Celilo : A Small, Affordable Two-Way Unity Horn

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
For those who are curious, although there were differences over the years, by the end I believe Dr. Geddes has constructing cabinets using either 2 x 3/4 MDF or Renshape board. Cross braces I believe were a finger locking style and damped between with ~ 50% polyurethane 50% 3M glass microspheres. Waveguides were polyurethane too, I believe <50A, poured into a mold.
 
image.jpg


My Summas were built like a surfboard : foam core, carbon fiber skin. If I'm not mistaken, the paint was latex, which worked surprisingly well.

The walls were crazy thick, something like two inches.

Before I sold them, I repainted them in black plastidip, because the original paint had become scuffed up.
 
Drk9q8w.jpg


FYI -

Celilo II seems to be a winner. I'll be uploading the STL files in the next couple days. I've been busy refining the actual speaker, haven't had a chance to assemple a repo. I will also be posting a box design.

A7CTr2H.png


Here's what the Celilo II polars looked like last week. This is really good performance, IMHO.

vBVSRLk.png


And here's how they look today. For a conventional waveguide, this would be excellent performance, but for a Unity horn with eighteen holes in the waveguide, it's kinda unbelievable! In a bandwidth of three octaves, from 1400Hz to 11khz, Celilo II is +/- 1.5dB in a beamwidth of 45 degrees. (yellow is 0 degrees, orange is 11 degrees, red is 22 degrees.)

Econowave.jpg


For comparison's sake, here's the Pyle clone of JBL's "progressive transition waveguide", used in the Econowave. A very nice waveguide, no doubt.
 
Last edited:
Thank you! Dare I say it, we haven't seen waveguide performance like this before.

Econowave.jpg


A really good waveguide combo, like Zilch used in the Econowave, will suffer from some 'wiggles' in the response that are caused by pathlength differences. Whether they're audible is anyone's guess, but they're there.

index.php


A speaker like the JBL M2 uses DSP to 'hammer' the response flat. And I could do that too, but I really wanted to make it as smooth as possible without EQ. Note that this measurement of the M2 uses a 65dB scale, and my scale is 50dB.

Io7Xf0w.png


Here's the completely unsmoothed response of Celilo 2. This is approaching the smoothness of a dome tweeter on a baffle, but with directivity control and high efficiency. The 'secret' to making it smooth is making the aperture of the waveguide as small as possible, just 1.6cm. That's less than 40% the size of a one inch compression driver.
 
Administrator
Joined 2004
Paid Member
Also not seeing the photobucket pics. It's much better to attach images here - at least they stick around. Limiting images to about 1300 pixels wide is also a good idea, as the forum software does not do a good job of dynamic resizing.
 
What's the situation with cross firing these? Do they need to be toed in less because of the design of the waveguide?

If they are facing forward-ish does the long path stay on the inside, but if crossfired it's on the outside?

There's a few reasons:

1) In my house, I have my Cosynes pushed all the way into the corners. One of the ideas I had with the last project was to have a baffle that was at a 45 degree angle. But I had a difficult time building the enclosure and making it WAF friendly. So in this project, I just had the waveguide itself fire off-center.

2) The way that I maximize the waveguide size is that I cut it in half, and print the two halves on the same printer. This is how I can print a waveguide that's 12 wide using a printer with a bed that's just 7" tall. But if the waveguide is symmetrical, that means there will be a seam right down the middle of the waveguide. Making the WG asymmetrical solves that.

3) A lot of my research into sound tells me that sound hates symmetry. So I try to break up that symmetry as much as I can. That's why I never make round waveguides or square waveguides.

hHtpv8I.png


Last but not least, this arrangement delivers even coverage from one side of the room to the other. The pic above is an actually sim of my living room, and shows how the waveguide works in a typical room.
 
I don't see any obvious advantage to symmetry.

It's easier to build, but it measures worse.

If you have a 3D printer, asymmetrical is superior.

In future projects, I will likely make things a little bit LESS asymmetrical; this waveguide has an asymmetry of four to one, I'll probably back that off to 1.62 to one, the Golden Ratio.
 
So if the audio channel had to be centered on the user vs to the side, would one just orient this waveguide on its side?

Celilo should be oriented so that it's "hot" axis is pointed to the opposite side of the room

If used as a center channel, it should be oriented at a 90 degree angle, so that the HOT axis goes UP

Note that the difference between the HOT axis and the center axis is fairly mild, just a couple of dB
 
LgQEctr.png


0AmDLSA.png


Last night I dragged my Waslo Cosynes out to the driveway and measured them. I also modded my Kali Audio LP6 with a roundover.

Then I put the Cosynes in the garage, and the LP6 in their place.

After listening all night, I am starting to think that there's really something to the use of a very large waveguide.

If you look at the measurements of these two speakers, the LP6 is exemplary, particularly for $150. It boggles my mind that you can get a powered speaker that's THIS good for so cheap.

But the Cosyne just runs circles around it in the imaging department.

Subjectively, the Cosyne's phantom center is just ROCK solid. I can move *towards* the speaker, *away* from the speaker, I can even move to the left and to the right a bit, it sounds like I have a physical center channel.

With the Kali, If I center myself perfectly, and get about half a foot closer to the floor, and move about a meter towards the speaker, the phantom center 'snaps' into place, but it disappears if I'm not in the sweet spot. And even IN the sweet spot, it's nowhere near as solid as the Cosyne.

The Kali soundstage is a little bit deeper and wider, but very diffuse. It basically sounds like every good loudspeaker you've heard at an audio show. Really great performance, but not much different than a hundred speakers you've already heard.

But the Cosyne....

I am starting to wonder if I just need to put up a false wall to hide these things or something, like Geddes did with his Summas...

vBVSRLk.png


Here's the current polar response of Celilo. This is the tweeter only.

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


For comparison's sake, Bill's "Small Syns" speaker. Note this is a 60dB scale, mine is 50dB.

Though Celilo's waveguide is much larger than Kali's, the Kali speaker maintains pattern control lower. I believe that a lot of this has to do with the beaming of the Kali woofer, the Kali crossover, the dimensions of the cabinet, etc. Charles Sprinkle (designer of the Kali and JBL M2 waveguide) is quite a wizard.

I'm a bit worried that Celilo will sound little different than the Kali, maybe worse, and maybe won't be an equal for the Cosyne unless I make it much, much bigger.

Bill Waslo has generally suggested going with a bigger waveguide.

Decisions decisions!
 
Patrick,
While symmetry does have 1/4 mode at a fixed frequency and asymmetry broadens that frequency range, but isn't it true that the effects are mostly on axis, with toe-in listening its not a big issue, agree?
Also, Earl's foam solution takes care of that.

Asymmetry also produces another problem, the lower pattern loss cut-off frequency is now not constant. On one side the wave guide is longer and so the cut-off is lower which is good, but on the other side the wave guide is smaller and cut-off is higher, not good.

How do you deal with it, or it does not matter that much or something else?
 
So I've had the Kali LP6s in place of the Cosynes for a few days now. Some random observations that might apply to the debate of "two way monitor with waveguide" versus "Unity horn":

The phantom center on the Cosynes is RIDICULOUS and if that's your thing, go buy or build a pair. I've never heard a speaker with such a rock solid center; about the only thing that comes close is the Danley SH-50. The waveguide on the Cosynes is something like 24" in diameter. The Lambda Unity horn is 16". The Danley Sh-50 is 28" IIRC.

The stage on the LP6 is significantly larger; it extends beyond the boundaries of their physical location. OTOH, the Cosynes can be placed further apart. IE, the boundaries of the Cosyne soundstage basically ends at where they're placed, but they can be placed *ridiculously* far apart and still keep a phantom center. My living room is pretty darn big (new houses have huge living rooms) and I had the Cosynes nearly ten feet apart.

This morning I was listening to some fairly mellow rock, like R.E.M. and 10,000 Maniacs, and the LP6 really shines on this stuff.

But then I noticed something ELSE that's interesting about the Cosynes: I put on some industrial rock (The Revolting Co cks) and the Kali LP6 just RUINED that music. The Kali LP6 somehow managed to make The Revolting Co cks sound "polite." This reminded me of something that happened to me a few years ago: I went and saw Nine Inch Nails in concert, and they brought a humongous L'Acoustic array. And it sounded *incredible*, it was the first time I've been to a rock concert that featured an actual sound stage. It was a live show that had actual width and depth to the sound.

But it also sounded incredibly sterile. I'm not sure if it was the absence of distortion, due to the ridiculous headroom, or something else. But it just sounded TOO CLEAN.

This might be A Good Thing, who knows? Maybe the LP6 is just really low in distortion. But it doesn't "slam" like the Cosyne does.

I noticed something similar with my Metlako speaker; when I went from four midbasses to two midbasses, it sounded a little bit less dynamic, at all volumes.

I noticed the same thing when I went from my Gedlee Summas to the Vandersteens; the Summas just seemed to have limitless dynamics, whereas the Vandersteens seemed to compress the dynamics, at all levels.

The weird part was that this seemed to happen at all volumes, not just loud SPLs. It's one of the reasons that I HATE it when audio shows use female voice to show off their speakers. If you want to see what a speaker can do, put on something with real dynamics, not Diana Krall.
 
Hi Patrick/John,
I have pretty much always found that for otherwise similar designs a larger waveguide almost always sounded better (to me anyway). The imaging of the bigger 3D printed waveguide speakers is a little better than the SmallSyns and the sound just seems more solid, even though they both use essentially the same drivers. Probably because pattern control goes down to a lower frequency(?). That's why I was starting to make a humongous waveguide with the printed stub (until the reality of the size of our living space occurred to me!).

One thing that might be helping with the Cosyne's strong center image is that the waveguide is higher up from the floor than with most speakers -- maybe try raising other waveuides to see if that makes any difference. I do remember that I could sit basically in front of one of the speakers with the other 10ft or more away and still hear a center image between the speakers. I get a similar effect with the SmallSyns, and have them sitting up with the horns a little above ear level.

As your plots show, the Cosynes frequency response isn't the flattest in the world (the crossover went more for phase linearity than for flatness at the time).
 
Yes, Celilo I and II both use the NE19.

The only real problem with the speaker is:

1) waveguide is too small; it's losing pattern control at 2khz

2) I think cross-firing inside of the waveguide may have cause the wavefront to "detach" from the waveguide. I've seen this with a lot of my waveguides that have a rapid expansion; basically if it expands too fast, the wavefront just "sails" right down the center of the waveguide as if it wasn't there. Charles Sprinkle alluded to this in an interview about the M2, basically saying that it's wide coverage would be impossible without the 'knuckles' in the center. Recently, JBL took this even further with their in-walls, using COMSOL to come up with a shape that barely even looks like a waveguide:

Acoustic Horn Design – The Easy Way (Ath4)
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.