Just some thoughts,
I remember reading one of Dr. Geddes' articles, wherein he discusses higher order modes. He concludes that HOMs are linear, however their audibility is level dependent and is non-linear.
Geddes, Le'Cleach, and others have discussed waveguide optimization on this forum. However, I'm not sure if their simulations include the full compressible flow of newtonian fluids equations (ie Navier-Stokes) or simply the progression of the pressure wave.
The reason I ask...
Waveguides contribute significant sound pressure levels, INTENSE levels...even at appreciable distances. I believe that is one of their main attributes. At a point in time (before it reaches your ears), all of that energy is abruptly compressed into the throat and then forced to abruptly expand. I would expect vortices and eddies to form at the boundaries (ie turbulence) under such circumstances, which is VERY level dependent (ie highly non-linear, chaotic). I would also expect this fact to lead to a gross error in simulations which do not include it, other than low levels (laminar flow).
Perhaps (significant?) turbulence in the throat is another reason horns may *sound good* at low levels (laminar flow) and *harsh* at high levels (turbulent flow).
Just thinking out loud...
Best Regards,
Thadman
I remember reading one of Dr. Geddes' articles, wherein he discusses higher order modes. He concludes that HOMs are linear, however their audibility is level dependent and is non-linear.
Geddes, Le'Cleach, and others have discussed waveguide optimization on this forum. However, I'm not sure if their simulations include the full compressible flow of newtonian fluids equations (ie Navier-Stokes) or simply the progression of the pressure wave.
The reason I ask...
Waveguides contribute significant sound pressure levels, INTENSE levels...even at appreciable distances. I believe that is one of their main attributes. At a point in time (before it reaches your ears), all of that energy is abruptly compressed into the throat and then forced to abruptly expand. I would expect vortices and eddies to form at the boundaries (ie turbulence) under such circumstances, which is VERY level dependent (ie highly non-linear, chaotic). I would also expect this fact to lead to a gross error in simulations which do not include it, other than low levels (laminar flow).
Perhaps (significant?) turbulence in the throat is another reason horns may *sound good* at low levels (laminar flow) and *harsh* at high levels (turbulent flow).
Just thinking out loud...
Best Regards,
Thadman
Last edited:
I'm a bit rusty on my fluid flow theory and haven't done any cfd in a few years, but I don't think this would be the cause of any effect you would see in high frequency horns. The air molecules are moving very small distances, bunching up and spreading out. But I think to say they're flowing anywhere is a bit of a stretch. On the other hand in bass reflex ports, that is a more appropriate way to look at things if you're interested in high output levels and reducing compression effects. JBL as well as some others applied CFD to this area in some AES papers. The thing to remember is that the air is flowing both ways through the port which complicates things when searching for a good solution.
As I remember, the main effect of high sound pressures in the throat of horns is due to local increases in the density of the air leading to changes in the speed of sound. This can cause sine waves to change to triangle waves. I believe Tom Danley posted about this, possibly at the AA HE forum, possibly in relation to acoustic levitation. He might have posted some SPL values where this started occurring.
As I remember, the main effect of high sound pressures in the throat of horns is due to local increases in the density of the air leading to changes in the speed of sound. This can cause sine waves to change to triangle waves. I believe Tom Danley posted about this, possibly at the AA HE forum, possibly in relation to acoustic levitation. He might have posted some SPL values where this started occurring.
The pressures are really quite small compared to those experienced by, for example, an airplane wing. 120 dB is only about .003 psi. As I recall from my basic aerodynamics courses, you are pretty safe treating air like an incompressible fluid as long as the flow velocity is well below the speed of sound.
skeptic43 said:
An excellent assumption. However, I have not seen it discussed on this forum.
John Sheerin said:
Interesting
I understand the pressures will be low. However, I believe the CD is a unique case.
The compression driver rapidly accelerates the fluid creating a pressure wave. The phase plug-throat-waveguide abruptly compresses the pressure wave into a very small area and then abruptly expands it, possibly leading to high fluid velocity in the throat. However, although excursion may be small, it possess tremendous acceleration.
The mass of air (ie molecules) within the waveguide (coupled to the diaphragm) oscillates. This requires that the fluid flow over the boundaries of the waveguide. My thinking was that the tremendous acceleration (instantaneous velocity) of the fluid may lead to a high Reynolds number (ie turbulence), which may disjoint the boundary fluid layer and form vortices and eddies. This effect may be pronounced at high levels and may be absent at low levels (laminar flow).
I believe turbulence is observed (ie chuffing) in ports at >20m/s. Wouldn't this be significantly lower than the speed of sound?
Last edited:
With regards to psychoacoustics...
Beyond 20ms, energy is processed as separate from the initial transient. It is perceived as ambient and is often preferred by listeners. However, it was observed that if the energy is much closer in time (<1ms) to the initial transient, a significant degradation in sound quality is observed.
I believe this is because pressure waves naturally encounter obstacles in nature. We have evolved to process the initial transient separate from the obstacles effects. However, if the obstacle is very close, we may not be able to make such a distinction and the initial transient is smeared. I believe the throat of the compression driver will contribute effects very close in time to the initial transient.
I'm aware that pressure waves are refracted through temperature gradients. Perhaps similar effects will be observed as a pressure wave passes through turbulence. I would expect the forces to be reactive.
Beyond 20ms, energy is processed as separate from the initial transient. It is perceived as ambient and is often preferred by listeners. However, it was observed that if the energy is much closer in time (<1ms) to the initial transient, a significant degradation in sound quality is observed.
I believe this is because pressure waves naturally encounter obstacles in nature. We have evolved to process the initial transient separate from the obstacles effects. However, if the obstacle is very close, we may not be able to make such a distinction and the initial transient is smeared. I believe the throat of the compression driver will contribute effects very close in time to the initial transient.
I'm aware that pressure waves are refracted through temperature gradients. Perhaps similar effects will be observed as a pressure wave passes through turbulence. I would expect the forces to be reactive.
Last edited:
Turbulence can be defined for an incompressible fluid too -- water has turbulence when it flows over a rock. Just doing it in my head, the max peak velocity in the throat of a 1" waveguide would be something on the order of 4 m/s or 1% of the speed of sound so I don't think you'd need to get too fancy figuring out how it reacts to the horn shape.
Of course you could use way too much power and make it sound bad if you wanted to but the whole point of a Geddes-style waveguide vs. a traditional horn is sound quality, not max SPL.
About Tom's levitation experiments, as I recall he was looking at 160+ dB SPL levels to get things to float and I'm sure that causes all kinds of audio nasties.
With regards to foam placed within the waveguide,
I'm aware that Geddes considers it simply as a resistive element for mathematical simplicity. However, due to its highly complex microstructure, I believe it may have other effects which may not be easily ignored.
I have access to several journal articles through my Universities research library. Numerical simulations which agreed most with experimental results modeled it as a *bed*. The fluid is forced through the windows of the foam between the ligaments. This rapid compression and expansion has several effects beyond a simple resistance. The foam structure at the surface can result in disruption of the boundary layer (ie turbulence).
I'm not aware of any experiments which could determine turbulence in the throat, other than numerically.
I'm aware that Geddes considers it simply as a resistive element for mathematical simplicity. However, due to its highly complex microstructure, I believe it may have other effects which may not be easily ignored.
I have access to several journal articles through my Universities research library. Numerical simulations which agreed most with experimental results modeled it as a *bed*. The fluid is forced through the windows of the foam between the ligaments. This rapid compression and expansion has several effects beyond a simple resistance. The foam structure at the surface can result in disruption of the boundary layer (ie turbulence).
I'm not aware of any experiments which could determine turbulence in the throat, other than numerically.
How did you determine the max peak velocity in the throat?
Have you considered the channels within the phase plug as a possible source of turbulence?
I have looked through Geddes' phase-plug patent application. All of his examples force the fluid through incredibly small ducts before expanding into the throat.
About Tom's levitation experiments, as I recall he was looking at 160+ dB SPL levels to get things to float and I'm sure that causes all kinds of audio nasties.
Sounds like he's having too much fun
I'm not attacking Dr. Geddes' research or the design of his Summa. For someone with his research experience and passion, I would feel it would be arrogant to assume he hasn't ranked the engineering problems in order of importance. His research has raised loudspeaker design to a much higher level. As such, I expect his speakers to sound fantastic
I'm simply trying to apply whats in my head to reality. I spent 4 hours driving home from the Colts game last night in the ice/snow and my mind found something to ponder upon. I find fluid behavior to be very interesting.
Last edited:
Crude in-the-head calc -- 1 mm xmax * 1 kHz * 4 (2" diaphragm squished into a 1" throat) = 4000 mm/sec. At higher frequencies at the same SPL, the stroke is shorter but the frequency is higher so peak velocity stays the same.
I thought this thread was about how to model waveguides, not compression drivers and phase plugs. For waveguide modeling, assume a flat piston is pushing air into the waveguide.
With regards to turbulence, absolute velocity is significant. However, I believe in the case of a compression driver, the velocity differential is much more significant. The maximum velocity may occur at the center of the ducts and a zero velocity (contributed by a single layer of air molecules) will occur at the boundary. This differential may lead to large fluid strains and result in disruption of the boundary fluid layer.
This thread is about optimization. We must consider the total system if we are to approach a global optimum solution. However inconvenient, I'm not sure its appropriate to separate the optimization of the compression driver from the waveguide.
Last edited:
it is actually easier than you think.........
3 rules to lower reflections ..........
no parallel walls.
no corners.
and no sharp breaks in the flare.
spherical horns should be best, but you also get a large standing wave due to the same distance across the mouth (like a square 12" x 12" mouth). You see impedence peaks on a horn/driver graph that are the standing waves of length to mouth and also height and width of the horn mouth.
For example I use a very good ev hr90 at 1khz
http://archives.telex.com/archives/EV/Horns/EDS/HR90 EDS.pdf
But I think the 18sound xt1464 should be better, not sure if I can cross it at 1khz.
To me I cross a horn where the d.i. radically drops off.
It has smooth walls and an eliptical mouth.
XT1464 - Constant Coverage HF Horn
here should be a great horn for 2khz (threaded)
I thought I wanted the spherical tractrix from stereolabs but the contour means you cannot go over the 10:1 freq range (i.e. the 400hz one) it really is 600hz to 6khz, while I run from 1khz up to 20khz on the same horn. Stereo lab makes a good contour for a 3 way (think avante garde loudspeakers).
To see the freq response of a horn and driver, simply overlay the plane wave tube response of the driver over the d.i. plot of the horn. If the di shoots up (like many horns) you will get a boost there, and where the horn d.i. falls off, the freq response also falls off, thats why I cross my huge 1' x 2' wide horn at 1khz. But unless the di is smooth, you will not get that response farfield or 10' away where the off axis freq response is what you hear. Or you will lose highs as you move back into the room or off axis. My 1' x 2' horn has the depth and mouth area for a bit lower but the vertical directivity falls off below 2khz. I tried crossing over lower, it sounded worse. The movie theater jbl horns that cross at 500hz have a mouth 3' tall by 3' wide. That's what it takes to keep directivity there. But those have a diffraction slot, parallel walls at the throat (yuck).
Norman
3 rules to lower reflections ..........
no parallel walls.
no corners.
and no sharp breaks in the flare.
spherical horns should be best, but you also get a large standing wave due to the same distance across the mouth (like a square 12" x 12" mouth). You see impedence peaks on a horn/driver graph that are the standing waves of length to mouth and also height and width of the horn mouth.
For example I use a very good ev hr90 at 1khz
http://archives.telex.com/archives/EV/Horns/EDS/HR90 EDS.pdf
An externally hosted image should be here but it was not working when we last tested it.
But I think the 18sound xt1464 should be better, not sure if I can cross it at 1khz.
To me I cross a horn where the d.i. radically drops off.
It has smooth walls and an eliptical mouth.
XT1464 - Constant Coverage HF Horn
An externally hosted image should be here but it was not working when we last tested it.
here should be a great horn for 2khz (threaded)
I thought I wanted the spherical tractrix from stereolabs but the contour means you cannot go over the 10:1 freq range (i.e. the 400hz one) it really is 600hz to 6khz, while I run from 1khz up to 20khz on the same horn. Stereo lab makes a good contour for a 3 way (think avante garde loudspeakers).
To see the freq response of a horn and driver, simply overlay the plane wave tube response of the driver over the d.i. plot of the horn. If the di shoots up (like many horns) you will get a boost there, and where the horn d.i. falls off, the freq response also falls off, thats why I cross my huge 1' x 2' wide horn at 1khz. But unless the di is smooth, you will not get that response farfield or 10' away where the off axis freq response is what you hear. Or you will lose highs as you move back into the room or off axis. My 1' x 2' horn has the depth and mouth area for a bit lower but the vertical directivity falls off below 2khz. I tried crossing over lower, it sounded worse. The movie theater jbl horns that cross at 500hz have a mouth 3' tall by 3' wide. That's what it takes to keep directivity there. But those have a diffraction slot, parallel walls at the throat (yuck).
Norman
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.