. The problem that I ran into with a prism shape is that the volume of the enclosure becomes impossibly large because of the depth of the magnets of the midbasses. Basically the smallest prism shaped enclosure that I could make wound up being about four cubic feet, which is way too big.

2) The midbasses and the waveguide are arranged like that so that directivity control extends about as low as an SPL-TD1. Basically the idea is to have a home speaker with directivity control that's comparable to an SPL TD1, but with a wider beamwidth, smaller footprint, much lower cost, and much lower output. I don't need a home theater speaker that's capable of 130dB, this speaker does 110dB, which are THX levels... Enough for me.

762834d1560471153-unitized-image-control-waveguide-ea3b3487-def8-42d7-8a4a-ccf4423a7c62-jpg


763679d1560872540-unitized-image-control-waveguide-unity-gento-sb19-tcp115-4-1-1-jpg


Here's some pics from others who've built these, for comparison's sake. My finishing skills are NOT GOOD, and most will be able to make these look better than I can.
 
Last edited:
As noted before, the goal of this project is to replace my Waslo Cosynes, as my wife doesn't want giant speakers in our living room. The Cosynes sound great.

SbGFCH9.png


Here's the polar response, using a rough xover that I put together yesterday. This xover is super early, so there's a lot of room for improvement. I haven't optimized it in XSIM, I literally guessed on the xover point and slopes to the midbasses. (I'm fairly happy with the passive xover that's used on the midranges and the tweeters, it has not changed.)

The crossover shown in the pic is an active xover, between the midbasses and the Unity waveguide. The xover for the mids and tweets is passive.

The images in the bottom left show the EQ settings in MiniDSP. I am largely using MiniDSP EQ to tweak the overall response. Nothing dramatic.

The reason that I am going hybrid on the xover is because delay is required between the midbasses and the Unity waveguide. This is because the depth of this speaker isn't as deep as an SPL-TD1. Also, inductors are expensive; going hybrid saves some money.

Horiz%20Polar%20unsm%20to%2045deg.png


I don't have horizontal polars of the Cosyne, so I'll use the Small Syns for comparison. This measurement also goes to 45 degrees. This measurement was done up against a wall, to simulate the boundary effect. (My measurement was not.) Data from https://www.diyaudio.com/forums/multi-way/292379-syns-8.html#post4787283

My measurement shows a 'floor bounce' around 200hz. The downward tilt in the response curve is intentional, as noted in earlier in the thread. I may revisit that, due to the narrowing directivity at high frequency.

simulated.png

Here's the individual filters from Bill's "Small Syns." I may need to make my slopes shallower, like his are. Bill wrote:

"The trick to this design is to blend between the midrange and the woofer pair to get reasonably flat, but dropping in level, off-axis mid-high response. It's not very easy to get! I've been simming this in XSim using separate driver objects for each driver type at each angle, connecting the matching types to balance the off-axis response by trial and error.. It takes a fair amount of time to work into something reasonable, but build results do tend to match the simulations quite well."

oJb9NRq.png

For comparison's sake, this is one of the best speakers that I've measured. This is a Behringer B2030A. This is the smaller version of the speaker that bested the Linkwitz Orion in a blind test. (Linkwitz Orions beaten by Behringer.... what!!?)

8zLd3dk.png

Here's the frequency response and phase, measured at 22 degrees off axis, which is the listening axis. (It's designed to be listened to off-axis.)

That phase curve is pretty awesome if I do say so :)

sVUmsZd.png

This is the frequency response and distortion. This one needs work. The box is buzzing audibly, and I think that's generating the distortion.
 
Last edited: