Unity Horn Designs

Status
Not open for further replies.
JessicaG said:
let us know sheldon....

what kind of speakers have you heard before these?

After a long hiatus, I got back into better audio gear a couple of years ago. At the time, I listened to a number of commercial speakers. I ended up getting some Alon Exotica's. I liked the sound, they worked well with tube gear (friendly impedence curve), and are relatively small for the output. It's a 3-way with a Raven ribbon on top, a 4 1/2" Seas mid, and an 8" Seas woofer. Despite the name, nothing too exotic. But I think the designer did a nice job with integration in the crossover. I've listened to a few speakers since, but no extensive trials. I also built a small pair of backhorns for a second system while I've been waiting for the Unity stuff.

So what basis do I have for choosing the Unity's? Not much really - indulging a whim. As said earlier, I'm a sucker for cool ideas, especially those that are based on sound logic. The Unity is one of those. Plus, I'm curious to try a system that should offer very good dynamic range - better than a box speaker. And I'm curious to see what the fuss is about SET's, so I'm planning on building a couple of low powered amps (based on 801a tubes) to power the horn (biamped). The Unity will have the efficiency to make them practical (if any of this can be said to be practical). I like making things, and I like making thing that are a little different, even more.

Sheldon
 
Sheldon said:

By the way, John Hancock, in post 11 here, alluded to a mod that traded some output for improved linearity. I'm assuming that he was referring to the design at the mouth of the horn, and what looks like a defraction grating? Is this the same issue you addressed in your AA post regarding placing acoustic foam at the mouth of the horn?

The mod I did is to the holes themselves, not the mouth of the horn and yes it does adress the same issue, which was a reflection off the holes at 4kHz. Since Tom is giving out some details on his implementation, I hope he won't mind me giving some more details on my mod. I've updated the thread on my system with a few more pitcures...

http://www.diyaudio.com/forums/showthread.php?s=&threadid=17642.


Tom Danley said:
John H. was one of Nick’s customers back then and I was impressed how fast he figured out how they work. He probably gets paid way more doing what ever he does but he ought to be in audio someday, when he can afford the big pay cut ha ha..

Ha Ha...I'll pretend this is true. One day I'll be an audio pro, one day.


moray james said:
To Tom and or John H. I would be interested in knowing if there are any mods that can be made to Nick's horn to bring it up to a more current version as I assume that much has been learned about the unity in the last few years? I guess it would be easy to try the foam idea (or John's ribs) without much if any trouble but I also wonder if cutting a chamfer or 1/4 round on the mid driver throat holes would make matters better or worse. I suppose that too could be tried and repaired with the aid of some bondo if push came to shove but I would rather have the "Nod" first before hacking away.

I modded my holes by filling them in with bondo (with a depth of 3mm (as I recall) and then drilled out a new smaller hole, then ground out the old hole to make it wider. Take a look at the above thread. Could easily be undone again without any cosmetic impact, but I am confident you won't want to undo it.

best regards, John
 
Hi John,

Glad you're back, and thanks for that explanation. When I first looked at your picture, I had not seen Nick's horn. Even if I had, that mod would take a very keen eye. Only problem is (aarrg), I just put the clearcoat on my painted horns, not 20 minutes ago. Looks like I have to get out the tools and get to work again).

thanks,
Sheldon

Hmm, a thought occurs to me (always a dangerous sign). The holes in the Lambda horns are about 20mm. If I just took a 3mm section of hose (or somesuch) with an OD of 20mm an ID of about 8mm, could I just insert that in the hole, flush with the inside of the horn- at least for trial purposes. Would the 20mm diameter of the hole behind the insert be adequate? Or should I go ahead and enlarge that?
 
Hey John,
I did roughly the same thing on my last unity (per your suggestion), but I had problems with getting flat response from the mids. It turned out I needed really large holes to get flat response with my mids, but of course this screwed up the tweeter's response. At least making the change allowed me to somewhat validate my modeling approach. I was just wondering if you got flat response using your mids or if you relied on eqing in your DSP filters.

After playing around with my model a bit the other day, the specs I came up with to get flat response from small holes (and even larger holes, like on the standard unity) were a bit different - heavy cone, high BL, Fs around 80hz (without a rear chamber). Maybe I only found a local optimum, but I was thinking about building some and giving them a try in the future. Any thoughts?
 
John Sheerin said:
Hey John,
I did roughly the same thing on my last unity (per your suggestion), but I had problems with getting flat response from the mids. It turned out I needed really large holes to get flat response with my mids, but of course this screwed up the tweeter's response. At least making the change allowed me to somewhat validate my modeling approach. I was just wondering if you got flat response using your mids or if you relied on eqing in your DSP filters.

After playing around with my model a bit the other day, the specs I came up with to get flat response from small holes (and even larger holes, like on the standard unity) were a bit different - heavy cone, high BL, Fs around 80hz (without a rear chamber). Maybe I only found a local optimum, but I was thinking about building some and giving them a try in the future. Any thoughts?

Hi John,

I hear you're out in California now...hope you're enjoying the sun. We don't get much here in Zurich. It's raining right now :-(

As an acoustical element, the holes are a parallel series and inductance. The inductance is proportional to the length of the hole divided by the area of the hole:

ind~l/A

The resistance is proportional to length divided by area times radius:

res~l/(A*r).

Now, way back when I did these calculations I decided that the inductance was the dominant impedance. I was careful, then to keep the ratio of l/A constant when I made the holes smaller so that the inductance stayed the same at the crossover frequency. Doing this, though raises the resistance. I cut the radius of the hole by about half, so the resistance would have doubled. If my calculations were right, though, the inductance still would be the dominant impedance so the total impedance wouldn't have changed. It's possible I got my units wrong, though, and the resistance dominates. I certainly didn't notice an effect when I made the mod. Since I am using FIR filters I may not have even noticed it, though (poor Tom has to expend a lot more brain power to design his analog filters).

If you are having trouble getting the f3 of the acoustic filter high enough, why don't you just make the volume of the air cavity formed by the back of the horn/front of the diaphragm smaller. Remember that cavity together with the holes form an LC (RC if it is the R that dominates for the holes) low-pass filter. The L is from the holes (again R if it is resistance that dominates) and the C is the air cavity. Weren't you using larger drivers than on the Sound Physics horns? This would have raised your C, so you would need a correspndingly lower L (R) to keep the f3 the same. Alternatively, you could just fill in your cavity to make that volume smaller.

best regards, John

John
 
Sheldon said:
Hmm, a thought occurs to me (always a dangerous sign). The holes in the Lambda horns are about 20mm. If I just took a 3mm section of hose (or somesuch) with an OD of 20mm an ID of about 8mm, could I just insert that in the hole, flush with the inside of the horn- at least for trial purposes. Would the 20mm diameter of the hole behind the insert be adequate? Or should I go ahead and enlarge that?

Nope, won't work...you have to keep the ratio of area of the hole to length of the hole constant. You could stick a short piece of pipe in there, but then the remaining part of the hole (without the pipe) will still form a significant impendance. You need to grind away this part of the old hole so that acoustically it isn't there (you can see what I did in the pics of my horns.)

John
 
Hi John,
Thanks for the reply. I was in California for the winter, but now I'm back in the midwest. California's just too expensive.

When I play around with my model, changing the holes definitely changes the response. That doesn't mean it happens in reality exactly the way my model predicts, though... I usually don't have trouble getting the F3 of the lowpass high enough. At least in my model I can always put the holes closer to the tweeter (which seems to dominate) or make the front chamber smaller. What I normally run into is a response that has a peak at the top end and the bottom end, like a mistuned bandpass. The sensitivity is then usually much lower in between the peaks, which is what I'm trying to avoid.

For example, see:
An externally hosted image should be here but it was not working when we last tested it.


This gets me flat response between about 200hz and 2khz. If I take the foam plug out to use it as a unity horn, that drops to around 1.2khz, iirc, which is as you would expect.

With these holes (same horn, drivers):
An externally hosted image should be here but it was not working when we last tested it.


I get a moderate peak at the top and bottom of the frequency range, as I described. I think these were 3/8" diameter holes, 0.1" long. Here's predicted vs. outdoors measured response:
An externally hosted image should be here but it was not working when we last tested it.


Red is measured response, green is predicted, blue is predicted including the horn's directivity. This is not too bad compared to a lot of other combinations I've modeled, but the larger holes do give much flatter response (which is what I'm shooting for with the smaller holes).

The tweeter response was really smooth with these holes, though.
 
John Sheerin said:
Hi John,
Thanks for the reply. I was in California for the winter, but now I'm back in the midwest. California's just too expensive.

When I play around with my model, changing the holes definitely changes the response. That doesn't mean it happens in reality exactly the way my model predicts, though... I usually don't have trouble getting the F3 of the lowpass high enough. At least in my model I can always put the holes closer to the tweeter (which seems to dominate) or make the front chamber smaller. What I normally run into is a response that has a peak at the top end and the bottom end, like a mistuned bandpass. The sensitivity is then usually much lower in between the peaks, which is what I'm trying to avoid.


hmmm...If the f3 of the acoustic low-pass is high enough, then the holes shouldn't be the issue. I think you may have an issue with having only one entrance hole per driver. That may not be the only issue, but it certainly is one issue. You want to keep the dimensions of that front chamber small relative to the wavelength of the frequency. Having two holes effectively halves the dimensions of the chamber. With one hole, you're setting up a transmission line, with associated resonances. What drivers are you using? What's the inductance/DC resistance? What are the dimensions?

John
 
hancock said:
Nope, won't work...you have to keep the ratio of area of the hole to length of the hole constant. You could stick a short piece of pipe in there, but then the remaining part of the hole (without the pipe) will still form a significant impendance. You need to grind away this part of the old hole so that acoustically it isn't there (you can see what I did in the pics of my horns.)
John


Thanks John,

When I saw how you removed area behind the new holes, I figured that might be the case but asked anyway, hoping to avoid carving away. Then another thought came to mind. You used the words "reflection off the holes" to describe the cause of the 4khz anomaly. Since the walls of the mids entrance holes are perpendicular to the tweeters wave front, the nature of the physical interaction seems straightforward, unless you were simplifying for brevity's sake. If it is just a simple reflection, I surmise that making the hole smaller just makes that area available to create a reflection smaller. Small enough to be invisible to 4khz or just simply reduced enough in area so that the reflection energy drops to insignificance?

Either way, by breaking up one larger area into several smaller ones should either reduce the reflected energy or, if the reflection angles are different, diffuse it. So, one last try to avoid getting out the cutting tools and bondo. How bout we insert several pieces of thin-walled tube into the entry hole? The tube pieces would need to be the same length as the entry hole, and fit snugly against one another and the wall of the hole. Would this break up and diffuse the reflection? If it worked, it would allow the hole area to remain essentially unchanged, so you wouldn't give up output. Cross section below.

Sheldon
 

Attachments

  • unityhole.gif
    unityhole.gif
    3.9 KB · Views: 641
Wow!

Then another thought came to mind. You used the words "reflection off the holes" to describe the cause of the 4khz anomaly. Since the walls of the mids entrance holes are perpendicular to the tweeters wave front, the nature of the physical interaction seems straightforward, unless you were simplifying for brevity's sake. If it is just a simple reflection, I surmise that making the hole smaller just makes that area available to create a reflection smaller. Small enough to be invisible to 4khz or just simply reduced enough in area so that the reflection energy drops to insignificance?

I guess it's similar to the straw port seen on some of the Mission
loudspeakers. It was done to cut down on port noise. Provided a low pass filter to reduce port noise.

No real reason that it couldn't work on some level. THe length would be rather critical. The whole length of tube may act as a resonator. And most probably as a 1/4 wavelength resonator. But if is above the passband who cares! Definitely worth experimenting with.

Mark
 
Re: Wow!

mwmkravchenko said:


I guess it's similar to the straw port seen on some of the Mission
loudspeakers. It was done to cut down on port noise. Provided a low pass filter to reduce port noise.

No real reason that it couldn't work on some level. THe length would be rather critical. The whole length of tube may act as a resonator. And most probably as a 1/4 wavelength resonator. But if is above the passband who cares! Definitely worth experimenting with.

Mark

Thanks Mark,

I was proposing that the length of the inserts be equal to the thickness of the current horn wall, which is about 1.25cm. That would make them a 1/4 wavelength resonator at about 6.8khz (343ms/0.05m). If the crossover is at 1.5k, we should be out of trouble by then.

Sheldon
 
Good question catapult - during my experiments I was planning to try the same thing.

I also wonder about getting away from circular holes altogether and going to either
a) a large hole, but covered with a minimally (10-20%) perforated grill - something like to cover for ceiling speakers. It could present the same net surface area, but impact the tweeter much less.
b) a slit running the width of the horn. this would in theory have a much worse diffraction effect, but again might be OK if covered with a minimally perforated grill.

The radiation pattern from the mids might be a problem, but at the wavelengths involved with the mids, it's not entirely clear to me that it's insurmountable.
 
hancock said:


hmmm...If the f3 of the acoustic low-pass is high enough, then the holes shouldn't be the issue. I think you may have an issue with having only one entrance hole per driver. That may not be the only issue, but it certainly is one issue. You want to keep the dimensions of that front chamber small relative to the wavelength of the frequency. Having two holes effectively halves the dimensions of the chamber. With one hole, you're setting up a transmission line, with associated resonances. What drivers are you using? What's the inductance/DC resistance? What are the dimensions?

John

Hi John, all,
After playing around some more with my model, I don't think resonances were the issue (at least what you'd think of as t-line resonances). I mean, the measured performance did correlate very well with lumped element modeling, so it's not as if the performance was unexpected - it was just unexpected at the time because I hadn't written a program to model the unity yet. It looks like I just had bad system tuning combined with interaction with the reflection from the rear of the horn. Tonight I found that if I enlarged the front chamber a bit, made the rear chambers about 1/3 the size (difficult), moved the holes towards the tweeter by about 1" (very difficult), and went to two 3/4" diameter holes 0.1" long, I would get much flatter, higher sensitivity response with more top end extension.

Switching to longer holes would degrade the peformance a bit, but not too bad (1/2" long). If I then tried to switch to the smallest, shortest holes possible while keeping the L/A ratio constant to keep the same performance, I could only go down to 0.55" diameter, 0.1" long, as going much smaller than this, it would not be possible to maintain the ratio due to the end correction on the holes swamping the physical length.

Here's a comparison of the performance of the above combination to what I previously built:

An externally hosted image should be here but it was not working when we last tested it.


So while not perfect, it would be quite a bit better. Now I'm back to trying to find a driver which will work with the small holes with better response than this.

To answer your other questions, I was using 5" mids - about 4.4" inner diameter on the gasket, about 1/2" from the dust cap to the outside wall of the horn. Le was about 0.25mH, Re around 7.3 ohms, so I don't think that was contributing to the problem.
 
Status
Not open for further replies.