Lynn Olson said:
Hi Lynn,
Stimulating thread, it’s become quite the resource. Hope I’m not pulling you too far off topic here.
Diffraction is an obvious first order effect, showing itself clearly in the frequency response. For a well designed box, internal reflections tend to be so far down that they’re hard to see in a frequency response measure or in a impulse response, but not hard to hear. They appear to be minimum phase at the gross level, as they don’t show up in the excess phase. However, they could still be non min phase but be so far down in level that visually in a graph their contribution to the sum response is visually imperceptible. As we all know, given there is some inherent delay involved, they may still be audible.
A test I use to illustrate how invasive box standing wave modes are, is to take an unstuffed golden ratio test box with driver, then kill the mls signal. The decay can be heard to audibly ring through the cone for almost a second. Some prefer unstuffed boxes, but to me it feels too much like running the preamp through a chorus pedal.
I use this golden box test, along with listening tests, to determine the best box stuffing, with my biases. I’ve tried audio polys, lamb’s wool, 3 grades of Fiberglass, fiberboard, Sonex lining, and blends thereof. For me, straight old fiberglass of 1 to 1.5 cuft/in is the ticket. I’ve yet to compare to OB though, I think this would be a great and simple test, to highlight the expected improvement with OB.
For diffraction, I use the 1.25” bevels when available and proper driver placement to spread the delayed diffraction. I’ve also tried full out audio felt on the front, carefully noting and compensating for how it changes the acoustic impedance, operating like a shallow waveguide. I wasn’t enamored with it. In my limited tests, I heard no immediate improvement. If anything, it tended to dry the sound up in an unpleasant way. I graphed out the change over angle provided by the felt and I’m theorizing that the dryness was due to the reduced dispersion in the high end.
For a large OB though, a colour matched and countersunk 2” strip of felt nicely applied along the length of the front side edges may work out quite well at reducing diffraction. McMaster Car sells the felt, perhaps in different colours.
But back to min phase. The reason it’s important to recognize diffraction as min phase is that it provides the knowledge and comfort that it can be corrected for with minimum phase equalization. Non-min phase eq can’t be corrected with min phase eq. So, we can use passive or active xovers and eq the diffraction. Good thing, as this is what we do when we design for a target response. Unfortunately, a secondary effect is that it mis-eq’es off axis as the time delay between the diffraction causing edges and the direct source change off axis. So, eqing flat on axis results in further error off axis. For this reason, I don’t EQ out all the diffraction. I leave just a bit in on axis, and create a smoother response launched into the room. Coupled with somewhat uncorrelated xovers, it works well for me, given my live room.
Sean Olive (and I think it was Daniel Queen) ran independent studies where they artificially introduced coincident and non-coincident delayed reflections. They tested audibility as difference limens, using different source signals. Depending upon the exact condition, different signals worked best at uncovering the audibility. I wrote a summary of 5 or 6 such papers once for a different news group. I’ll repost here tonight. It’ll provide some guidance as to which test signal works best.
john k... said:
Hi John, this is what's meant by K<1, in Feyz's analysis. The only thing I could think of is a very large resonance encountered by the reflected or diffraction signal, but not the direct. I can't even think of how that my possible, but I don't lie in bed worrying about it.
Cheers,
Dave
Hi
Dave, does cepstral analysis also work on the shorter delayed edge diffraction artefacts, thus providing an efficient tool to minimise them?
-----------------
Graham, I'll keep that in mind. Thanks.
In general I think that the wide range speaker for Lynn's OB should have NO dust cup as the rear sound from any ventilation path to the back side contributes to the overall performance of the speaker.
This will be hard to find in the PRO / neodym area.
In the Dynaudio 21W54 example the area of the ventilation hole is roughly ¼ of the VC area accelerating the air escaping from the back to 4 times the membrane speed – kind of backwards aimed compression driver in other words.
Greetings
Michael
Dave, does cepstral analysis also work on the shorter delayed edge diffraction artefacts, thus providing an efficient tool to minimise them?
-----------------
Graham, I'll keep that in mind. Thanks.
In general I think that the wide range speaker for Lynn's OB should have NO dust cup as the rear sound from any ventilation path to the back side contributes to the overall performance of the speaker.
This will be hard to find in the PRO / neodym area.
In the Dynaudio 21W54 example the area of the ventilation hole is roughly ¼ of the VC area accelerating the air escaping from the back to 4 times the membrane speed – kind of backwards aimed compression driver in other words.
Greetings
Michael
MTM Dipoles
Hi John, thanks for the posting and the comprehensive design information on the Music and Design website. If I read your comments accurately, there's a small spectral region (about an octave or so wide) where the dipole radiation pattern widens out (increasing total power into a sphere), and you arrange the MTM spacing so the narrowing vertical directivity offsets the widening (in the horizontal plane).
That must account for what at first glance looks like fairly large vertical spacing in the NaO II system - this must be calculated to provide just the right degree of vertical directivity. When combined with an acoustic 3rd-order crossover, this appears to provide considerably improved power spectra compared to a MTM in a conventional enclosure.
As you say, a dipole and monopole MTM are quite different - but the dipole MTM needs to have the baffle width, vertical spacing, and crossover frequency carefully selected so the MTM narrowing and dipole pattern widening are complementary. No simple equation for this comes to mind - I surmise that simple vector calculations for 30, 45, and 60 degrees off-axis (V & H) would not be sufficient.
It also makes the mesh-edge approach look even more interesting, since that may provide a softer transition between the different operating regimes, especially for the front-and-back dipole tweeters.
P.S. Dave, looking forward to the Sean Olive study you're going to be publishing this evening. Lots of good information in this thread - it's definitely attracted some talented contributors from all over the industry. Thanks again to Dave, John, Earl, MBK, and all the others who've joined in.
john k... said:
Hi John, thanks for the posting and the comprehensive design information on the Music and Design website. If I read your comments accurately, there's a small spectral region (about an octave or so wide) where the dipole radiation pattern widens out (increasing total power into a sphere), and you arrange the MTM spacing so the narrowing vertical directivity offsets the widening (in the horizontal plane).
That must account for what at first glance looks like fairly large vertical spacing in the NaO II system - this must be calculated to provide just the right degree of vertical directivity. When combined with an acoustic 3rd-order crossover, this appears to provide considerably improved power spectra compared to a MTM in a conventional enclosure.
As you say, a dipole and monopole MTM are quite different - but the dipole MTM needs to have the baffle width, vertical spacing, and crossover frequency carefully selected so the MTM narrowing and dipole pattern widening are complementary. No simple equation for this comes to mind - I surmise that simple vector calculations for 30, 45, and 60 degrees off-axis (V & H) would not be sufficient.
It also makes the mesh-edge approach look even more interesting, since that may provide a softer transition between the different operating regimes, especially for the front-and-back dipole tweeters.
P.S. Dave, looking forward to the Sean Olive study you're going to be publishing this evening. Lots of good information in this thread - it's definitely attracted some talented contributors from all over the industry. Thanks again to Dave, John, Earl, MBK, and all the others who've joined in.
Re: MTM Dipoles
Well, as much as I love the theory part I mostly use it as a guide to tell me what is happening and gain understanting of how these things interact. With the NaO II and the Mini the baffle was designed with these things in mind but ultimately long term listening was the gage of success.
The development of the NaO II began with the NaO prototype. There have been a number of updates to the crossover and woofer system along the way. But the baffle configuration and driver layout has remained constant. While it is very simple, it has proven itself.
Now it's your turn Lynn. 1283 replies to yor thread since March. Time to start cutting baffles.
Lynn Olson said:
Well, as much as I love the theory part I mostly use it as a guide to tell me what is happening and gain understanting of how these things interact. With the NaO II and the Mini the baffle was designed with these things in mind but ultimately long term listening was the gage of success.
The development of the NaO II began with the NaO prototype. There have been a number of updates to the crossover and woofer system along the way. But the baffle configuration and driver layout has remained constant. While it is very simple, it has proven itself.
Now it's your turn Lynn. 1283 replies to yor thread since March. Time to start cutting baffles.
Some Time Soon ...
Progress on recovery is slow but sure. Still using the walker, but now need it mostly for balance. I'm guessing the left leg is good for 60~100 lbs, and now I can stand by myself with the walker nearby, can lean over, and pick up things. Sounds trivial, but all of them are things that I last did on January 7th.
The latest distraction is the failure of my main computer, with the motherboard, dual processor unit, and power supply all failing at once. This happened three weeks ago, thus no access to my graphics programs, and the repair process has resulted in three in-home visits and 3rd complete replacement of the aforementioned PS, m/board, and CPU's. Finally got Apple to replace it with a new machine - I'll be talking to them on the phone tomorrow, to set it up.
Think I'll get a 1 KW UPS tomorrow to protect the new computer. Colorado has plenty of lightning storms - there's been several times I've heard a faint "click" come out of the computer speakers and seen a lighting bolt at the same time. I've even heard a "click" in my head as I drove along and saw lightning a few miles away - must be some kind of ground-surge that can be picked up directly. Rare, but it happens. Much more lightning here than the West Coast - there's a storm now as I type this.
The tech that came out here thinks it's likely the PS was damaged enough to lose regulation on some critical low-voltage rails, which then in turn took out the m/board and dual CPU's.
Boy, I can tell all of you reading this, DON'T EVER BREAK A LEG! It takes a long time to get better, and being disabled is a real drag - and scary too. I thank Bud Purvine, who had a similar experience at age 18, for encouraging me to post on the diyAudio forums. Being able to collaborate with the community here has been mentally stimulating - something Bud knew I needed - and given me something to work on in the months to come. At the rate things are going, I expect I will be walking unsupported in the next several weeks, and re-learning to go up and down stairs.
john k... said:
Now it's your turn Lynn. 1283 replies to yor thread since March. Time to start cutting baffles.
Progress on recovery is slow but sure. Still using the walker, but now need it mostly for balance. I'm guessing the left leg is good for 60~100 lbs, and now I can stand by myself with the walker nearby, can lean over, and pick up things. Sounds trivial, but all of them are things that I last did on January 7th.
The latest distraction is the failure of my main computer, with the motherboard, dual processor unit, and power supply all failing at once. This happened three weeks ago, thus no access to my graphics programs, and the repair process has resulted in three in-home visits and 3rd complete replacement of the aforementioned PS, m/board, and CPU's. Finally got Apple to replace it with a new machine - I'll be talking to them on the phone tomorrow, to set it up.
Think I'll get a 1 KW UPS tomorrow to protect the new computer. Colorado has plenty of lightning storms - there's been several times I've heard a faint "click" come out of the computer speakers and seen a lighting bolt at the same time. I've even heard a "click" in my head as I drove along and saw lightning a few miles away - must be some kind of ground-surge that can be picked up directly. Rare, but it happens. Much more lightning here than the West Coast - there's a storm now as I type this.
The tech that came out here thinks it's likely the PS was damaged enough to lose regulation on some critical low-voltage rails, which then in turn took out the m/board and dual CPU's.
Boy, I can tell all of you reading this, DON'T EVER BREAK A LEG! It takes a long time to get better, and being disabled is a real drag - and scary too. I thank Bud Purvine, who had a similar experience at age 18, for encouraging me to post on the diyAudio forums. Being able to collaborate with the community here has been mentally stimulating - something Bud knew I needed - and given me something to work on in the months to come. At the rate things are going, I expect I will be walking unsupported in the next several weeks, and re-learning to go up and down stairs.
Hi,
As promised some notes from the literature, looking at the audibility of resonance and reflection. Apologies that this isn't re-interpreted directly for the application in hand, being much older posts of mine. I still think there’s a wealth of info to mine here, to help us develop and voice our speakers in a more controlled manner.
This was an area of interest for me at one time, when I helped design acoustic echo cancellers. If anyone would like to know something more specific from the papers, post away and I could always double back and take a look.
Lynn, a friend here in town fell down the stairs in January and just got off her crutches, still with a ways to go. I hope you have a speedy recovery, but not so speedy that we don’t hear how this saga ends.
OK, on to the old posts.
To: bass@lunch.engr.sgi.com (BNR400)
Wrom: ZFSQHYUCDDJBLVLMHAALPTCXLYRWTQTIPWIGYOKSTTZRCLBDXRQBGJSNBOHM
Subject: re:Bob Neidorff/Re: Ken's desert
"Modification of Timbre by resonance: Measurement and Perception" by Floyd Toole and Sean Olive, JAES, '88.
This landmark paper describes the audibility of resonances that are superimposed on the signal. Box panel resonances, cone modes etc can fall into this category. They used several types of music, pink noise and transients (drums do nicely here) as sources. Music was always the least revealing of resonances, especially music that contained any natural reverberation in the recording. For resonances delayed less than 1 ms (i.e. most speaker box resonances and enclosure standing waves), pink noise was the most revealing source. For resonances delayed beyond about 30 ms (i.e. room acoustical phenom), transient signals worked well. In fact, the presence of room reverberation led to easier audibility of delayed resonances when transient sources were used.
Read this paper if you want to know exactly why pink noise and transients are such useful signal sources in loudspeaker qualification. The utility of this paper is enormous and extends to many other aspects of loudspeaker design. For example, I've interpreted the results to tell me how far down I need the band stop of a high pass filter when used with a kevlar woofer showing a high Q break up mode.
Dave Dal Farra (gpz750@bnr.ca) "I was moving so fast I started
Bell Northern Research using Him as a braking marker"
Audio Design Group FJ1200/GPz750
From Dave Dal Farra:
Here are some of the better references, and I've gone one better and listed some of the more interesting conclusions they reached. I once did jury duty, and read and summarized all these papers on the bus to and from a long trial. Enjoy!
"The Modification of Timbre by Resonances: Perception and Measurement", Olive, Toole, JAES Vol 36, no 3, 1988
- shows how the ability to hear peaks is related to Q, and what signals best allow us to hear particular types of resonance;
- example: resonances delayed less than 1 ms are easier to hear using pink noise;
- in semi-reverberant fields, pink noise or transients work equally well in detecting resonances.
"The detection of Reflections in Typical Rooms", Olive, presented at the 85th AES convention, preprint # 2719 (F-1)
- reflections greater than10ms best heard using impulses
- reflections less than 10ms easier to hear using pink noise.
- reflections in the same direction as the first incidence can be 5 to 10 dB above the first incidence before detection;
- lateral reflections increase spaciousness when just above the noticeability threshold;
- vertical reflections in the median plane affect timbre more than spaciousness;
- lateral reflections less than 10 ms lead to image spreading;
- lateral reflections from 10 to 40 ms lead to image spreading and spaciousness;
- echo is detectable above 40 ms.
- room RT60 has almost no effect on these effects for reflections delayed no more than 30 ms.
- reflections have to be 7 dB above the absolute detection threshold before they cause an image shift
"Electroacoustic simulation of listening room acoustics; Psychoacoustic design criteria", Soren Bech, presented at the 89th AES convention, preprint 2989
- to maximize the subjective perception of diffusion, angle the speakers in 23 degrees, and minimize inter-aural cross correlation
- narrower bandwidth signals will sound more diffuse than wideband ones, if the environment is mainly non-diffuse.
"A psychoacoustically optimized loudspeaker", Kantor, presented at the 77th AES convention, preprint # 2190
- early reflections up to 600uS cause localization distortion due to image spreading
- the ear's peak sensitivity to reflections is 2 kHz
- the floor notch causes vertical localization errors only if the signal envelope is longer than several ms.
- 20 to 40 ms reflections increase ambience
- recommends constant directivity in the mids where the ear is most sensitive to reflections
- subjectively, 20 ms reflections create half the spectral colouration of 250 us reflections
"A perceptual Criteria for Loudspeaker Evaluation", Kates, JAES Vol 32, No 12, 1984
- binaural effects lead to echo suppression (its why the floor notch is so audible)
"The effect of Loudspeaker Radiation Patterns on Stereo Imaging and Clarity", Queen, JAES 1978
- for 0 to 3 ms, images are smeared following an intensity rule, phase of the reflection didn't appear to affect this.
- short term reflections at least 10 dB below the first incidence do not create image shift (Olive found the threshold was 7 dB)
- horizontal mounting of woofer and tweeter caused 15 degree image wander when side wall reflections were present. Main cause was strong side lobes due to woofer/tweeter interaction (lesson here: use steep xovers if mounting woofer and tweeter horizontally).
- very narrow baffles reduced image shift in real rooms
- unconscious head move aids low frequency localization ability
- multiple vertical sources reduce image ambiguity (there's one for the line-source lovers)
"Localization of Sound in a Room with reflecting Walls", Wagenaars, JAES Vol 38, No.3, 1990
- room reflections first impact the perception of depth, before lateral direction
"Spaciousness and Localization in Listening rooms and their effects on the recording technique", Griesenger, JAEs Vol 34, No4, 1986
- we localize below 700 Hz by detecting phase differences between the ears, induced by the diffraction of plane waves around the head.
- Above 1.5 kHz, time differences, amplitude differences, and pinnae effects dominate localization.
- again, spaciousness is a lateral phenomena: its increased by decreasing inter-aural correlation.
- to add spaciousness, augment left minus right lateral reflected energy.
- he was most pleased with the added spaciousness shown by a 4 dB bass boost in left minus right below 600Hz, in concert with a 4 dB L+R bass cut below 600Hz. Again, dipole fans take note.
- BART: Dipole speakers sounded far more spacious than bipoles.
- smaller rooms decrease the ability to localize low frequencies, pulling them to the center of the stereo pair.
- room modes at low frequencies alter the phase relationship between pressure and velocity (relative to free field), interfering with the ability to localize.
"Temporal Localization Cues and their role in auditory perception", Wilde, presented at the 95th AES convention, preprint 3708
- this one is a favourite as its particularly contentious: the human ear can resolve 3 us changes in localization; A 44 kHz sample rate can resulve infinitely well but if the assumption holds that >20kHz energy results in envelope changes below 20 kHz, then there may be merit to this limitation.
- shows how hearing is a learned phenomena, as subjects can adapt to others' head related transfer functions, through training.
"Optimum Loudspeaker Directional Patterns", Kates, JAES, 1980
- for you dipole fans, Kates postulates that if you're 1.5 to 2x farther from the speakers than they are from each other, a cosine pattern provides best localization.
"Effect of early reflections from upside on auditory envelopment", Journal of the Acoustical Society of Japan, 16, 2 1995
- again, lateral spaciousness is mainly produced by early lateral reflections with high inter-aural independence.
- while Olive showed short vertical reflections affect timbre, this paper shows vertical images are spread if a vertical reflection is delayed by at least 10 ms, and the reflected power is at least 1/2 the incident.
Cheers,
Dave Dal Farra
As promised some notes from the literature, looking at the audibility of resonance and reflection. Apologies that this isn't re-interpreted directly for the application in hand, being much older posts of mine. I still think there’s a wealth of info to mine here, to help us develop and voice our speakers in a more controlled manner.
This was an area of interest for me at one time, when I helped design acoustic echo cancellers. If anyone would like to know something more specific from the papers, post away and I could always double back and take a look.
Lynn, a friend here in town fell down the stairs in January and just got off her crutches, still with a ways to go. I hope you have a speedy recovery, but not so speedy that we don’t hear how this saga ends.
OK, on to the old posts.
To: bass@lunch.engr.sgi.com (BNR400)
Wrom: ZFSQHYUCDDJBLVLMHAALPTCXLYRWTQTIPWIGYOKSTTZRCLBDXRQBGJSNBOHM
Subject: re:Bob Neidorff/Re: Ken's desert
"Modification of Timbre by resonance: Measurement and Perception" by Floyd Toole and Sean Olive, JAES, '88.
This landmark paper describes the audibility of resonances that are superimposed on the signal. Box panel resonances, cone modes etc can fall into this category. They used several types of music, pink noise and transients (drums do nicely here) as sources. Music was always the least revealing of resonances, especially music that contained any natural reverberation in the recording. For resonances delayed less than 1 ms (i.e. most speaker box resonances and enclosure standing waves), pink noise was the most revealing source. For resonances delayed beyond about 30 ms (i.e. room acoustical phenom), transient signals worked well. In fact, the presence of room reverberation led to easier audibility of delayed resonances when transient sources were used.
Read this paper if you want to know exactly why pink noise and transients are such useful signal sources in loudspeaker qualification. The utility of this paper is enormous and extends to many other aspects of loudspeaker design. For example, I've interpreted the results to tell me how far down I need the band stop of a high pass filter when used with a kevlar woofer showing a high Q break up mode.
Dave Dal Farra (gpz750@bnr.ca) "I was moving so fast I started
Bell Northern Research using Him as a braking marker"
Audio Design Group FJ1200/GPz750
From Dave Dal Farra:
Here are some of the better references, and I've gone one better and listed some of the more interesting conclusions they reached. I once did jury duty, and read and summarized all these papers on the bus to and from a long trial. Enjoy!
"The Modification of Timbre by Resonances: Perception and Measurement", Olive, Toole, JAES Vol 36, no 3, 1988
- shows how the ability to hear peaks is related to Q, and what signals best allow us to hear particular types of resonance;
- example: resonances delayed less than 1 ms are easier to hear using pink noise;
- in semi-reverberant fields, pink noise or transients work equally well in detecting resonances.
"The detection of Reflections in Typical Rooms", Olive, presented at the 85th AES convention, preprint # 2719 (F-1)
- reflections greater than10ms best heard using impulses
- reflections less than 10ms easier to hear using pink noise.
- reflections in the same direction as the first incidence can be 5 to 10 dB above the first incidence before detection;
- lateral reflections increase spaciousness when just above the noticeability threshold;
- vertical reflections in the median plane affect timbre more than spaciousness;
- lateral reflections less than 10 ms lead to image spreading;
- lateral reflections from 10 to 40 ms lead to image spreading and spaciousness;
- echo is detectable above 40 ms.
- room RT60 has almost no effect on these effects for reflections delayed no more than 30 ms.
- reflections have to be 7 dB above the absolute detection threshold before they cause an image shift
"Electroacoustic simulation of listening room acoustics; Psychoacoustic design criteria", Soren Bech, presented at the 89th AES convention, preprint 2989
- to maximize the subjective perception of diffusion, angle the speakers in 23 degrees, and minimize inter-aural cross correlation
- narrower bandwidth signals will sound more diffuse than wideband ones, if the environment is mainly non-diffuse.
"A psychoacoustically optimized loudspeaker", Kantor, presented at the 77th AES convention, preprint # 2190
- early reflections up to 600uS cause localization distortion due to image spreading
- the ear's peak sensitivity to reflections is 2 kHz
- the floor notch causes vertical localization errors only if the signal envelope is longer than several ms.
- 20 to 40 ms reflections increase ambience
- recommends constant directivity in the mids where the ear is most sensitive to reflections
- subjectively, 20 ms reflections create half the spectral colouration of 250 us reflections
"A perceptual Criteria for Loudspeaker Evaluation", Kates, JAES Vol 32, No 12, 1984
- binaural effects lead to echo suppression (its why the floor notch is so audible)
"The effect of Loudspeaker Radiation Patterns on Stereo Imaging and Clarity", Queen, JAES 1978
- for 0 to 3 ms, images are smeared following an intensity rule, phase of the reflection didn't appear to affect this.
- short term reflections at least 10 dB below the first incidence do not create image shift (Olive found the threshold was 7 dB)
- horizontal mounting of woofer and tweeter caused 15 degree image wander when side wall reflections were present. Main cause was strong side lobes due to woofer/tweeter interaction (lesson here: use steep xovers if mounting woofer and tweeter horizontally).
- very narrow baffles reduced image shift in real rooms
- unconscious head move aids low frequency localization ability
- multiple vertical sources reduce image ambiguity (there's one for the line-source lovers)
"Localization of Sound in a Room with reflecting Walls", Wagenaars, JAES Vol 38, No.3, 1990
- room reflections first impact the perception of depth, before lateral direction
"Spaciousness and Localization in Listening rooms and their effects on the recording technique", Griesenger, JAEs Vol 34, No4, 1986
- we localize below 700 Hz by detecting phase differences between the ears, induced by the diffraction of plane waves around the head.
- Above 1.5 kHz, time differences, amplitude differences, and pinnae effects dominate localization.
- again, spaciousness is a lateral phenomena: its increased by decreasing inter-aural correlation.
- to add spaciousness, augment left minus right lateral reflected energy.
- he was most pleased with the added spaciousness shown by a 4 dB bass boost in left minus right below 600Hz, in concert with a 4 dB L+R bass cut below 600Hz. Again, dipole fans take note.
- BART: Dipole speakers sounded far more spacious than bipoles.
- smaller rooms decrease the ability to localize low frequencies, pulling them to the center of the stereo pair.
- room modes at low frequencies alter the phase relationship between pressure and velocity (relative to free field), interfering with the ability to localize.
"Temporal Localization Cues and their role in auditory perception", Wilde, presented at the 95th AES convention, preprint 3708
- this one is a favourite as its particularly contentious: the human ear can resolve 3 us changes in localization; A 44 kHz sample rate can resulve infinitely well but if the assumption holds that >20kHz energy results in envelope changes below 20 kHz, then there may be merit to this limitation.
- shows how hearing is a learned phenomena, as subjects can adapt to others' head related transfer functions, through training.
"Optimum Loudspeaker Directional Patterns", Kates, JAES, 1980
- for you dipole fans, Kates postulates that if you're 1.5 to 2x farther from the speakers than they are from each other, a cosine pattern provides best localization.
"Effect of early reflections from upside on auditory envelopment", Journal of the Acoustical Society of Japan, 16, 2 1995
- again, lateral spaciousness is mainly produced by early lateral reflections with high inter-aural independence.
- while Olive showed short vertical reflections affect timbre, this paper shows vertical images are spread if a vertical reflection is delayed by at least 10 ms, and the reflected power is at least 1/2 the incident.
Cheers,
Dave Dal Farra
mige0 said:
Hi Michael,
I’ve spent some time investigating how it works, and how to post process the files, but I’ve yet to spend any time performing cepstral editing on direct measures. For this I defer to john k. John did send a cepstrum response once where I interpreted that reflections as far out as 1ms could be edited, but I highly doubt this would apply to diffraction. Diffraction is modeled as a number (approaching infinity) of point source reflections, each delayed based upon the distance to the baffle edge, for each point. This will create a multitude of reflections, spread over time, vs a discrete singular reflection such as floor bounce.
John will have a more definitive answer, but my judgement at this point is that, no, it wouldn’t work well for diffraction, unless the baffle were round, the speaker centrally mounted and the mic directly on axis.
Cheers from Ottawa,
Dave
Re: Re: Yust gotta know.
Which is pretty much what I was suggesting before: a "disappearing edge". (..still, its nice to know someone far more qualified suggests the same! )
IF the front and rear output are symmetric, BUT the edge isn't quite "0" - then its likely you will have a *very* narrow band diffraction effect (..thats likely to be high in spl).
john k... said:
Which is pretty much what I was suggesting before: a "disappearing edge". (..still, its nice to know someone far more qualified suggests the same!
)IF the front and rear output are symmetric, BUT the edge isn't quite "0" - then its likely you will have a *very* narrow band diffraction effect (..thats likely to be high in spl).
I tried a few sharpened edge baffles a while back. The edges were lit up from diffraction more than the same baffle with squared edges. I suspect the cause to be 2 possible causes:
1. With a pointed edge, instead of the transition being from 1/2 space on the baffle face to 3/4 space formed by the front and side, the transition is from 1/2 space to full space. Plus, while the diffraction effects for the front and rear may be equal and directly out of phase, they are too directional to cancel each other.
and/or
2. The counterpart wave from the other side increases the pressure change at the edge along with the corresponding diffraction effects, because for each high pressure region there is a matching low pressure region meeting it from the other side.
1. With a pointed edge, instead of the transition being from 1/2 space on the baffle face to 3/4 space formed by the front and side, the transition is from 1/2 space to full space. Plus, while the diffraction effects for the front and rear may be equal and directly out of phase, they are too directional to cancel each other.
and/or
2. The counterpart wave from the other side increases the pressure change at the edge along with the corresponding diffraction effects, because for each high pressure region there is a matching low pressure region meeting it from the other side.
Re: Re: Re: Yust gotta know.
Scott
From a scientific standpoint, I don't believe that what you claim is the case. The diffraction at a OB edge would double because of the pressure difference at the edge, not vanish. There are two reversals here, not one and the net effect is a doubling of the diffraction field not the elimination of it.
The edge becomes a "pressure release" boundary condition where the excess acoustic pressure at the edge sees the exact same, but opposite acoustic field from the other side of the diaphragm. This exactly nulls the acoustic pressure at this edge to zero, which causes a great deal of diffraction. A normal baffle edge for a closed box, only sees a small pressure change at the edge because of the loss of the boundary, but does not go all the way to zero. The net diffraction amount is always proportional to the pressure change.
ScottG said:
Which is pretty much what I was suggesting before: a "disappearing edge". (..still, its nice to know someone far more qualified suggests the same!
IF the front and rear output are symmetric, BUT the edge isn't quite "0" - then its likely you will have a *very* narrow band diffraction effect (..thats likely to be high in spl).
Scott
From a scientific standpoint, I don't believe that what you claim is the case. The diffraction at a OB edge would double because of the pressure difference at the edge, not vanish. There are two reversals here, not one and the net effect is a doubling of the diffraction field not the elimination of it.
The edge becomes a "pressure release" boundary condition where the excess acoustic pressure at the edge sees the exact same, but opposite acoustic field from the other side of the diaphragm. This exactly nulls the acoustic pressure at this edge to zero, which causes a great deal of diffraction. A normal baffle edge for a closed box, only sees a small pressure change at the edge because of the loss of the boundary, but does not go all the way to zero. The net diffraction amount is always proportional to the pressure change.
DDF said:
Thanks Ottawa Dave for all that cool stuff! A lot to chew on there.
I have been reading Dr. Dave's (the big kahuna at Lexicon) papers lately because the subject of spaciousness is dear to me. Why do some (few) systems do it well and other do not? Size helps, but isn't the final determinant.
Much studying to do.....
Re: Re: Re: Re: Yust gotta know.
Earl, I think you need to reconsider. Given that linear theory applies, we can consider the front and rear radiation separately, and the radiation from the main sources and from the diffraction sources separately as well. Start by considering perfectly anti-symmetric source. That is, the front and rear sources are identical in amplitude but inverted in phase. Along the front baffle surface the wave propagates outward towards the edge. When it reaches the baffle edge a diffracted wave is generated and propagates out from the point of diffraction summing with the original wave at the observation point. This diffracted wave will nominally be inverted relative to the incident wave. We can consider the sum of the original wave and the diffracted wave by the principal of superposition since both of these waves obey the linear wave equation.
Now we can look at the rear wave. When the front and rear sources are anti-symmetric the same result occurs as for the front, but the both the original rear wave from the source and the rear diffraction wave are 180 decrees out of phase with respect to their front counter parts, relative the points of diffraction. Thus, these two additional waves can be summed to the two waves from the front side. All than needs to be considered is the relative path length differences. The point here is that the problem can always be decomposed into these 4 waves, regardless of baffle thickness, and the result expressed as the superposition of the 4 at any point in space. (More than 4 if you want to consider secondary diffraction, but that only complicates the issue without introducing new information.)
As long as the front and rear primary sources are anti-symmetric and the baffle is symmetric, the front and rear the diffractions sources will also be anti-symmetric and will form a secondary dipole with separation equal to the baffle thickness, T, and axis aligned with the primary dipole axis. For the primary dipole the separation (for a circular baffle) is R+T. The relative strength of the secondary (diffraction) to primary dipoles below the primary dipole peak is 20 Log(T/(R+T)). Below the dipole peak the dipole response follows a 1st order high pass response with pole at F = C/(2S) where C = sound speed and S = the separation. For the secondary, diffraction dipole S = T and as T goes to zero the pole goes to infinity yielding a secondary dipole with zero amplitude at all frequencies. Or, if you like, the diffraction sources cancel everywhere since there are equal in strength, opposite in phase and with T = 0 there is not propagation delay between them.
In the real case the anti-symmetric behavior of the front and rear sources is lost as the frequency rises due to the asymmetries of the front and rear of a conventional driver, but for the highest quality dipole system this lost of asymmetry should be above the useful range of the dipole source.
gedlee said:
Earl, I think you need to reconsider. Given that linear theory applies, we can consider the front and rear radiation separately, and the radiation from the main sources and from the diffraction sources separately as well. Start by considering perfectly anti-symmetric source. That is, the front and rear sources are identical in amplitude but inverted in phase. Along the front baffle surface the wave propagates outward towards the edge. When it reaches the baffle edge a diffracted wave is generated and propagates out from the point of diffraction summing with the original wave at the observation point. This diffracted wave will nominally be inverted relative to the incident wave. We can consider the sum of the original wave and the diffracted wave by the principal of superposition since both of these waves obey the linear wave equation.
Now we can look at the rear wave. When the front and rear sources are anti-symmetric the same result occurs as for the front, but the both the original rear wave from the source and the rear diffraction wave are 180 decrees out of phase with respect to their front counter parts, relative the points of diffraction. Thus, these two additional waves can be summed to the two waves from the front side. All than needs to be considered is the relative path length differences. The point here is that the problem can always be decomposed into these 4 waves, regardless of baffle thickness, and the result expressed as the superposition of the 4 at any point in space. (More than 4 if you want to consider secondary diffraction, but that only complicates the issue without introducing new information.)
As long as the front and rear primary sources are anti-symmetric and the baffle is symmetric, the front and rear the diffractions sources will also be anti-symmetric and will form a secondary dipole with separation equal to the baffle thickness, T, and axis aligned with the primary dipole axis. For the primary dipole the separation (for a circular baffle) is R+T. The relative strength of the secondary (diffraction) to primary dipoles below the primary dipole peak is 20 Log(T/(R+T)). Below the dipole peak the dipole response follows a 1st order high pass response with pole at F = C/(2S) where C = sound speed and S = the separation. For the secondary, diffraction dipole S = T and as T goes to zero the pole goes to infinity yielding a secondary dipole with zero amplitude at all frequencies. Or, if you like, the diffraction sources cancel everywhere since there are equal in strength, opposite in phase and with T = 0 there is not propagation delay between them.
In the real case the anti-symmetric behavior of the front and rear sources is lost as the frequency rises due to the asymmetries of the front and rear of a conventional driver, but for the highest quality dipole system this lost of asymmetry should be above the useful range of the dipole source.
Member
Joined 2003
Hi John,
For a typical cone driver, is the underlying assumption of a perfectly anti-symmetric source a fairly big leap? Acoustic center offset, probable diffraction at the rear inside edge of the driver mounting hole, basket diffraction, as well as spider and/or magnet interference seems to generate a fairly chaotic environment for this to work up to a typical mid/tweeter crossover frequency. (panel ESL etc = different case.)
Paul
For a typical cone driver, is the underlying assumption of a perfectly anti-symmetric source a fairly big leap? Acoustic center offset, probable diffraction at the rear inside edge of the driver mounting hole, basket diffraction, as well as spider and/or magnet interference seems to generate a fairly chaotic environment for this to work up to a typical mid/tweeter crossover frequency. (panel ESL etc = different case.)
Paul
JohnK,
Respectfully, I'm not sold on the idea than an infinitely thin baffle should result in 0 edge diffraction, especially since I've tried it and get greater diffraction. The wavelengths involved are long enough that asymmetries don't explain the increase. Since out of phase sound waves net to zero only if they are travelling in the same direction, I see only 2 explanations.
If 4 waves exist then the 2 diffraction waves are travelling in different direction and the greater pressure change at the edge causes the increase.
Or
Only 3 waves are created, the front and rear wave along with 1 diffraction wave that has a greater change in pressure at the edge.
The edges of these baffles vibrate quite noticeably to the touch. The drivers are mounted in a decoupled manner to the baffles, and the baffles made of solid wood. My only explanations for the increase in vibrations are increased edge diffraction, or at the driver cutout energy is transmitted into the baffle and somehow the shape of the edge focuses that energy to escape right at the pointed edge.
Respectfully, I'm not sold on the idea than an infinitely thin baffle should result in 0 edge diffraction, especially since I've tried it and get greater diffraction. The wavelengths involved are long enough that asymmetries don't explain the increase. Since out of phase sound waves net to zero only if they are travelling in the same direction, I see only 2 explanations.
If 4 waves exist then the 2 diffraction waves are travelling in different direction and the greater pressure change at the edge causes the increase.
Or
Only 3 waves are created, the front and rear wave along with 1 diffraction wave that has a greater change in pressure at the edge.
The edges of these baffles vibrate quite noticeably to the touch. The drivers are mounted in a decoupled manner to the baffles, and the baffles made of solid wood. My only explanations for the increase in vibrations are increased edge diffraction, or at the driver cutout energy is transmitted into the baffle and somehow the shape of the edge focuses that energy to escape right at the pointed edge.