Or just use a large baffle! 🙂
It may not be narrow and sleek looking, and goes against modern design trends, but a moderately wide baffle can put the baffle step frequency below this range, constraining the entire midrange to no greater than half space.
For example a 50cm wide baffle puts the baffle step frequency at ~230Hz, below the critical parts of the midrange, and approximately at the Schroeder frequency of most listening rooms. Below that modal smoothing is more relevant than directivity anyway, and we are perceiving the steady state response.
Sure, a 50cm wide baffle would look rather silly with small drivers, but if you have a 12 or 15" woofer in a 3/4 way design it can look well proportioned.
The advantage over both the dipole and cardioid (above the 230Hz baffle step frequency) is that you have a DI of 6dB instead of 4.8dB, and unlike the dipole you have no reverse phase radiation to the rear that is going to bounce off the front wall and cause comb filtering if the speaker is too close to the wall.
On the contrary, a wide baffle speaker can be moved quite close to the front wall (closer than a narrow baffle design) with very little detriment to tonal balance or imaging in the midrange, because there's little to no midrange radiating to the rear of the speaker to generate comb filtering or time delayed reflections, as well as front wall proximity not boosting the lower midrange as it would on a narrow baffle design. (You would still have to equalize the bass with close positioning though)
There are drawbacks, like increased time delay of edge diffraction that would be noticeable at high frequencies if not addressed, but those can be solved by using controlled directivity drivers at higher frequencies.
Last edited:
Acoustic Treatment
In my slightly small sized living room (maybe 13 X 18 X 9), I nailed 2 inch cotton rope into 5 corners (along 3 sides at the ceiling, and ceiling to floor at the front wall), and it dramatically reduced the approx. 15mS slap echo and substantial ringing that was blatently obvious when you stood anywhere and snapped your fingers or clapped your hands. It seems that the corners are the main culprit in rooms this size. That ringing sounded primarily midrange. I think the cancellations in the lower mid and bass are a whole different thing, and can be improved by the use of flanking drivers in that frequency range. I find that my center channel speaker helps this particular issue a bunch. Some say you don't need a center speaker to have a good phantom image, but I find that it's real helpful in the lower mid and bass cancellation region. My 42 inch HD LCD TV sits on top of it.
In my slightly small sized living room (maybe 13 X 18 X 9), I nailed 2 inch cotton rope into 5 corners (along 3 sides at the ceiling, and ceiling to floor at the front wall), and it dramatically reduced the approx. 15mS slap echo and substantial ringing that was blatently obvious when you stood anywhere and snapped your fingers or clapped your hands. It seems that the corners are the main culprit in rooms this size. That ringing sounded primarily midrange. I think the cancellations in the lower mid and bass are a whole different thing, and can be improved by the use of flanking drivers in that frequency range. I find that my center channel speaker helps this particular issue a bunch. Some say you don't need a center speaker to have a good phantom image, but I find that it's real helpful in the lower mid and bass cancellation region. My 42 inch HD LCD TV sits on top of it.
I should add that the 2 inch cotton rope that I bought at Fabric Depot is VERY flamable. Auralux makes some foam that would be real good and has the legal amount of fire retardant in it. It comes in 2, 3, and 4 inch square by 2 feet long chunks. Not too expensive. I guess they use a spray glue to hold it in place. If the room is so large that echos get longer than 50mS, where speech intellegability gets bad, then you might want to consider covering an entire end wall with damping material.
I checked out Elias wavelet software page and I got confused. I guess it's not clear to me what the test signal is and how it's being used. I'm not up on that and would need more explanation. Perhaps more coffee too.
I've often wondered if tone bursts wouldn't be a great way to analyze a room (maybe that's what Elias is doing), looking specifically at start up and decay times of ringing, over frequency. People say that the cumulative spectral decay 3-D graphs are of limited use or accuracy, but I suspect that something like that done with tone bursts could be very revealing when used to analyze listening room acoustics. I theorize that a frequency that rings will be perceived louder than would show using pink noise (impulsive) and a calibrated mic. But because there's also start up time with ringing, it gets more complicated. I was thinking a gaussian burst envelop might be ideal. Linkwitz seems to like the "Blackman" tone burst better. They're very similar in effect. Any thoughts?
I've often wondered if tone bursts wouldn't be a great way to analyze a room (maybe that's what Elias is doing), looking specifically at start up and decay times of ringing, over frequency. People say that the cumulative spectral decay 3-D graphs are of limited use or accuracy, but I suspect that something like that done with tone bursts could be very revealing when used to analyze listening room acoustics. I theorize that a frequency that rings will be perceived louder than would show using pink noise (impulsive) and a calibrated mic. But because there's also start up time with ringing, it gets more complicated. I was thinking a gaussian burst envelop might be ideal. Linkwitz seems to like the "Blackman" tone burst better. They're very similar in effect. Any thoughts?
Member
Joined 2009
Ardor - Point Source Monitor
Horizontal Frequency response at 0=blue, 60=red, 90=green
An externally hosted image should be here but it was not working when we last tested it.
My speakers a very wide dispersion design. I was aiming to do omni but I was told I can't call them that because the cabinet obstructs the backwave. Currently the sweet spot is about 60degrees off-axis(the red). I'd like to maybe minimize to 45 if i can.
Conventional speakers in all methods of placement have this inherent trouble in the bass and midrange because of their decreasing directivity in the lower spectrum. I think most will agree that cardioid is the ideal directivity pattern for that region. It's the direction I'm looking in for my next speakers.
For the time being I think my current speakers are well suited for toe-out because I put emphasis on very wide uniform radiation. I selected the crossover points to use the drivers before they begin to be directional. I get a flat Frequency Response at the sweet spot and the side wall, the part that worries me is the reflection from the front wall.
You need to record your system's impulse response (say, with HolmImpulse) and feed that to his excellent application software... which runs in the octave programming enviroment as the only thing special compared to a straight Windows executable file (.exe). The package is IHMO excellent (the multiresolution constant-Q wavelet visualization is really killer, quite superior in information content to most other impulse visualizations), big thanks to Elias for sharing this, and his insights in general.
I'm using gapless shaped tone burst sequences for many years now (with 2..3 periods of raised cosine envelope corners, varying period count of bursts, also sometimes changing polarity after each burst). This is used in a sort of "wavelet analysis by ear", also having some the discussed modulation-domain properties (for the in-phase sequences the tone becomes one continuous drone, effectivly unmodulated, when exiting a mode and sitting on the peak of it. Elias' work is putting this in a pretty scientific context wrt visualization and analysis.
- Klaus
With my open baffle dipoles, I find that front wall reflections that are greater than about 6-7mS give more than they take (speakers being at least 3 ft. out from the front wall). You obviously still have some comb filter effects, although mine measure very small, but I get a more 3-D nature to any embeded reverbs in the music, which is the main reason I designed and built these speakers. 3-D means more emotional impact.
Member
Joined 2009
phase of front wall and sidewall reflections
If I understand correctly the front wall reflection of a monopole speaker comes back in opposite phase. The sidewall reflections are with the original phase. I'm not sure how that affects the image but I don't think the front wall reflection of a monopole and dipole are comparable. Maybe somebody more knowledgeable than me can shed light on this issue?
If I understand correctly the front wall reflection of a monopole speaker comes back in opposite phase. The sidewall reflections are with the original phase. I'm not sure how that affects the image but I don't think the front wall reflection of a monopole and dipole are comparable. Maybe somebody more knowledgeable than me can shed light on this issue?
Although I haven't proven this, I believe there is a phase reversal any time an acoustic signal is bounced off any wall. I'd love to hear from someone who knows 1st hand from an actual experiment. But either way, front wall and side wall reflections (any reflections) are going to produce comb filter effects when combined with any other time displaced signal at your ear, regardless of which wall or phase. The only instance when reflections are a good thing that I know of is when they give you a false spaciousness that you decide you like, or when it's because many reflections cause a filling in of each others cancellation notches better than just a few. If the reflection paths cause delays in the 6mS - 20mS range, they don't seem to be particularly subjectively damaging, and some people might actually like the effect they have. Personally I like the >6mS reflections on the Z axis, but prefer to get my lateral spaciousness from cues in the recording combined with the use of an inter-aural cancellation circuit.
One specific effect of comb filters is that with the pitch shifts as they are typical in most any kind of music (there is FM-modulation as well as AM-modulation at work in many instrument sounds, and in voices). When we have a significant reflection this causes a phasing-effect which tends to enrich the sound both in attack an vividness (electric guitar players know, a very short -6dB single delay thickens solo lines as well power chord drones when properly fine tuned).
In my book of personal preferences some reflections, real ones as synthesized ones, offering different degrees of freedom, can be a basically good thing, as well as the resulting gentle(!) comb filter effects. I strived long until I reached a point of listening in a pretty well damped room (exept for LF) only to find out that it sounds too lifeless, unengaging but with extremely accourate but flat imaging at speaker distace(no height and not too much depth). Put this in context that I was already using the dipoles point sources with their virtues and the stereo already had improved by a magnitude over the classic 3-ways used before.
Then I set out and dialed in artifical reflections (there is a sweet spot that combines nicely with the front wall delay) and a slightly modulated reverb for the vividness of an outdoor reverb ambience, chained into a few % of in-phase channel crosstalk as well as speaker-crosstalk cancelling signals to get some low-level comb filtering from both interchannel level and time differences (this does the main space and image size illusion, phantom sources become balloons in a 3D space compared to single points on a line). For a visual analogy, see Head Tracking for Desktop VR Displays using the WiiRemote - YouTube , of course not a literal analogy, but I think you get the message.... which is: you get a feeling that the phantom sound sources have size and are layered both in their positions as well as in their acoustic ambience (reverb of the recording space or reverb added to individual instrument tracks). You can clearly distingish the different reverbs and phantom source positioning methods used in music production. With listening to point sources in semi-anechoic space this is much less the case.
All that comes with only slight (IHMO) impact on precision, but with a need to dial in higher volumes as the image energy is mostly concetrated behind the speakers, but occasionally streaches out beyond the speakers in any direction (including in front of them, and way back behind the front wall).
That really put the fun back into listening, even if this is a no-go approach in a purist way of thinking how audio reproduction should be and to which I fully agree of course, but I don't bother anymore... every single time I open my eyes after listening I hardly can believe the sound really came from any speakers at all, let alone from speakers that close and in that moderate sized room... and want could I want more, this effect refuses to decrease even after many months of use whereas other problems of the speaker and system became more and more apparant over time.
EDIT:
I see, you use a different but basically equally "artifical" and personal sculpturing of the delivered illusion... I like that because I think this way of dealing with our hobby is very creative and satisfying, and educational too, compared with swapping cables "to get the soundstage right".
- Klaus
Last edited:
ScottG wrote;
Most "hit" the bottom of their baffle-step loss somewhere between 600-300 Hz. At that point they are radial (or horizontally "omni"). Their non-directional behavior at these lower freq.s does little for maintaining imaging character and tonal behavior. (..and it gets particularly bad with loudspeakers employing significant baffle step correction.)
DBmandrake wrote;
“Or just use a large baffle!
It may not be narrow and sleek looking, and goes against modern design trends, but a moderately wide baffle can put the baffle step frequency below this range, constraining the entire midrange to no greater than half space.” And “On the contrary, a wide baffle speaker can be moved quite close to the front wall (closer than a narrow baffle design) with very little detriment to tonal balance or imaging in the midrange”
Boris, you went to considerable thinking in the design of your point source speakers, few follow that spacing criteria needed for coherent addition but our horns do as well..
Consider that a flat baffle is a 180 degree horn at least from the standpoint of the pressure radiating away. While the “high pass” corner for horn loading may not be compatible with typical driver parameters when mounted this way, it is still a horn so far as governing directivity.
By high pass I mean that for a 30 hz bass horn to work as an acoustic transformer, it’s cross section cannot increase any faster than doubling every two feet while a 60Hz horn would be doubling every foot. In any case, for a point source (too small to have it’s own directivity) on a flat baffle, it will have a radiation pattern like any other simple straight sided horn (a portion of a sphere) and exhibit a pattern loss point where the pattern expands (doubles angle for an octave lower), the point related to the baffle step in direct radiator parlance.
In the approach I developed for commercial sound at work where pattern control is critical, one can combine as many frequency ranges as one wishes into one horn shape so that it behaves like a single wide band driver in a large conical horn. While a different principal of operation, it radiates a spherical segment like a quad ESL-63 except >+30dB louder and has a more or less constant radiation angle down to pattern loss F (set by mouth size and angle).
Boris, consider a point source system like your driver arrangement but on a large flat baffle. Now picture folding that baffle into a wide cone with the hf driver at the apex and the mid and low drivers mounted on the walls but retaining the same ¼ wl separation. AS long as each source is less than ¼ wl across, it fills the horn like one point source even if there were several drivers. Now build a crossover that adds these together so that what radiates on the big end appears to be from one source at the apex.
With a horn built like that, 28 inches across at the mouth and 50 degree angle like an SH-50, a significant degree of directivity is produced. Some numbers from the polar plot of an SH-50;
The pattern width or beam width defined as the -6dB point relative to the center.
The data wasn't taken at 20KHz but at 16KHz the pattern is about 60 degrees wide and about -25dB down at 90 degrees off axis.
At 10 KHZ the pattern is about 50 degrees wide and -35dB at 90 degrees off axis.
At 5KHz, the pattern is about 50 degrees wide and at 90 degrees off axis is about -30dB (1/1000) down.
At 1KHz the pattern is widening but still about 50 degrees wide and at 90 degrees off axis is about -20dB.
At 500Hz the pattern is about 70 degrees wide and 90 degrees off axis is about -20dB down.
At 250Hz the pattern is about 120 degrees wide and is about -8dB 90 degrees off axis.
I don’t have a way to measure how much difference this directivity makes in the home but it is a lot in my living room. I see some folks have posted an ETC at the listening position, I’ll see if I can do one of those over the weekend.
With some of the Synergy horns, the cabinet wall angle is the same as the horn wall and the distance between them fairly small. Given the directivity and lack of any lobes or nulls in it’s radiation, one can place it on the angled side hard against the wall and then effectively no side reflections are produced. This has the effect of allowing a much wider sound stage due to the larger spacing but lack of side wall reflections.
Best,
Tom Danley
Danley Sound Labs
Boris, page 4 here is the point source approach as applied to a full range CD horn;
http://www.danleysoundlabs.com/pdf/danley_tapped.pdf
Most "hit" the bottom of their baffle-step loss somewhere between 600-300 Hz. At that point they are radial (or horizontally "omni"). Their non-directional behavior at these lower freq.s does little for maintaining imaging character and tonal behavior. (..and it gets particularly bad with loudspeakers employing significant baffle step correction.)
DBmandrake wrote;
“Or just use a large baffle!
It may not be narrow and sleek looking, and goes against modern design trends, but a moderately wide baffle can put the baffle step frequency below this range, constraining the entire midrange to no greater than half space.” And “On the contrary, a wide baffle speaker can be moved quite close to the front wall (closer than a narrow baffle design) with very little detriment to tonal balance or imaging in the midrange”
Boris, you went to considerable thinking in the design of your point source speakers, few follow that spacing criteria needed for coherent addition but our horns do as well..
Consider that a flat baffle is a 180 degree horn at least from the standpoint of the pressure radiating away. While the “high pass” corner for horn loading may not be compatible with typical driver parameters when mounted this way, it is still a horn so far as governing directivity.
By high pass I mean that for a 30 hz bass horn to work as an acoustic transformer, it’s cross section cannot increase any faster than doubling every two feet while a 60Hz horn would be doubling every foot. In any case, for a point source (too small to have it’s own directivity) on a flat baffle, it will have a radiation pattern like any other simple straight sided horn (a portion of a sphere) and exhibit a pattern loss point where the pattern expands (doubles angle for an octave lower), the point related to the baffle step in direct radiator parlance.
In the approach I developed for commercial sound at work where pattern control is critical, one can combine as many frequency ranges as one wishes into one horn shape so that it behaves like a single wide band driver in a large conical horn. While a different principal of operation, it radiates a spherical segment like a quad ESL-63 except >+30dB louder and has a more or less constant radiation angle down to pattern loss F (set by mouth size and angle).
Boris, consider a point source system like your driver arrangement but on a large flat baffle. Now picture folding that baffle into a wide cone with the hf driver at the apex and the mid and low drivers mounted on the walls but retaining the same ¼ wl separation. AS long as each source is less than ¼ wl across, it fills the horn like one point source even if there were several drivers. Now build a crossover that adds these together so that what radiates on the big end appears to be from one source at the apex.
With a horn built like that, 28 inches across at the mouth and 50 degree angle like an SH-50, a significant degree of directivity is produced. Some numbers from the polar plot of an SH-50;
The pattern width or beam width defined as the -6dB point relative to the center.
The data wasn't taken at 20KHz but at 16KHz the pattern is about 60 degrees wide and about -25dB down at 90 degrees off axis.
At 10 KHZ the pattern is about 50 degrees wide and -35dB at 90 degrees off axis.
At 5KHz, the pattern is about 50 degrees wide and at 90 degrees off axis is about -30dB (1/1000) down.
At 1KHz the pattern is widening but still about 50 degrees wide and at 90 degrees off axis is about -20dB.
At 500Hz the pattern is about 70 degrees wide and 90 degrees off axis is about -20dB down.
At 250Hz the pattern is about 120 degrees wide and is about -8dB 90 degrees off axis.
I don’t have a way to measure how much difference this directivity makes in the home but it is a lot in my living room. I see some folks have posted an ETC at the listening position, I’ll see if I can do one of those over the weekend.
With some of the Synergy horns, the cabinet wall angle is the same as the horn wall and the distance between them fairly small. Given the directivity and lack of any lobes or nulls in it’s radiation, one can place it on the angled side hard against the wall and then effectively no side reflections are produced. This has the effect of allowing a much wider sound stage due to the larger spacing but lack of side wall reflections.
Best,
Tom Danley
Danley Sound Labs
Boris, page 4 here is the point source approach as applied to a full range CD horn;
http://www.danleysoundlabs.com/pdf/danley_tapped.pdf
It is very similar in principal although they have taken it to a higher degree. I corresponded with Ulrich Horbach last year about his paper, which I greatly admired. He mentioned my prior work at Snell as part of the inspiration. I had also discussed those design with Don Keele several times, including the fractal nature of the arrays: If you get a "recipe" for a 3 element array that works, then that cluster becomes the center for the next larger array, and so on. Most of this work comes from the lobe free THX arrays I worked on at McIntosh (HT-1 and HT-3). None of it is an earthshaking invention, more an application of common sense, once you see the relationship between lobe patterns and element spacing relative to the crossover frequency.
All these designs become very appealing, whether CD horns, symmetrical expanding arrays, wide range dipoles, well tapered line arrays, or cardiod bass cabinets. The question remains whether we are becoming enamored with a pretty polar curve or whether a particular polar patern actually has merit from a psychoacoustic point of view.
That is why the fundamental studies of rooms, room reflections, timing and direction are so important. Its not what we can do but what we should do that points the way to progress.
David S.
Member
Joined 2009
Hi Tom,
I understand that my design is inherently flawed in the lower midrange at the point where the wavelength grows to no longer be supported by the front baffle. I can see it on ungated measurements and I can hear it. I'm planning to try a U-shape baffle for cardioid response in that region.
-----------
I'm familiar with the operation principles of the Unity/Synergy horns. I can't wait to hear them. It's a groundbreaking piece of engineering and I admire your work!
Given the narrow directivity of the Synergy horns and their ability to stack together you are probably in the best position to answer the question of this thread. Have you tried pairing 2 horns per channel to increase the horizontal dispersion angle? You can get insight on how important sidewall reflections are by pairing 2 Synergy horns per channel and having one radiate directly and the other one illuminate the wall. If you can switch on/off the outer horns you should be able to hear exactly how important room interactions are.
I understand that my design is inherently flawed in the lower midrange at the point where the wavelength grows to no longer be supported by the front baffle. I can see it on ungated measurements and I can hear it. I'm planning to try a U-shape baffle for cardioid response in that region.
-----------
I'm familiar with the operation principles of the Unity/Synergy horns. I can't wait to hear them. It's a groundbreaking piece of engineering and I admire your work!
Given the narrow directivity of the Synergy horns and their ability to stack together you are probably in the best position to answer the question of this thread. Have you tried pairing 2 horns per channel to increase the horizontal dispersion angle? You can get insight on how important sidewall reflections are by pairing 2 Synergy horns per channel and having one radiate directly and the other one illuminate the wall. If you can switch on/off the outer horns you should be able to hear exactly how important room interactions are.
Klaus, it's so nice to hear from someone who appreciates the improvement of good dipoles. So many people seem to believe that technical accuracy as is commonly understood gives them the most enjoyable or "correct" listening experience. As a musician myself, I know that I would always welcome having a better sound out of my guitars. Most musicians I've met have very limited knowledge about getting good sound. I have two good friends who despise me for "cheating", using my own brew of the Carver inter-aural cancellation circuit to get a vastly improved spread of soundstage and air between instruments and singer. They refuse to sit in the sweetspot... It's against their religion or something... I can only laugh. Tell me more about your dipoles.
All that talk about what frequency ranges do the most imaging, and which don't (about ten pages back now) has got me willing to try an experiment that relates to what you said. I've been wanting to build a better center speaker for under my TV for a while now. I'm going to have a front that is broken up into 3 similar sized panels, with a five inch on each panel, but also a 2 inch dome at the top of the center panel (1kHZ - 6-7kHZ). On top of the cabinet will be a horizontal array of 4 3/4 inch dome tweeters, with two more tweets right above and in the middle (sorta Bozak style - each angled differently). The gamble is that I'm going to cross over at about 6-7 kHZ, which means lobing due to the distance from the 2 inch dome (which will be minimal but still more than would be nice for this xover freq.). But, since there are six tweeters, each with a different interference pattern than the others (depending on the listening angle), at the crossover frequency, I'm hoping they will fill in each others cancellation frequencies, and sound fine. But the reason I picked 6-7kHZ is because that's just above where we do most of our ear-brain image location analysis. It's a place in frequency where we can screw it up a bit, without seriously damaging imaging info that we can use. I'm using Dayton ND20FB-4 tweets which have almost no flange at all, so they can be pretty close to each other. The center speaker I built before (8 inch in a box with a ribbon on top X=about 4kHZ) sounds too contained. I want a more spread out sound. I think this will give me what I want.
That's probably the best working and still practical approach, 2 horns covering the listening area with toe-in (to make the sweet spot bigger and somewhat stabilize extreme listening positions to the left and the right) and 2 very narrow directivity, directable horns firing at the first reflection points of ipsilateral walls.
Spaciousness very noticeable starts to increase when the level of the reflections created by the side speakers is louder than -5dB compared to the direct sound.
http://www.diyaudio.com/forums/mult...y-pattern-stereo-speakers-74.html#post2702519
I think the most basic reason why stereo fails is it's trying to reproduce a mono sound (usually multiple mono sounds) with stereo.
Every sound in nature is mono. Never stereo.
(Yes, you hear with 2 ears)
Trying to achieve a "soundstage" can be fun. I enjoy it.
But I put the emphasis on the "sound" and a lot less on the "stage".
- Status
- Not open for further replies.