Ribbon Unity Horn

In my entire life, I've never blown up a conventional tweeter or a compression driver.

The second bit of trickery is that I figured I could extend the bandwidth by getting the efficiency really high and then using a filter to 'flatten out' the bandpass response. Basically flatten out the midrange in the center of it's passband, while leveraging the efficiency of a fairly large array to increase the output at the limits of it's passband.


Hope that makes sense.

TLDR: I'm doing everything thing I can to push the midranges UP, so that I can keep the crossover point as high as possible, to keep the ribbon from blowing itself up.


Thanks for your design intention explanation. Seems to be coming to fruition.
 
Basically all of my speakers these days are active-passive hybrids. I use a passive xover for the midrange lowpass and the tweeter highpass, and then I use an active xover for the speaker's highpass and for EQ. This allows me to use a single channel of amplification for the midrange and tweeter, it protects the tweeter from turn-on thumps, and allows me to do fairly basic passive xovers, since I can shape the overall response later.

jr2E9LI.jpg

Here's the response of the speaker as it is today. This has a 3rd order high pass on the ribbon and a 2nd order low pass on the midranges. The two traces show in-phase and out-of-phase, basically demonstrating that the drivers are pretty well aligned, timewise. (That's what creates the big out-of-phase null.)

u0Y9YlV.jpg

Here's the horizontal polars. Pretty good response I think! One thing that I noticed with my 3D printed speakers is that using a big ol' roundover seems to extend the point at which the directivity collapses.

J7MqHfI.jpg

Here's the vertical polars. In this measurement, I think you can see why I'm not a big fan of square ribbons - they beam like crazy and they also have off-axis dips. You really have to listen to a big ribbon with your head in a vise. The Fountek NEO CD 3.5H has a horizontal width of 3/4", which gives it really nice horizontal polars. On the flip side, the vertical directivity is narrow. Which is fine for a room, as long as the ribbon is mounted at a reasonable height.

43PMvBi.jpg

Here's the impedance of the midrange array with a 2nd order lowpass. I basically did this measurement to make sure the impedance didn't have any dips.
 
YouTube

Here's how it sounds today. This is at a much higher volume, and I also filmed it up close to show that the ribbon doesn't seem to be modulated by the midranges. I'm actually shouting over the speaker in the video but you can't hear me at all when I'm close to it!

Also, around the two minute mark, the speaker starts to sound kinda 'boomy' as I walk backwards. That's because I was walking backwards into my garage, so you were starting to hear the concrete walls of the garage coloring the sound.

For comparison's sake, here's a Danley SH50:

YouTube
 
Last edited:
u0Y9YlV.jpg

If anyone's curious why there's a dip at 6.8khz, here's the reason:

zRoA6jM.jpg

The midrange has an out of band peak. I added a 2nd order lowpass, but it looks like I may need to increase that to third order to tame it.

jr2E9LI.jpg

At 6.8khz the midrange and tweeter are out-of-phase, creating a dip. If you reverse the phase of the tweeter, you can see the dip goes away. But it's replaced by a dip at the xover, which we definitely don't want.

I hate crossovers :(
 
I don't think it is your crossover, I think something is up in your horn. Crossing the mids down more (steeper slope) will probably help it (the 6.8khz dip). Looking at all of your graphs, something is happening in both the mids and highs at, or about that point, and it gets weirder when when you are off axis, because it gets additive, instead of subtractive when you are off of vertical. How tall is that tweeter? The wavelength that is affected is about 2". Looking at the original post, I don't think this is it, as it happens in your first sweeps, and this was only half way done. Is it the length from your doscs to the tweeter? This might explain why the midrange has the overall bump in this area, as it is reinforcing that frequency in particular, and the nearby frequencies somewhat, and then there are all the harmonics that are cancelling. The HF could be cancelling at this frequency because it hits the dosc exit, then comes back, and cancels when it re-hits the tweeter, out of polarity.

Maybe the distance from your dosc to the tweeter is the same length as the height of the tweeter, so it goes additive instead of subtractive when you are off axis vertically, and goes subtractive when you are on axis, or off axis horizontally.
 
In a Unity horn, the midrange drivers are firing into a coupling chamber, which rolls off the highs. You get an out-of-band peak, like you would in any bandpass enclosure.

The fix is simple: lowpass the midranges.

The measurement posted here has no crossover whatsoever on the midranges. No highpass, no low pass.

Yeah, I get that. But even so, you'll need a heck of a steep low-pass filter to effectively get rid of those huge peaks. Not a good starting point, IMHO.
 
This next batch of pics will give you a glimpse into my sad home theater. Don't laugh, the whole reason I'm working on this project is to address this issue. I'm rockin' a very ritzy Ikea cabinet for my HT gear, topped with a piece of plywood because my new TV won't fit on it. (The old one did.)

The cherry on top is that my cat has destroyed my Vandersteens.

w5DvZCC.jpg


aWsM2vd.jpg


An idea of the size of this ribbon unity

JcBXqMv.jpg


qgldHtA.jpg


07PoeFK.jpg


The dividers in the enclosure are cut on my K40 laser. The pieces are constructed the same way that Ikea furniture is, a CLD composite of cardboard skinned with plywood. The reason that the layers don't line up perfect is that I kept adjust the diameter until it fit perfect.
 
I once made a floor to ceiling cat pole, which was a cardboard tube from the center of a carpet roll, covered in carpeting, and it had a platform on the top that was the leftover from an 18" speaker hole cutout. Perhaps you could cover the bottom of the project in carpet, to sate the cat's needs. /snark
I would decry the desecration of good speaker grills, but the cat overlords at my house have free run, and would probably find a way to make me pay for such comments.
 
Somehow I thought I'd updated this project, but I didn't.

I ran into a bevy of issues:

aWsM2vd.jpg


The primary issue was my assumption that the waveguides would print perfectly. They didn't. The 3D prints were warped. And the weight of the ribbon warped them further. It led to a situation where the entire speaker leaned backwards. You can see it in the pic above.

J7MqHfI.png


The other issue was that the vertical polars were very narrow - about twenty to forty degrees, which narrowed quite a bit above 10khz.

u0Y9YlV.png


Here's the horizontal polars of the ribbon unity from page one. The horizontal beamwidth is around fifty degrees.

GEyhBfb.png



vonhhwe.png


So I redid the design. Black is the old design. White is the new. New waveguide has a beamwidth that's nearly twice as wide. The walls of the new waveguide measure 120 degrees (horizontal) by 60 degrees (vertical.)

9klmtsm.png


L9ef9C5.png


Here's some pics to give you an idea of it's size. It's very small.
 
PfEeQ3t.png

Here's the horizonal polars of the new deisgn

u0Y9YlV.png

Here's the horizontal polars of the old design

2SOBPOJ.png

For comparison's sake, here's the horizontal polars of a Pyle PH612, which is a clone of a JBL PT waveguide.

These polar measurements tell me that the JBL PT waveguide has the best polars. It's really a wonderful design. But my new waveguide 'loads' the tweeter to a lower frequency, by virtue of it's diffraction slot. My new waveguide suffers from some broadening and narrowing of the beamwidth across it's range, due to geometry. But there's good news:

1) my waveguide is a Unity horn, and it will easily play down to 500Hz once I add the midranges. Try doing THAT with a JBL PT waveguide.

2) My waveguide doesn't suffer from peaks or dips. This is really impotant; thought I'd love it if my waveguide had the epic polar response of a JBL PT waveguide, I'm happy with the fact that there are no obnoxious peaks or dips in the passband, on axis OR off. I'd say it's well behaved, particularly considering it measures less than three liters in volume (one tenth of a cubic foot.)
 
Last edited:
Very cool project!
What mid drivers are you using?

I have been thinking about a similar setup while I was playing with the Dayton PTMini-6 planar and 3.3" Visaton FRM 8M drivers.
But since I'm still fighting with winisd and VituixCAD and only had a look at hornresp , I decided to go hornless for now.

I got the mini planars as I was not satisfied with the highs of the broadband 3.3"; not much air/detail.

I still am amazed at how great all these cheap ($15) drivers sound. I chose the FRS 8M (€10 here) because of their great response, on and off axis response and their amazing sensitivity; they deliver.


2 mids, each in their own piece of pvc pipe, 1 planar and a simple ebay 1st/2nd x-over worked out so well that it is still my main listening system while in 'to be finished stage'. I'm now working on a 4 driver, 1 planar tweeter version in a piece of pvc pipe 5"dia and 14"long. ~1 gallon/4 liter. When that is finished I will finalize the 2 mid version.

Would these drivers work in a mini unity design like yours crossed over/ horn filtered@ 4 - 5KHz?
I guess one advantage of this planar is that it is so small and flat that it can placed very near to the mids. Maybe just put 2 mids in a V, frames touching each other and the planar just in front of them? The slits for the mids and the planars membrane can be very close then. (~1/4" or about 6mm) The planars frame looks like a very short horn to me, probably boosting the highest octave.
 
Last edited: