You guys are starting to "get it". The proposed design would have a lot of HOM due to the "holes" in the side of the waveguide for the HF. I would never do this. To me an ABSOLUTE minimum of HOM and difraction above 1 kHZ. is an absolute requirement. The proposed design has numerous sources of both.
Hi Angeloitacare
I see your design and question about the holes in the walls.
So far as the holes, these have at least two observable effects.
It is desirable to make the holes as small as possible given the acoustic requirement, the holes are a “port” which combines with the trapped volume under the cone driving it.
This forms an acoustic low pass filter, which has to be set just above the electrical crossover. That Helmholtz filter, being connected to the same acoustic load, also shows up in the response of the hf driver as well. AS the upper and lower drivers are connected to the same acoustic load but separated in time slightly, both the acoustic magnitude and phases for both ranges and there electrical impedance mag and phase are needed to work out the crossover for the synergy alignment which eliminates the phase shift associated with a crossover. These must "hand off" to each other without interference but thankfully usually the mutual effects like the Helmholtz filter are partly / fully complimentary.
At high frequencies, one finds why there is a need to make the holes small, the larger they are, the more of a reflection they can produce higher up in frequency.
In experimenting with holes and placement, I found that these typically were not broad band effects but frequency / dimension specific.
The gap you show in your idea speaker would certainly cause a big problem as shown however, like the Synergy / Unity horns, that penetration into the horn should be as small as you can get away with.
Also, I found that the corners in a square or rectangular horn are "partially hidden" at high frequencies, if one places the holes in the corners, one can make them much larger than if they penetrated on the center of a flat surface, for a given pattern disturbance.
In other words (and at least for the conical horn I play with) , at a frequency where the horn has set the radiation angle, the corners in that horn are partially hidden, why a square horn can only project a partially square pattern up high..
Bottom line, as opposed to an exposed driver etc, the holes make a low pass filter which greatly reduces harmonic distortion from the driver, size, shape and placement of the holes allows the perturbation in the horn as small as possible.
Facing the need to have a wide angle horn speaker for work, I sampled a number of coax drivers, picked one that had the lowest degree of “problems” at the transition to the cone being the horn (showing up in the 3-5KHz range generally) and went high enough and laid out a horn that simply extended the cone profile.
The measured result is below ; although it appears the time was not quite right (rising phase at hf end) and is inverted, oh well.
If you open the CLF file for that speaker you can see that even now the transition in directivity's is not totally smooth, look in the 3-5KHz region on the spherical balloon, move it around, also by 16KHz, one can see that the inner horn in the pole is governing the pattern with the cone and outer horn essentially uninvolved.
http://www.danleysoundlabs.com/pdf/SH 100 Spec Sheet.PDF
Because of the compression driver behind the lf driver, I was able to make a Synergy style "phase less" crossover for it
Anyway, if your going to make what you have drawn, figure out how to make the mid entry as small as possible and in a round horn, symmetry is not your friend.
This wide angle baffle as you have drawn has a hidden advantage in that the pattern loss frequency goes down as the angle increases for a fixed width and that as you compensate for the horn gain in the 150-700Hz range, you are reducing the power going to the driver for a given SPL.
Best,
Tom Danley
I see your design and question about the holes in the walls.
So far as the holes, these have at least two observable effects.
It is desirable to make the holes as small as possible given the acoustic requirement, the holes are a “port” which combines with the trapped volume under the cone driving it.
This forms an acoustic low pass filter, which has to be set just above the electrical crossover. That Helmholtz filter, being connected to the same acoustic load, also shows up in the response of the hf driver as well. AS the upper and lower drivers are connected to the same acoustic load but separated in time slightly, both the acoustic magnitude and phases for both ranges and there electrical impedance mag and phase are needed to work out the crossover for the synergy alignment which eliminates the phase shift associated with a crossover. These must "hand off" to each other without interference but thankfully usually the mutual effects like the Helmholtz filter are partly / fully complimentary.
At high frequencies, one finds why there is a need to make the holes small, the larger they are, the more of a reflection they can produce higher up in frequency.
In experimenting with holes and placement, I found that these typically were not broad band effects but frequency / dimension specific.
The gap you show in your idea speaker would certainly cause a big problem as shown however, like the Synergy / Unity horns, that penetration into the horn should be as small as you can get away with.
Also, I found that the corners in a square or rectangular horn are "partially hidden" at high frequencies, if one places the holes in the corners, one can make them much larger than if they penetrated on the center of a flat surface, for a given pattern disturbance.
In other words (and at least for the conical horn I play with) , at a frequency where the horn has set the radiation angle, the corners in that horn are partially hidden, why a square horn can only project a partially square pattern up high..
Bottom line, as opposed to an exposed driver etc, the holes make a low pass filter which greatly reduces harmonic distortion from the driver, size, shape and placement of the holes allows the perturbation in the horn as small as possible.
Facing the need to have a wide angle horn speaker for work, I sampled a number of coax drivers, picked one that had the lowest degree of “problems” at the transition to the cone being the horn (showing up in the 3-5KHz range generally) and went high enough and laid out a horn that simply extended the cone profile.
The measured result is below ; although it appears the time was not quite right (rising phase at hf end) and is inverted, oh well.
If you open the CLF file for that speaker you can see that even now the transition in directivity's is not totally smooth, look in the 3-5KHz region on the spherical balloon, move it around, also by 16KHz, one can see that the inner horn in the pole is governing the pattern with the cone and outer horn essentially uninvolved.
http://www.danleysoundlabs.com/pdf/SH 100 Spec Sheet.PDF
Because of the compression driver behind the lf driver, I was able to make a Synergy style "phase less" crossover for it
Anyway, if your going to make what you have drawn, figure out how to make the mid entry as small as possible and in a round horn, symmetry is not your friend.
This wide angle baffle as you have drawn has a hidden advantage in that the pattern loss frequency goes down as the angle increases for a fixed width and that as you compensate for the horn gain in the 150-700Hz range, you are reducing the power going to the driver for a given SPL.
Best,
Tom Danley
gedlee said:
Angelo, this is a significant post, for several reasons. There is very little published on HOM's, multipath, internal reflections and diffraction in a waveguide or horn, and Dr. Geddes is the primary researcher on the topic. You could look far and wide for what has been expressed in the few sentences quoted above.
The other thing that is significant is the strong dependence on throat size. Dr. Geddes is saying he'd expect to see 5-8 modes in the audible band (I presume 1~20 kHz) for a driver with a 1" throat, and four times as many for a driver with a 2" throat. This means the HOM's depend not on diameter of the throat, but on its area.
The implications for a bass horn are severe, due to the square relation between the throat size and number of HOMs. Simple extrapolation tells us that a 5" driver will have 25 times as many, and larger drivers even worse, in geometric proportion. A bass horn with a cone driver is doomed from the start, unless you are willing to tolerate a very dense thicket of HOM's over the operating band. Maybe HOM's are less audible at lower frequencies - I don't know the answer to that one.
I'm grossly oversimplifying the picture, but the trend appears clear. In a horn or waveguide with essentially perfect transitions between the exit from the phase plug and a theoretically minimum-diffraction OS profile, you get the 5-8 modes mentioned above. With a 1" throat.
If there are nonuniform transitions between the exit from the phase-plug and the profile of the waveguide, you get many more HOM's, not less. Every bump and ripple between the HF section of the coax driver and the entry to the OS waveguide counts as a disruption and source of HOM's - even tiny mounting errors between the flange of a compression driver and the horn create reflections, much less grosser things like the multiple discontinuities between the cone flare, the ripples in the surround, the flat mounting flange, and the throat of the OS waveguide.
I dimly recall a post in another thread where Dr. Geddes mentioned in passing that the thickness of a coat of paint makes a difference; elsewhere in this thread he mentioned that the simple act of rotating the compression driver on the flange would move the polar pattern around.
Going back to personal experience, I've heard very expensive (US$10,000/pair) high-end horn systems where visibly poor finish in the throat region made the HF horn completely unlistenable from gross colorations, so I take Dr. Geddes comments quite seriously. The throat region of the horn is much more critical than mentioned in the traditional horn literature if low diffraction (and low coloration) is an important criterion.
What complicates the subjective discussion is that some long-time horn enthusiasts either don't hear, or don't mind, HOM's, so they can listen to traditional systems like the Altec A7 or the big professional horn-lens JBLs with complete enjoyment. For them, all this hairsplitting about HOM's probably just sounds silly, and a waste of time. You have to decide for yourself how you feel about horn colorations, and what your tolerance level for HOM's is.
The other thing that is significant is the strong dependence on throat size. Dr. Geddes is saying he'd expect to see 5-8 modes in the audible band (I presume 1~20 kHz) for a driver with a 1" throat, and four times as many for a driver with a 2" throat.
Dr.Geddes
might you explain : have you and is it possible to measure this?
This means the HOM's depend not on diameter of the throat, but on its area.
Does the area not depend on the diameter ?
The implications for a bass horn are severe, due to the square relation between the throat size and number of HOMs. Simple extrapolation tells us that a 5" driver will have 25 times as many, and larger drivers even worse, in geometric proportion. A bass horn with a cone driver is doomed from the start, unless you are willing to tolerate a very dense thicket of HOM's over the operating band. Maybe HOM's are less audible at lower frequencies - I don't know the answer to that one.
Lynn
( sorry if some thoughts are offtopic )
I don't think HOM's are a issue at low frequencies; the waves are long. This is proven by the fact, that horns with edgy foldings wont sound much different, ore not even any difference can be perceived, compared to horns with round , smooth foldings. What much more is a issue , is how low frequencies load the room, if the source can be easily detected. The Tannoy Autograph was one of the speakers, that left me remarkable impressions. It uses a backloaded basshorn, the bass is clean, radiating in a nicely integrated way and even into the room, since the horn mouth exit has vertically a height of about 1,4m. Nothing bothered me, pure enjoyment. I think basshorns with small mouth's close to the floor are also not so good solutions. The best i regard, if the low frequencies load the room evenly in vertical axis, from the ceiling to the floor, but also horizontally. But to accomplish that is not so easy. Multiple sub's, placed to the floor, as Dr.Gueddes suggests, i think that is very good , too.Direct radiating woofers can be fast and catch up with horns, but the small surface makes it easy to detect the source. Only multiple drivers solve this. A basshorn is ideal, since it usually has a big radiating area from the mouth, that alouds a even coupling to the room, and is more difficult to pointing out where the source is. AND they sound always fast and accurate. Time delay is a small disadvantage compared to the benefits.
.
I dimly recall a post in another thread where Dr. Geddes mentioned in passing that the thickness of a coat of paint makes a difference; elsewhere in this thread he mentioned that the simple act of rotating the compression driver on the flange would move the polar pattern around..
I do give far less importance to HOM's. It's indicative that most Audiophiles that listen to horns don't even know these exist.
I had a pair of Tractrix horns, with a BMS 4592nd coax compression driver, covering 300hz - 20khz. The treble had some disturbing sound, which bothered me, as more as i listened with this horn. Initially, i thought it was just because of beaming. But i think HOM were the cause of this. However, as soon as i replaced this solution with a separate horn for the midrange, and a supertweeter, this was no issue anymore, and treble was sounding satisfying again, . And the midrange, using a compression driver with tractrix horn, was ok as well. Of course, using a separate tweeter brings integration problems . So a OS wave guide , that covers from 1khz up, and resolves HOM , is indeed a very valid and good solution. What i still doubt, is , if there is a audible and relevant difference of a OS to a other Wave guide or horn of the same size , covering the same frequency band, in regard of HOM's. The most advantage of the OS is in my opinion the controlled and uniform dispertion.
Going back to personal experience, I've heard very expensive (US$10,000/pair) high-end horn systems where visibly poor finish in the throat region made the HF horn completely unlistenable from gross colorations, so I take Dr. Geddes comments quite seriously. The throat region of the horn is much more critical than mentioned in the traditional horn literature if low diffraction (and low coloration) is an important criterion.
I made a test of a compression driver with 1,4" exit that i bought recently, on a tractrix horn with 2" exit, just putting it close to the throat , and the discontinuity at the throat did not bother me at all.
I cannot say how the performance would differ, using a perfect throat transition, since i did not compare, but i doubt very much, at a blind listening test, someone would be able to detect a difference. I know anyone will agree on this with me, but i am being sincere; i am just reporting what i experienced.
Angelo
Lynn Olson said:
Actually I meant to 10 kHz. To 20 KHz the numbers would at least quadruple.
HOMs would be maximally audible in the range from 1 kHz to 10 kHz, peaking - predictably - at about 3 kHz. Below 500 Hz we would not be too sensitive to this group delay effect. But what you say about bass horns is quite true, they would have lots of HOMs since very few of them are designed with detail in mind.
"nonuniform transitions " increase the "level" of the HOMs not the number as you state, but I think that you realize that and simply mis-stated.
The situation with the subjective is complex. We come to like what we are used to and tend to find that pleasant - which is of course why the Summas don't audition well at shows. Horns have always had a great dynamic sound quality, but poor coloration and temporal quality. Anyone who denys this simply hasn't looked at the problem very much. If the horn proficianodos were to hear what a colorless waveguide sounded like then there would be no controversy about HOMs etc. This is why five of five audiophies who came to my home said that the speakers were the best that they had ever heard. This is not a coincidence.
angeloitacare said:
I am not so much an experimentalist, but these have been measured by numerous researchers in many different ways all over the world. You only need to review the literature to see this. There is no lack of evidence that they exist
angeloitacare said:
Ignorance is bliss I suppose.
If the horn proficianodos were to hear what a colorless waveguide sounded like then there would be no controversy about HOMs etc. This is why five of five audiophies who came to my home said that the speakers were the best that they had ever heard. This is not a coincidence.
i don't know if someone has made direct comparisons between tractrix, JMMLC, and OS Wave guide.
Since i have tractrix horns, a JMMLC is beeing made, it is very probable that i will afterwards make a OS Wave guide as well, and compare each other. I think this will be a very interesting experiment.
i don't know if someone has made direct comparisons between tractrix, JMMLC, and OS Wave guide.
Since i have tractrix horns, a JMMLC is beeing made, it is very probable that i will afterwards make a OS Wave guide as well, and compare each other. I think this will be a very interesting experiment.
Excellent idea, Angelo. I plan to audition - and measure in the time and frequency domain - several different types, including the OS, and with and without 35 ppi foam (and other damping materials).
The poor time-domain performance (with long decay intervals) of traditional (multicell, sectoral) and diffraction (Mantaray, BiRadial) horns is well known inside the industry, and a large part of the reason that TAD and JBL keep time-domain graphs proprietary (they've owned and used TDS and MLS systems for at least 20 years, they just don't publish the data). The poor time-domain performance is also the reason that 1/3 and 1/6 octave frequency-response "smoothing" (and before that, "pen damping") is commonly seen in prosound specifications - the raw, unsmoothed data would look much worse, with many narrow peaks and dips due to energy storage.
On the other hand, distortion performance, efficiency, and available headroom are an order of magnitude better than direct-radiators - this the reason for the subjective clarity and impact of horn systems. By improving the time-domain performance (instead of pretending it isn't a problem, as has been industry practice for decades), you can have the best of both worlds. Not many people have been doing this - Dr. Geddes is one of the first, if not the first, to really zero-in and focus on this area.
P.S. The reason I criticize the top-of-the-line audiophile speakers is that many of them are the worst of both worlds - the low efficiency and limited headroom of small-diaphragm direct-radiators, combined with poor time performance thanks to ill-chosen diaphragm materials and inept crossover design. (Brand "W", for example.)
The two best reasons to choose small-diaphragm direct-radiators are compact size (better domestic acceptance and a more rigid, less resonant cabinet) and a simpler path to good time-domain performance. A big, inefficient, direct-radiator with poor time performance has the worst of all types of speaker design - and this describes the majority of ultra-expensive speakers at hifi shows.
The poor time-domain performance (with long decay intervals) of traditional (multicell, sectoral) and diffraction (Mantaray, BiRadial) horns is well known inside the industry, and a large part of the reason that TAD and JBL keep time-domain graphs proprietary (they've owned and used TDS and MLS systems for at least 20 years, they just don't publish the data). The poor time-domain performance is also the reason that 1/3 and 1/6 octave frequency-response "smoothing" (and before that, "pen damping") is commonly seen in prosound specifications - the raw, unsmoothed data would look much worse, with many narrow peaks and dips due to energy storage.
On the other hand, distortion performance, efficiency, and available headroom are an order of magnitude better than direct-radiators - this the reason for the subjective clarity and impact of horn systems. By improving the time-domain performance (instead of pretending it isn't a problem, as has been industry practice for decades), you can have the best of both worlds. Not many people have been doing this - Dr. Geddes is one of the first, if not the first, to really zero-in and focus on this area.
P.S. The reason I criticize the top-of-the-line audiophile speakers is that many of them are the worst of both worlds - the low efficiency and limited headroom of small-diaphragm direct-radiators, combined with poor time performance thanks to ill-chosen diaphragm materials and inept crossover design. (Brand "W", for example.)
The two best reasons to choose small-diaphragm direct-radiators are compact size (better domestic acceptance and a more rigid, less resonant cabinet) and a simpler path to good time-domain performance. A big, inefficient, direct-radiator with poor time performance has the worst of all types of speaker design - and this describes the majority of ultra-expensive speakers at hifi shows.
My only concern with these tests is that you get the waveguide right. I actually ended up making them because of all the people who tried them and said that they didn't work well. When I looked into it I found out that they hadn't understood the concepts and had designed the waveguide wrong (they "improved it") then when it didn't work well said that my theory was "no good". My results prove that if you get it right the theory is very very good. But it is easy to get it wrong.
Here's an example of poor time performance, with long settling times. The speaker sounds impressively "fast", dynamic, and exciting to listen to, but is also quite colored on pink-noise and symphonic music, and has a diffuse, vague image quality, with no rendition of depth at all. The settling time is about 3 to 4 times longer, and the MF and HF drivers are out of alignment.
Attachments
