I've owned a bunch of horns over the years, and I've built a TON of them.
About five years back I broke down and bough a set of commercial loudspeakers, a set of Gedlee Summas. One of the things that I noticed about the Summas was that they sounded different than my own designs, which had used the exact same curve for the waveguide.
I chalked it up to the roundover on the cabinet, and didn't give it much thought after that.*
This afternoon I was messing around with the wavefront simulator in Hornresp, and discovered something very interesting:
If there's any diffraction at the mouth of a waveguide, the wavefront doesn't radiate the way that it's supposed to.
This was interesting to me, as I'd always assumed that mouth treatment was like 'icing on the cake.' Something you might add to a loudspeaker to give it that last ten percent.
Go ahead and fire up hornresp and see it for yourself; draw any type of waveguide. Now add a big fat roundover to that waveguide.
The wavefront is completely different now isn't it?
^^ it's interesting that a lot of the hifi speakers don't have one
^^ but it appears that the prosound companies are aware of this
Even my $200 Kefs have a roundover on the waveguide
The easiest way to visualize this effect is with hornresp. If you want to mess around with it, do this:
1) Go into hornresp
2) Select the 'wavefront' simulator. It doesn't matter what the record says, because we're going to draw our own
3) Draw a shape in the wavefront a simulator. A flat baffle, a horn, whatever
4) Now add a roundover to that shape
Completely changes the wavefront doesn't it? It takes a chaotic wavefront and makes it smooth. It improves the polar response. It improves the on-axis response.
If there was ever a 'free lunch' in audio, this is it.
* that's not me, that's Earl Geddes and Duke LeJeune. I don't have many pics of my speakers, they're hard to photograph because they're so slippery and dark.
About five years back I broke down and bough a set of commercial loudspeakers, a set of Gedlee Summas. One of the things that I noticed about the Summas was that they sounded different than my own designs, which had used the exact same curve for the waveguide.
I chalked it up to the roundover on the cabinet, and didn't give it much thought after that.*
This afternoon I was messing around with the wavefront simulator in Hornresp, and discovered something very interesting:
If there's any diffraction at the mouth of a waveguide, the wavefront doesn't radiate the way that it's supposed to.
This was interesting to me, as I'd always assumed that mouth treatment was like 'icing on the cake.' Something you might add to a loudspeaker to give it that last ten percent.
Go ahead and fire up hornresp and see it for yourself; draw any type of waveguide. Now add a big fat roundover to that waveguide.
The wavefront is completely different now isn't it?
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
^^ it's interesting that a lot of the hifi speakers don't have one
^^ but it appears that the prosound companies are aware of this
Even my $200 Kefs have a roundover on the waveguide
The easiest way to visualize this effect is with hornresp. If you want to mess around with it, do this:
1) Go into hornresp
2) Select the 'wavefront' simulator. It doesn't matter what the record says, because we're going to draw our own
3) Draw a shape in the wavefront a simulator. A flat baffle, a horn, whatever
4) Now add a roundover to that shape
Completely changes the wavefront doesn't it? It takes a chaotic wavefront and makes it smooth. It improves the polar response. It improves the on-axis response.
If there was ever a 'free lunch' in audio, this is it.
* that's not me, that's Earl Geddes and Duke LeJeune. I don't have many pics of my speakers, they're hard to photograph because they're so slippery and dark.
Interesting.
I have a pair of 12" two-way PA speakers, where the CD horn sits proud (by about 1/4") of the front baffle. Adding a roundover would be difficult, as the front grill has to attach, but would dampening have a similar effect to a roundover?
Chris
I have a pair of 12" two-way PA speakers, where the CD horn sits proud (by about 1/4") of the front baffle. Adding a roundover would be difficult, as the front grill has to attach, but would dampening have a similar effect to a roundover?
Chris
Jean Michel L'Clearch (pardon me if I spelled it wrong again) showed the effect at the end of the waveguide clearly in the discussion of his design in that continuing thread, fwiw.
However, I don't think that is the "whole story".
_-_-bear
However, I don't think that is the "whole story".
_-_-bear
An externally hosted image should be here but it was not working when we last tested it.
Like this? Yeah, that's probably a fine idea.
Thanks for getting back so quickly.
I'll find some suitable materials and try the modification on one of the cabinets, and then I'll listen to the two next to each other to see how the sound has changed.
I'll find some suitable materials and try the modification on one of the cabinets, and then I'll listen to the two next to each other to see how the sound has changed.
All pictured speakers have bad transition from tweeter to horn's throat. Air turbulence created at the throat will be amplified. Also, shape of compression driver phase plug is very important, if turbulence created at that stage, none of the horn designs will help. I think Summas sound the best, because of foam plug, it levels out all air turbulence.
Last edited:
Back in the 1970's a certain Mr. Iwata in Japan came to similar conclusions. The story goes that he was given the task of fixing the shock waves caused by the Bullet Train coming out of tunnels at high speed. Apparently quite a problem. He came up with a series of slots or vanes that tapered at the edge of the tunnel. Problem solved, shock wave killed.
He then turned his attention to his horn system, added the same series of vanes at the exit and smoothing the abrupt transition form horn to air. I've seen photos of his system, all the horn mouths are done that way.
As Bear stated, Jean-Michel LeCleac'h took this further and came up with a profile that has a large round-over at the edge, making for a better and less abrupt exit. "Rabid Horns" (foam at the mouth) can also do a lot to smooth the transition. You can test it ourself with a couple of towels at the mouth.
I have no idea what works best, soft edge, rounded edge or a combo of both. They do seem to work, tho.
He then turned his attention to his horn system, added the same series of vanes at the exit and smoothing the abrupt transition form horn to air. I've seen photos of his system, all the horn mouths are done that way.
As Bear stated, Jean-Michel LeCleac'h took this further and came up with a profile that has a large round-over at the edge, making for a better and less abrupt exit. "Rabid Horns" (foam at the mouth) can also do a lot to smooth the transition. You can test it ourself with a couple of towels at the mouth.
I have no idea what works best, soft edge, rounded edge or a combo of both. They do seem to work, tho.
Went out and got a decent-sized soft blanket for £6.
I'll get the scissors out and sort something out. Hopefully I'll get some photos/impressions up tonight.
I'll get the scissors out and sort something out. Hopefully I'll get some photos/impressions up tonight.
Foam is a semi useless material (imho) for absorption at the exit of anything HF. Perhaps better than nothing in one regard, but as pictured by Patrick, it creates an abrupt transition and assuming it is absorptive at HF it serves to effectively shorten the horn.
As a thought experiment you can imagine applying a 100% effective absorptive material (imaginary stuff) that would be the equivalent of hacking off the end of the horn! Of course it would go hacked off into infinite space...
It's best to just fabricate a properly designed horn than to hack one that is not quite right...
As far as the Geddes foam plug, there are various thoughts about how that really functions... one thing for sure it is an acoustic LP filter, a resistance and certainly attenuates any reflections within the horn volume (they have to pass through the foam more than once).
As a thought experiment you can imagine applying a 100% effective absorptive material (imaginary stuff) that would be the equivalent of hacking off the end of the horn! Of course it would go hacked off into infinite space...
It's best to just fabricate a properly designed horn than to hack one that is not quite right...
As far as the Geddes foam plug, there are various thoughts about how that really functions... one thing for sure it is an acoustic LP filter, a resistance and certainly attenuates any reflections within the horn volume (they have to pass through the foam more than once).
I dunno Bear, foam or felt or something like that which actually extends the profile of the horn, while at the same time rounding and softening it does well pretty well. Inside the horn, maybe not so much.
There were some techniques from the early days that used a rough wall surface to create a cushion of turbulent air right along the walls. Or so went the story. No idea if it actually did anything.
There were some techniques from the early days that used a rough wall surface to create a cushion of turbulent air right along the walls. Or so went the story. No idea if it actually did anything.
Assuming the transition is essentially perfect there may be real benefit.
Of course if it sounds good, it is good. 😀
Think golf balls?
Of course if it sounds good, it is good. 😀
Think golf balls?
Oh, yeah - golf ball dimples. Good thinkin'! You know Myth Busters did a golf ball dimple car and much to their amazement - it worked! (better gas mileage)
Horn dimples. Hmmmmmmm................
Horn dimples. Hmmmmmmm................
Horns/waveguides are not fundamentally flawed; but many/most have poor implementation regards coupling to free space.
Diffraction is well covered; Linkwitz provides very nice web page summarizing the fundamentals.
Straight sided is the hardest case. The simplest straight sided waveguide is circular disc. Assuming <1.5" radiator at center, results will be very similar to Linkwitz's 12" circular baffle measurement:
With increasing frequency, driver beaming reduces energy seeing edge of baffle and the radiation impedance drop from 2pi space to 4pi space and the energy reflected is lower, reducing magnitude of response ripple at measurement point.
As a conical profile is formed energy seen by the baffle is redirected. As cone angle decreases a proportional reduction of radiation occurs from 2pi space conditions, increasing the radiation impedance drop to 4pi space at the cones edge. The waveguide results in energy being focused, and in waveguide seeing more high frequency energy from the driver, leading to more ripple at higher frequencies. Focusing and low frequency reflective driver loading improve steady state low frequency response.
If waveguide is inserted into infinite baffle, impedance drop reduces to going to 2pi space instead of 4pi space. As flat baffle is reduced in size to that of typical box speaker, baffle step returns. In typical little box speaker with small dome tweeter, box edges with edges smoothed with radius of 1/4 wavelength are easy to achieve and are effective for frequencies seen at edge of flat baffle. For 12" focusing waveguide ripple is seen strongly down to 1kHz zone, and radius needs to be on order of 3".
Diffraction from circular opening makes mouth region of waveguide at increasing frequencies behaves as ring radiator that is coaxial to driver. Apparent source location and width becomes both frequency and spatially dependent. This impacts frequency range for directional cues of both temporal and intensity basis. Spectral coloration is obvious result. Image smear is another result.
Diffraction is well covered; Linkwitz provides very nice web page summarizing the fundamentals.
Straight sided is the hardest case. The simplest straight sided waveguide is circular disc. Assuming <1.5" radiator at center, results will be very similar to Linkwitz's 12" circular baffle measurement:
With increasing frequency, driver beaming reduces energy seeing edge of baffle and the radiation impedance drop from 2pi space to 4pi space and the energy reflected is lower, reducing magnitude of response ripple at measurement point.
As a conical profile is formed energy seen by the baffle is redirected. As cone angle decreases a proportional reduction of radiation occurs from 2pi space conditions, increasing the radiation impedance drop to 4pi space at the cones edge. The waveguide results in energy being focused, and in waveguide seeing more high frequency energy from the driver, leading to more ripple at higher frequencies. Focusing and low frequency reflective driver loading improve steady state low frequency response.
If waveguide is inserted into infinite baffle, impedance drop reduces to going to 2pi space instead of 4pi space. As flat baffle is reduced in size to that of typical box speaker, baffle step returns. In typical little box speaker with small dome tweeter, box edges with edges smoothed with radius of 1/4 wavelength are easy to achieve and are effective for frequencies seen at edge of flat baffle. For 12" focusing waveguide ripple is seen strongly down to 1kHz zone, and radius needs to be on order of 3".
Diffraction from circular opening makes mouth region of waveguide at increasing frequencies behaves as ring radiator that is coaxial to driver. Apparent source location and width becomes both frequency and spatially dependent. This impacts frequency range for directional cues of both temporal and intensity basis. Spectral coloration is obvious result. Image smear is another result.
It's good that you have this sorted.
I've been studying mouth terminations for a few years and this is my build thread, posted today.
http://www.diyaudio.com/forums/multi-way/243680-build-thread-corner-guided-3-way.html
it works by creating air turbulence
don't know if you want that
also, I think it needs a minimum of speed/air movement to create this effect
they tried it on surfboards once
Okay, so I've set both up, one with blanket, one without.
I layered the blanket up so that it was level with the protruding horn, then put a wider bit over the top so that the outer 1" (roughly) of the horn was covered.
The result is a subtle improvement. The hi-hat work on Buena by Morphine and the guitar on Keith Don't Go by Nils Lofgren were enough to convince me. There seemed to be slightly more clarity there, as if some frequencies that were previously attenuated have been brought into line. I don't have any measurements to back this up, just my ears. I'll listen for a while to double-check my findings, and then consider doing the second one.
I used gaffer tape to hold the blanket down, which of course isn't a particularly permanent solution. I'd like something that can be removed if necessary, but is more secure. Any ideas?
Cheers
Chris
I layered the blanket up so that it was level with the protruding horn, then put a wider bit over the top so that the outer 1" (roughly) of the horn was covered.
The result is a subtle improvement. The hi-hat work on Buena by Morphine and the guitar on Keith Don't Go by Nils Lofgren were enough to convince me. There seemed to be slightly more clarity there, as if some frequencies that were previously attenuated have been brought into line. I don't have any measurements to back this up, just my ears. I'll listen for a while to double-check my findings, and then consider doing the second one.
I used gaffer tape to hold the blanket down, which of course isn't a particularly permanent solution. I'd like something that can be removed if necessary, but is more secure. Any ideas?
Cheers
Chris
I've used towels rolled up and sewn into a tube. Also thin felt tube with poly stuffing. Microfiber towels seem to work well.
The change can be subtle, but it's measurable. On my Altec 811 horns the towel mouth removed a kink in the response at 1KHz. You'll hear that. I suppose, but didn't measure, that diffraction was lowered.
The change can be subtle, but it's measurable. On my Altec 811 horns the towel mouth removed a kink in the response at 1KHz. You'll hear that. I suppose, but didn't measure, that diffraction was lowered.
Hi Pano,
Perhaps I didn't word my post so well. I'm looking for ways of holding the blanket to the cabinet.
Perhaps a staple gun?
Chris
Perhaps I didn't word my post so well. I'm looking for ways of holding the blanket to the cabinet.
Perhaps a staple gun?
Chris
Really! Care to share what those might be? I only know of one. It would be great to know the other ones.
- Status
- Not open for further replies.