A dome isn't a good spherical source. At least a CD can supply a decent flat wavefront wiht efficiency.
does any manufacturer make affordable 3/8", 1/2" and 3/4" treble drivers?
Surely using much smaller moving elements than the 40mm to 50mm used in the 1" exit drivers will help overcome some of the problems being discussed?
Thanks.
With respect to 2), then I clearly don't understand Makarski's comments about the driver being the problem more than the horn. Maybe they relate back to his comments on 1). I'll go and re-read the paper tomorrow.
On 3), I phrased the issues/questions badly. I can see the HOMs start lower in frequency (and travel along the horn length but not parallel to the horn axis), but Makarski suggests that because these are much lower in dB than the main mode that they don't really matter until the 'power' in these modes exceeds that of the main mode (at about 14kHz for a 1 inch driver). He then clearly ties an on axis response dip at this point with an off axis directivity spread. Again, I'll check this tomorrow.
The OS generates the least of any device with two different angles at the throat and the mouth. The conical generates none, but the angles don't change.
This is the crux of the part that I think people have the most trouble wrapping their head around. A plane wave is the shape of the main mode at the entrance to a tube or an OS waveguide, but it is entirely wrong for the entrance to a conical horn. The main mode in a conical horn is a section of a sphere, which isn't even close to a plane wave. So what is a perfectly pure wave for one device is completely wrong for another. One must find that waveguide shape that has the correct main mode to match that of the driver.
The OS changes the shape of the main mode as it propagates from flat to spherical. The conical starts out spherical and remains spherical. The role of a waveguide is to shape the wave front exiting from the driver. The slower this is done the less HOM are created and the less reflection there will be. A diffraction slot does this instantly and creates large reflections and HOMs and poor sound quality. The goal then is what is the slowest that this can be done? The answer is the OS.
Another example is transitioning from a round cross section to a square or rectangular one. This will generate HOMs and reflections as well. Best is to not do it at all, but if you must then do it as slowly as possible - which, of course would then be an ellipse.
Last edited:
As stated, a dome vibrates axially and a spherical wave front vibrates radial. It is not a perfect match, but its not too bad. I did an exact analysis of this situation that is available on my website as an addendum to my book.
B&C makes a 3/4" driver. A 1" is a good compromise. The 3/4" would have trouble getting down below 1 kHz, but a good 1" can do that.
You have to understand what Markarski was looking at. His concern was the degree to which HOMs will change the directivity or frequency response of a waveguide. His conclusion was "Not much". That is probably correct, but it is not the question that I asked. I asked "Could an HOM be audible?" which is entirely different. We think we hear 1% THD (in some cases we can), is it so far fetched to think that maybe we could hear 1% HOM? 1% is not going to change the directivity at all, but it could be audible. So Makarski's comments may not always coincide with what we are talking about here. I brought him up to debunk the false claim that HOMs don't exist or have never been proven to exist. They have.
Could the desired wavefront/boundary interface be defined more organically to cylindrically challenged individuals? Is the goal something as simple as requiring the edge of the wavefront to remain normal to the boundary? Can the wavefront-geometry to waveguide-geometry be expressed in such a pictoral manner without cylindrical coordinates?
You stated that HOMs are less desirable the further into the horn (from the mouth?) they are generated. Ignoring any desired directivity criteria, would not, say, a longish hypex-like horn with very little curvature from the throat and well along the length to the mouth generate less detrimental HOMs than the OS, since the hypex has its greater curvature at the mouth?
Last edited:
Sorry, I don't know how to express the shapes that the various geometries take on. Spherical should be easy and planar as well, we see these every day. But the rest are mathematical derived topologies and the math is not trying to be kind to your imagination. It took ne years to envision what the contour and surfaces of the ellipsoid coordinates look like. I can see them in my head but I can't describe them in simple terms.
The edge of the wave front of the main mode is always normal to the boundary but you have to understand that this is hardly the only requirement. I can draw billions of curve for which the interior space is normal to the boundary, but it turns out that only 13 definable coordinate systems meet this requirement AND have 3 mutually orthogonal coordinates. 13 out of billions is a pretty exclusive club.
The derivation of these 13 coordinates is one of the most elegant in all of physics. This chapter (6) in Morse and Feshbach astounds me to this day. All 13 coordinates are shown in 3D at the end of this chapter. When you get through that chapter you really know that you have learned something.
I have always held that with the same tensor approach that Einstein used for General Relativity one could free the waveguide problem from this limited 13 coordinate problem. This is because his formulation is coordinate independent, as it had to be. But that task is well beyond my aging mental capabilities (just as a discrete time formulation of General Relativity was beyond Einstein later in life, but yet he knew that it would be important. Today they are a very hot topic in theoretical Physics.)
"Ignoring any desired directivity criteria" and the whole problem changes, to the point that its not worth talking about. To me "Desired directivity criteria" is the problem. You can't just exclude it and expect the discussion to still make sense.
The edge of the wave front of the main mode is always normal to the boundary but you have to understand that this is hardly the only requirement. I can draw billions of curve for which the interior space is normal to the boundary, but it turns out that only 13 definable coordinate systems meet this requirement AND have 3 mutually orthogonal coordinates. 13 out of billions is a pretty exclusive club.
The derivation of these 13 coordinates is one of the most elegant in all of physics. This chapter (6) in Morse and Feshbach astounds me to this day. All 13 coordinates are shown in 3D at the end of this chapter. When you get through that chapter you really know that you have learned something.
I have always held that with the same tensor approach that Einstein used for General Relativity one could free the waveguide problem from this limited 13 coordinate problem. This is because his formulation is coordinate independent, as it had to be. But that task is well beyond my aging mental capabilities (just as a discrete time formulation of General Relativity was beyond Einstein later in life, but yet he knew that it would be important. Today they are a very hot topic in theoretical Physics.)
"Ignoring any desired directivity criteria" and the whole problem changes, to the point that its not worth talking about. To me "Desired directivity criteria" is the problem. You can't just exclude it and expect the discussion to still make sense.
the question related to using a smaller treble driver to possibly reduce some of the problems.
Nothing to do with whether 1kHz midband ability is retained.
If I want a well performing treble horn, then why should I compromise that treble performance by forcing it to double up as a mid driver?
Reducing a problem is great, but not if it creates more problems. The 1" seems to be the sweet spot for home use. Covers the full upper range with a single driver. Larger or smaller can't do that.
I don't see why not. It makes sense to reduce HOMs in a horn, doesn't it? The discussion was about HOMs, not directivity.
It makes a difference to me. I'm not interested in horns that have no directivity control. They are not usable to me. So I don't think about, study or contemplate them. I guess I just have "no comment".
So the way I understand it, HOM's are reflections, like SWR in the RF world. Reflections can be in or out of phase, it would seem that these reflections would affect the frequency response of the system. To me the Horn has always seemed like a higher Q device than a direct radiator, and I think that this higher Q is what people don't like about the sound of horns. They are too good at reproducing the 2-4Khz range that the ear is most sensitive to. Very broad wideband low Q speakers seem to cause the least listening stress.
I have a set of Summas here, a set of Vandersteens, some JBL Control Nows.
The thing I notice with the Summas is that you can listen for longer, and at louder volume levels.
It's a bit of a catch-22, because a lot of the speakers that grab you by the balls are the same speakers that become fatiguing after 30 minutes of listening. The Summas don't do that. You just find yourself listening for hours and hours and hours...
I've built a lot of speakers, and there's this phenomenon where you listen to your music over a new set of speakers, and you notice a bunch of things that you didn't notice before. Perhaps the soundstage is wider or deeper, or the speaker brings out detail, or it's rhythmically 'tighter.'
You find yourself listening to album after album after album, with a newfound appreciation for songs you know and love.
The thing that sucks is that this phenomenon doesn't necessarily indicate that a speaker is *better*, only that it is *different.* IE, if you ate nothing but fish for a week and then had a steak, the differences would be more noticeable than if you ate noting but salmon for a week and then switched to trout.
I notice this with the Vandersteens and the Summas. They're both very different, so when I listen to one it tends to exaggerate weaknesses in the other.
I don't think that this is an accurate portrayal. HOMs can exist with no reflection at all and they can be created by a reflection, but the two things are quite different. Horns are not high-Q any more than a direct radiator and in fact if anything they are lower Q because the damping is much higher. Waveguides are "Very broad wideband low Q speakers".
The ultrasonic levitation demos that abound use a standing wave node to equalize gravity. A HOM is nothing more than a vector point of a full wavelength and is the same as a standing wave. If diffraction occurs a reflection and refraction results. This is also the same as in RF, where diffraction, impedance mismatch or damage will result in a high vSWR ratio. Your foam plugs through both diffraction and direct attenuation reduce this back reflection. In RF design we are normally limited to a 1 octave bandwidth design for optimal results. Waveguides and horns function the same way but are faced with the daunting task of the wide bandwidth and thus are more critical to HOMs SWR rarefactions.
A horn will increase the gain of the speaker just like a director and reflector will increase the gain of a driven element (a dipole). A horn be it an audio, or an RF horn narrows the radiation to increase the energy going in one direction, this increase the gain. An RF horn is high Q. Perhaps the rules for RF antennas and RF horns do not apply to audio horns. I do not feel that the horn is as wide band as a direct radiator, a horn always sound "peaked" compared to a direct radiator to me. I have felt that the gain and bandwidth limiting of the horn accounts for this "peaked" sound. Although a horn can measure flat within its operating range they always sound as if they have higher gain in the upper midrange. The higher the gain of the horn the more pronounced this effect is to me.
Squeezing all the sound in one narrow direction like a horn does, just does not sound as good to me as a direct radiator that is not narrowing its polar plot to increase gain. To me a very broad wideband low Q speaker is flat panel type of large size.
I will have to see if HOM's are talked about in the RF world, and called HOM's.
Squeezing all the sound in one narrow direction like a horn does, just does not sound as good to me as a direct radiator that is not narrowing its polar plot to increase gain. To me a very broad wideband low Q speaker is flat panel type of large size.
I will have to see if HOM's are talked about in the RF world, and called HOM's.
Thanks Greebster, I can tell you have worked in the RF world, as I have, that was a good post. So HOM's are standing waves.
- Status
- Not open for further replies.