Suitable midrange cone, for bandpass mid in Unity horn.

Cone break up is poor fidelity because it is full of distortion products and is not a faithful reproduction of the original signal. Just because you can make some noise at an elevated frequency doesn't mean it usable. The reason why your horn "appears" to defy beaming is because it consists of break up and resonances which are being sprayed all over the place. Edgar has shown multiple times that the Tractrix beams to a point of non-usability at approximately 20 times it's flare rate. I have found the same to be true.

The claims you are making are misleading. It is impossible to get true horn loading to the point you have claimed. You are misleading several less experienced forum members into believing something that is not true because you do not understand your measurements. The attached graphic is from Don Keele's AES article on Low Frequency Horn Design Using Thele/Small Driver Parameters. There are several frequency break points along the bandwidth of a horn. Trying to go outside of these boundaries is a recipe for disaster.

It doesn't matter if you think it sounds "fine"; I'd be able to pick it out in an instant because I know exactly what it sounds like. I made the same mistakes 20 years ago. I did not begin to make really good horns until I accepted and followed what the horn greats laid down years before me.
 

Attachments

  • Horn_roll_off.png
    Horn_roll_off.png
    42 KB · Views: 609
Founder of XSA-Labs
Joined 2012
Paid Member
poor fidelity because it is full of distortion products and is not a faithful reproduction of the original signal

When you say distortion products, do you mean harmonic distortion? Below 1kHz I still have lots of panel vibrations because I am using thin foam core to prototype and it has not been braced. Above 1kHz, which I believe is above the mechanical resonance modes of the material, the harmonic distortion is quite low at circa <0.5% THD.

I hear what you are saying about the upper cutoff of a horn as described by the masters of horn design. What do you mean by
Tractrix beams to a point of non-usability at approximately 20 times it's flare rate ?

Here is the new EQ applied with polar data:

434705d1408852650-prv-5mr450-ndy-fast-applications-tractrix-polar-data-new-eq.png
 
Last edited:
Founder of XSA-Labs
Joined 2012
Paid Member
The design was for a 175Hz tractrix (Volvotreter's tractrix spreadsheet) and the driver has fs of 175hz. How would one check to see if it is loading? It is sealed up pretty well and the backs are sealed so all the output is coming from front horn. My PC sound card is fried otherwise I would do an impedance sweep to check for characteristic impedance plot showing loading. If you are asking whether or not the driver diaphragm is flapping around it is not - it has a very stiff suspension and is running through a 175hz -24dB/oct high pass filter. I was surprised to find that if I touched the cone it barely moves but a lot of sound pressure can be felt.
 
Last edited:
xrk971,

What you think you have there and what it really is, is two different things. I guess that makes me one of the "naysayers". Do you find it even remotely interesting that I predicted the drop off at 1400Hz? Do you know how I was able to predict that even before you built your horn? It's because I've been building horns for a very long time and know how they work. All the response above 1400Hz is a combination of cone break up and horn resonance. Everything above 1400Hz is not usable bandwidth and does not qualify as good fidelity. If you believe you're actually getting usable response out to 15KHz then I can only surmise you also believe Klipsch, Keele, Leach, Salmon, Edgar, Olson and Cohen are all idiots. All your claims fly in the face of what has been established as fact for a very long time.

"You can lead a horse to water, but you can't make them drink."

Back in the 90s, I built tractrix horns. I made my horns mostly via inspiration from people on the Internet. I didn't know anything about mass rolloff, or EBP, or any of the basic horn math.

This stuff takes a loooooooooooooooooooooooooooooooong time to learn.

Go to the first page of this very thread, and you'll see that GM (who is older and wiser) clued me in to why my horn project from 2006 didn't work.


Right now XRK972 is at a fork in the road:

1) The first option is to try and figure out how to extend the bandwidth of a driver as far as possible. This option eventually leads to back loaded horns that use drivers with light cones and carefully controlled break up. (Fostex, Lowther, Mark Audio.) Front loaded horns aren't good candidates for these designs because the horn in front of the driver rolls off the highs (which you are trying to preserve.)

2) The second option is to use multiple drivers on multiple horns within their optimum passband. The Synergy is a version of this; the Synergy is a horn on horn on a horn.
 
"You can lead a horse to water, but you can't make them drink."

Back in the 90s, I built tractrix horns. I made my horns mostly via inspiration from people on the Internet. I didn't know anything about mass rolloff, or EBP, or any of the basic horn math.

This stuff takes a loooooooooooooooooooooooooooooooong time to learn.

Go to the first page of this very thread, and you'll see that GM (who is older and wiser) clued me in to why my horn project from 2006 didn't work.


Right now XRK972 is at a fork in the road:

1) The first option is to try and figure out how to extend the bandwidth of a driver as far as possible. This option eventually leads to back loaded horns that use drivers with light cones and carefully controlled break up. (Fostex, Lowther, Mark Audio.) Front loaded horns aren't good candidates for these designs because the horn in front of the driver rolls off the highs (which you are trying to preserve.)

2) The second option is to use multiple drivers on multiple horns within their optimum passband. The Synergy is a version of this; the Synergy is a horn on horn on a horn.

What wil do a exponential or tractrix horn do with a wideband speaker of 3 inch and a woofer for below in a synergy setup in stead of a 3 inch alone? same problem with frequentie fall off, and when use a compression driver in stead, or a very good tweeter who have most of the time no theiel small parameters..

regards

kees
 

Attachments

  • Tractrix-build-2.png
    Tractrix-build-2.png
    350.8 KB · Views: 552
Founder of XSA-Labs
Joined 2012
Paid Member
It's interesting how folks who haven't tried a 5MR450NDY and who haven't coupled it to a tractrix with a 2.78 in square throat that is 22 in long x 28 in wide x 17 in tall with a sealed 5.25 liter back chamber, who haven't made measurements, who haven't listened to this particular setup in person, continue to say it doesn't work. I have been down the BLH road already. I have built other horns and have seen the limitations you describe. For some reason, this particular driver seems to work exceedingly well. There is clear 15khz reaching my ears and my mic even 4m away and it sounds clean. Less than 0.5% HD. Not sure why I am at a fork in the road, I think I found a road that for some reason wasn't open before because we did not have this driver before.
 
That horn looks fairly decent to me. Between 6k-9k it looks like there's some nastiness with the polar response, but from 3k-6k it looks like fairly constant directivity and then it widens out to omni below there. Exactly what you'd expect. Of course you have that problem at 1.4k in the response. It looks like below 6k you're about -6dB at 30 degrees, so 60 degrees overall. That's a moderately tight beam, but very useable if you set the system up right. Looks like you'd want a super tweeter above that - something with matching DI at the crossover ideally.

Figuring out where it's loading - look at the acoustic impedance predicted by hornresp. Otherwise you can measure it, but that's not as easy. Using a horn where the driver's power response is falling is not a bad thing. JLH et al do it all the time with their Unity / Synergy horns - most compression drivers have their power response rolling off above ~4kHz. You just need to make sure your driver isn't going into breakup. Easiest way to do that is to look at the frequency response of the raw driver (not on the horn) and look for big dips and peaks. Looking at the plot in the manufacturer's data sheet, it looks extremely smoothed, so I wouldn't pay too much attention to that. A better way would be to measure your horn using boinks and see if the boinks stretch on much longer in time than the input signal. See linkwitzlab.com for details. Or you could look at an energy time curve, but I always have less luck with those.
 
It's interesting how folks who haven't tried a 5MR450NDY and who haven't coupled it to a tractrix with a 2.78 in square throat that is 22 in long x 28 in wide x 17 in tall with a sealed 5.25 liter back chamber, who haven't made measurements, who haven't listened to this particular setup in person, continue to say it doesn't work. I have been down the BLH road already. I have built other horns and have seen the limitations you describe. For some reason, this particular driver seems to work exceedingly well. There is clear 15khz reaching my ears and my mic even 4m away and it sounds clean. Less than 0.5% HD. Not sure why I am at a fork in the road, I think I found a road that for some reason wasn't open before because we did not have this driver before.

And this is exactly what I would have said in 2004, or even 2008.

In 2014, I know better, and I agree with JLH.

You got some learnin' to do son.

The most valuable articles I've read are here : http://www.quarter-wave.com/Horns/Horn_Theory.html

If you don't believe that I made a lot of the same mistakes that you are now making, read this thread : http://www.audiogroupforum.com/csforum/showthread.php?t=62789

Jason is on that thread, so is Puggie, who started this thread. I would argue that this very thread was inspired by my Unity horn from 2006, and I was inspired by John Sheerin, who just responded to you.
 
Last edited:
Founder of XSA-Labs
Joined 2012
Paid Member
That horn looks fairly decent to me. Between 6k-9k it looks like there's some nastiness with the polar response, but from 3k-6k it looks like fairly constant directivity and then it widens out to omni below there. Exactly what you'd expect. Of course you have that problem at 1.4k in the response. It looks like below 6k you're about -6dB at 30 degrees, so 60 degrees overall. That's a moderately tight beam, but very useable if you set the system up right. Looks like you'd want a super tweeter above that - something with matching DI at the crossover ideally.

Figuring out where it's loading - look at the acoustic impedance predicted by hornresp. Otherwise you can measure it, but that's not as easy. Using a horn where the driver's power response is falling is not a bad thing. JLH et al do it all the time with their Unity / Synergy horns - most compression drivers have their power response rolling off above ~4kHz. You just need to make sure your driver isn't going into breakup. Easiest way to do that is to look at the frequency response of the raw driver (not on the horn) and look for big dips and peaks. Looking at the plot in the manufacturer's data sheet, it looks extremely smoothed, so I wouldn't pay too much attention to that. A better way would be to measure your horn using boinks and see if the boinks stretch on much longer in time than the input signal. See linkwitzlab.com for details. Or you could look at an energy time curve, but I always have less luck with those.

Thanks for the constructive comments and glad that you think there is some potential here rather than dismissing it. The manufacturer's frequency response data actually seems quite accurate and matches what I measured. Another member actually superimposed them on the other thread. For a driver that has a very stiff cone, stiff suspension, and huge Nd motor, would not horn loading it, reduce excursion below what it would be in a regular sealed alignment? And would this not further reduce HD due to small displacements?

Here is manufacturer's and my measurements superimposed:
6570224_orig.jpg
 
If I do akabak with this kind of horn, indeed I have a fall off above 3 Khz with a 3 inch wideband speaker. It behave not like a synergy who indeed can be seen as three horns in one system, or in my case two. I am however curious about what a exponentional flar does in synergy form, two sides are conic, and the width exponentional flare, so I can mount woofers also.

use a complete synergy script in akabak do give something whole different thing then a single tweeter in the same horn, making it a to long horn for that who not reach 15 Khz or higher.

Nice links Patrick, I like to learn, sometimes to much ;-).
 

Attachments

  • ScreenHunter_187 Aug. 25 00.50.jpg
    ScreenHunter_187 Aug. 25 00.50.jpg
    82.5 KB · Views: 475
Last edited:
I have fill in that horn with the 5MR450NDY speaker on hornresp and export to akabak for extend frequenty sim, and iot looks that it do fall off above 3 kiloherz.

I do not understand for such speaker with a big motor that qts is fairly high, heavy cone? then it can not go wel untill 15 kilohertz but the measure let see output there.

Horn need light cone, big motor and low qts, 0.20 is very good.

confusion sickness for me.

regards

kees
 

Attachments

  • ScreenHunter_189 Aug. 25 01.35.jpg
    ScreenHunter_189 Aug. 25 01.35.jpg
    83.9 KB · Views: 473
Kees,
What you describe is the 'radial horn' in AkAbak - it's a complex element with lots of switches and parameters so you will need to read the manual on that section to use it. It will simulate an exponential with two flat sides and two curved walls with a single element.

I have do export from hornresp to akabak to get higher frequenty sims.

I do not now much about akabak, make a script myself is not a option, to complicated, now that I am a amp designer and boxes are new for me, I do like the sybergy however and want to make it some day tegehter with a tapped horn for home use.

I did saw the exponential horn with two flat sides and thatw as interesting because on the flat sides I can put woofers for the low mid as synergy, however a normal conic synergy sim looks better.

I did use this idea in hornresp see picture, what you call a tractrix horn, how did you sim this? because making a synergy with it is a nice test, putting two 6 inch woofers extra on it..



regards

kees
 

Attachments

  • Tractrix-build-2.png
    Tractrix-build-2.png
    350.8 KB · Views: 462
Last edited:
Founder of XSA-Labs
Joined 2012
Paid Member
I modeled the tractrix in AkAbak using multiple segments of a comic horn. With enough short conics you can approximate any shape. I ended up using 16 segments. AkAbak is not hard to model horns - you just need to look at examples. In HR there are max 4 segments. Each one has a starting CSA and ending CSA and a length. Same in AkAbak - need to specify starting and ending CSA and length and type of segment but you can have about 50 segments total and you can branch segments and loop segments and connect the driver face and back to any arbitrary node anywhere. So it is very flexible. I will post the tractrix script in the other thread. Regarding the 5MR450ndy - it certainly goes well above 3k in a horn in simulation and in reality.
 
I (like most around here) have learned *so* much from JLH, Bateman and XRK. I want to bring a third perspective that might help move the discussion forward rather than it getting lodged at "there is only one proper way to do this and the masters already decreed it". :D

From talking with XRK I'm aware that he's not only read MJK's paper but understands the math. So rather than forcing credentials or experience on each other to prove points why don't we reconsider the stated goals and outcomes and look at this process as explorative rather than needing to fit into a mold.

Obviously, high end reproduction is a commonly shared goal by everyone participating but, like most work in audio it's a matter of design tradeoffs that lead to aesthetic decisions (after all the math and logic has been applied).

In looking at XRKs experiment (PRV+Tractrix) we see positives and negatives across several dimensions as compared to other ways to accomplish high sensitivity 300-15khz.

On another hand (yeah, there are 20), If we take JLH's purist approach we end up with 2 or more top horns (depending on how to-the-letter we want to take the bandwidth recommendations) and the required crossovers, time alignment and spacial positioning. This makes horns a difficult thing for the amateur to experience or explore as the investment in building and setup becomes large.

I hear JLH saying, "don't mislead people and give them a bad taste of what a horn is" while I see XRK making an accessible horn loaded project for people to quickly duplicate and enjoy. It doesn't seem there is any violation of principles going on, but it's certainly not conforming to horn loading above 1.6k as JLH has pointed out, but then is that necessary given the drivers response and is the constrained directivity of the HF enough to satisfy the user/application?

Further and this isn't great logic, but it's a real phenomenon, that geometries with high diffraction get used all the time, such as the jbl2360 and it's predecessor the altec mantaray and further the EV constant directivity horns and that proceeded them. They are not purist in their implementation of Leach's maths but the most recent versions such as the 2360 sound good (to most) and get the job done. (this without going into diffraction lenses, slots, paraline, etc). How much combing is occurring in this tractrix and how offensive is it compared to the distortion incurred by a 2360? Is it really offensive to consider this question?

Looking at XRKs project for my own uses I lean towards the inclusion of a phase plug to get HF loading to match the 300-1.4khz band, but it might be difficult to maintain above 10khz (intuition not math speaking)...

As I understand cone breakup, it has to do with point A and point B on the cone surface not arriving at the same time setting up transverse waves. It seems in a horn with phase plug and moderate compression that the much reduced travel and air resistance would help conform the cone's movement to a pistonic mode until very high SPLs are needed no?

As was pointed out earlier a falling response in horns is often the norm and bumping the HF with EQ isn't unheard of, but in XRK's horn it does seem to approach comfortable limits and is possibly accentuating the breakup modes at 7 and 9khz.

I do find it funny that 'prophecy' of a dip at 1.6khz has been extended to support JLH's dismissive gestalt. Come on, everybody knows the driver and throat width is going to create combing...
bonk.

The world didn't end. We might even enjoy this horn that leaves 1p equations and ideal wavefronts to those more interested in perfection_ and the tradeoffs they had to make in order to get it :D
 
Last edited:
Founder of XSA-Labs
Joined 2012
Paid Member
I think Anthony Bisset brought up a great point about the foam core fullrange tractrix horn being a very accessible horn for many folks. Especially people who already enjoy fullrange drivers and are familiar with their limitations. For many of us, having a fullrange driver that has -50dB levels or less of HD while playing fairly loud is quite a find. Here are the latest measurements showing the low HD once the constrained layer foam-caulking-foam is applied to the horn. More details here: http://www.diyaudio.com/forums/full-range/259293-prv-5mr450-ndy-fast-applications-13.html#post4036324

Here is the constrained layer damping (CLD) being applied:

434884d1408929326-prv-5mr450-ndy-fast-applications-tractrix-build-cld-6.jpg


434944d1408971865-prv-5mr450-ndy-fast-applications-tractrix-cld-new-eq.png
 
Last edited:
I modeled the tractrix in AkAbak using multiple segments of a comic horn. With enough short conics you can approximate any shape. I ended up using 16 segments. AkAbak is not hard to model horns - you just need to look at examples. In HR there are max 4 segments. Each one has a starting CSA and ending CSA and a length. Same in AkAbak - need to specify starting and ending CSA and length and type of segment but you can have about 50 segments total and you can branch segments and loop segments and connect the driver face and back to any arbitrary node anywhere. So it is very flexible. I will post the tractrix script in the other thread. Regarding the 5MR450ndy - it certainly goes well above 3k in a horn in simulation and in reality.

Hi

I see that these segments can be seen as what I do in sketchup to make the flare, I do use also short segments of say 1 cm.

I want not use your speaker, but a visaton frs 8 m with strong motor, I go put in these in the akabak script. I do use a old phillips for the bass, 8 inch put that also in.

FRS 8 M - 8 Ohm

regards

kees
 
Founder of XSA-Labs
Joined 2012
Paid Member
Hi

I see that these segments can be seen as what I do in sketchup to make the flare, I do use also short segments of say 1 cm.

I want not use your speaker, but a visaton frs 8 m with strong motor, I go put in these in the akabak script. I do use a old phillips for the bass, 8 inch put that also in.

FRS 8 M - 8 Ohm

regards

kees

Give it a try but note that the throat size was based on the dia of the surround for the 5MR450NDY. Use the Volvotreter spreadsheet for tractrix design and make your own flare for the Visaton.
 
When you say distortion products, do you mean harmonic distortion? Below 1kHz I still have lots of panel vibrations because I am using thin foam core to prototype and it has not been braced. Above 1kHz, which I believe is above the mechanical resonance modes of the material, the harmonic distortion is quite low at circa <0.5% THD.

I hear what you are saying about the upper cutoff of a horn as described by the masters of horn design. What do you mean by

Here is the new EQ applied with polar data

There is a combination of problems with your horn and what you are doing to it. When a horn appears to defy what it's directivity should be it means you have scatter problems within the horn that are not being constrained by the horn walls. Second, you are laying on heavy DSP processing to cover up major problems with the horn. You have taken a poorly performing horn and slapped on the DSP band-aid. If you have to make "corrections" bigger than a couple dB you have reached the limitations of where the horn should be used. Boosting in narrow bands using high Qs will cause Doppler distortion with neighboring frequencies. While it might measure flat, it will never sound right. You seem content on continuing down this path so I will not make any further comments on the matter.