Paul W said:
Yes there are other effect to conside, particualry as the frequency rises. However, I was trying to response to Earl's comment that diffraction is worse for a dipole because of the of pressure being zero at the baffle edge. If the response isn't symmetrical you don't have a true dipole and you don't have zero pressure.
johninCR said:
No, that isn't correct. If two waves are out of phase at the point of summation it makes no difference which directions they are traveling. Summation only concerns amplitude and phase. Not direction or velocity.
But before you can quantify that diffraction is worse or not you have to ascertain whether or not the response is a dipole response. You have to measure the response at 90 degrees and see if there is a good, deep null there. If there isn't, you don't have a dipole response and the front and rear radiation isn't symmetric. So what you get is not a dipole diffraction effect.
As an aside, perhaps what Earl was thinking of was the strength of the individual diffracted waves and not the sum of the two (front and rear). It is true that the diffraction of the front (or rear) wave will be stronger for a zero thickness baffle at higher frequency than it would be for a box speaker where the box depth is much greater than the wave length in question. This is because for the box speaker (with square edges) the short wave lengths only turn a 90 degree corner and then must propagate to the rear of the box where they would turn another 90 degrees. With a zero thickness baffle these same waves would turn 180 degrees, resulting is a stronger diffracted signal. But the front and rear diffraction sources will still cancel, regardless of strength, for a true dipole response generated from symmetric front and rear sources.
John,
They're just small flat baffles, so they're about as close to dipole as I can get. I used a nautilus shape when viewed from the front, so the resulting different driver to edge difference gives them a smooth net response, however, the edges are lit up quite obviously a secondary sound source.
The only thing further I'm going to try is open up the driver cutouts more and add some kind of damping material to prevent sound from going directly from the cone into the wood. Since it's less than an inch away the spl's are quite high, so sound may very well be getting into the baffle there.
Since the effect seemed to increase after I bevelled the edges to a point, I suspect the real answer is that the edge diffraction is just a single source instead of 2 that are out of phase. It would then be greater than with a box because as the compression part of the wave sees the pressure drop from the larger space for expansion, it meets the low pressure portion of the rear wave and vice-versa (a greater pressure change). A further increase may also come from the true full space for the pressure change instead of just 3/4 space at the edge of the box.
A less likely possibility is that since the edge, where the front and rear waves meet, has the highest particle movement that it stimulates more vibration in the relatively thin edge.
I'll report back if the non-drying clay I will use to dampen the driver cutouts, but otherwise I'm not going to spend any more time with this design approach since it didn't work for me. The smooth response due to the infinite driver to edge distances is nice, but not at the expense of quite ugly edge diffraction.
They're just small flat baffles, so they're about as close to dipole as I can get. I used a nautilus shape when viewed from the front, so the resulting different driver to edge difference gives them a smooth net response, however, the edges are lit up quite obviously a secondary sound source.
The only thing further I'm going to try is open up the driver cutouts more and add some kind of damping material to prevent sound from going directly from the cone into the wood. Since it's less than an inch away the spl's are quite high, so sound may very well be getting into the baffle there.
Since the effect seemed to increase after I bevelled the edges to a point, I suspect the real answer is that the edge diffraction is just a single source instead of 2 that are out of phase. It would then be greater than with a box because as the compression part of the wave sees the pressure drop from the larger space for expansion, it meets the low pressure portion of the rear wave and vice-versa (a greater pressure change). A further increase may also come from the true full space for the pressure change instead of just 3/4 space at the edge of the box.
A less likely possibility is that since the edge, where the front and rear waves meet, has the highest particle movement that it stimulates more vibration in the relatively thin edge.
I'll report back if the non-drying clay I will use to dampen the driver cutouts, but otherwise I'm not going to spend any more time with this design approach since it didn't work for me. The smooth response due to the infinite driver to edge distances is nice, but not at the expense of quite ugly edge diffraction.
mige0 said:
Hi Michael,
As a starter before you cut wood, if you want to look in the time domain, the following might be a good short cut.
SLs posted frequency response captures with circular baffles at:
http://www.linkwitzlab.com/diffraction.htm
If you buy into the premise that edge diffraction is min phase, then SLs curve can be converted to a table using:
http://www.pvconsultants.com/audio/utility/spl.htm
This freq domain can then be converted to an impulse response using just about any free ware measurement/analysis system that implements the Hilbert Transform.
Cheers,
Dave
panomaniac said:
My pleasure Maui Michael, glad to help.
DG's site is great, he's so generous with it. I'm a fan of his research and spent a good deal of time there myself, and only absorbed a fraction of it.
DG seemed to nail down the root cause of spaciousness as velocity (not pressure) differences between the right and left ear. How to use that to advantage is the hard part.
This might be an argument to try two gradient subs, aimed at each other from opposite sides of the room, if maximizing LF spaciousness is the goal. Mind you, the standing wave signature may get ugly in such a scenario.
Cheers,
Dave
john k... said:
Perhaps the baffle vibrates because the pressure on onside is + sin(wT) when it's - sin(wt) on the other?
Another thing I had mentioned!
johninCR said:
JohninCR - I'm having a hard time visualizing your baffle & transducer (and its range of operation).
IF its a small baffle then its highly unlikely that the radiation is symmetric at the edge (unless of course you are using two small (sd) drivers (like tweeters) with equal spl.s - the front driver in-phase and the rear out of phase relative to the front driver).
ScottG said:
Here's the only good picture I have. This one is for the 8" B200, which I haven't even bothered to mount. I made 3 pairs almost identical except for the driver cutout, but for the smaller drivers the shortest edge is much sharper since I had more material to work with. The open ring is just decorative, since the nautilus shape alone was way too ugly. I tried mounting both the FE108EZ and FE167 with similar results. Both were quite dipole in response with deep nulls directly at the sides. The 108's were probably closest to exact, since the cone was pretty well centered within the baffle thickness. I don't recall the low end cutoff I used, but that was more as protection from overexcursion, so I matched the baffle roll-off with a corresponding shallow slope on the woofer before rolling off the woofers more steeply around the baffle cutoff of the main driver.
I liked the look, but disliked the sound. I chalked it up to being a good idea that didn't work. I made some out of cardboard first, but ignored that they vibrated a lot expecially at the edges. I thought it was just a materials problem, not a design problem.
I did try a foam ring around the edge which helped to some extent. It was too ugly on it's own, and I didn't like the look of covering the whole thing in grill cloth either. I got disgusted and moved on to my dipole WG's, which I'm happy enough with to stop experimenting.
An externally hosted image should be here but it was not working when we last tested it.
DDF said:
Hey Dave,
It is hard to do. I've heard the soundfield reproduced with with such incredible vastness and detail that all listeners present just blinked, shook their heads and look at each other in disbelief.
For over 20 years I've tried to reproduce that, but haven't. I want to know why.
So I'm trying to put together some intelligent questions for Dr. Dave about the subject. But that's a whole other topic for another thread!
ScottG said:
Maybe not where the driver to edge distance is shortest, but at the longest point it's over 12". Surely the rear radiation is equal to the front at that frequency, and along that longer edge is where the problem seems greatest. You helped me figure out why, and is further evidence to me that dipole does increase the problem of edge diffraction as I have thought all along. Diffraction is caused by the pressure differential at the edge, and with dipole that differential is double.
SL: "Much is hypothesized, little is proven and much is overrated when it comes to di
Hi
Dave, SL and EDGE widly agree as it seems. Unfortunately what is NOT covered is the edge thickness tending to become knife like or the edge becoming absorbent mesh.
If I understood SL right he sees edge treatment rather cosmetic.
Greetings
Michael
Hi
Dave, SL and EDGE widly agree as it seems. Unfortunately what is NOT covered is the edge thickness tending to become knife like or the edge becoming absorbent mesh.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
If I understood SL right he sees edge treatment rather cosmetic.
Greetings
Michael
I think part of the problem is in terminology.
First, are the classic dipole ripples that are maximized with a circular baffle, like displayed by the Edge. As SL states and is commonly accepted, rectangular shapes or off center mounting resulting in varying driver to edge differences easily smooths the responce. To me that isn't edge diffraction, but instead is simply dipole reinforcement and cancellation above the baffle cutoff point.
Then there's what occurs in the region considered baffle step for boxed speakers. The effect of this diffraction at the edges is more pronounced with small speakers because the pressures are higher (because it occurs closer to the source) and because it occurs closer in frequency to where our ears are most sensitive. For my own benefit using numerous test baffles and different edge geometries, I've concluded that with narrow baffles the difference is worth pursuing, and that it's more important with open alignments than with boxes.
JohnK said over on the AC forum that an infinitely thin baffle was the ideal, because the edge diffraction effects would net to zero, so I built some baffles with this in mind. Now I'm out alone on a limb, since I disagree with both SL & JohnK. Depite being lonely, I'll never build another baffle that doesn't make an attempt at minimizing edge diffraction except in the case of large speakers, where the pressures are low enough that it goes unnoticed.
Maybe a pourous edged speaker is the answer, but for now I'll leave that for someone else to try and stick with edge geometries that unfortunately or not result in getting away from dipole.
First, are the classic dipole ripples that are maximized with a circular baffle, like displayed by the Edge. As SL states and is commonly accepted, rectangular shapes or off center mounting resulting in varying driver to edge differences easily smooths the responce. To me that isn't edge diffraction, but instead is simply dipole reinforcement and cancellation above the baffle cutoff point.
Then there's what occurs in the region considered baffle step for boxed speakers. The effect of this diffraction at the edges is more pronounced with small speakers because the pressures are higher (because it occurs closer to the source) and because it occurs closer in frequency to where our ears are most sensitive. For my own benefit using numerous test baffles and different edge geometries, I've concluded that with narrow baffles the difference is worth pursuing, and that it's more important with open alignments than with boxes.
JohnK said over on the AC forum that an infinitely thin baffle was the ideal, because the edge diffraction effects would net to zero, so I built some baffles with this in mind. Now I'm out alone on a limb, since I disagree with both SL & JohnK. Depite being lonely, I'll never build another baffle that doesn't make an attempt at minimizing edge diffraction except in the case of large speakers, where the pressures are low enough that it goes unnoticed.
Maybe a pourous edged speaker is the answer, but for now I'll leave that for someone else to try and stick with edge geometries that unfortunately or not result in getting away from dipole.
BudP said:
Bud,
I think I understand what goes on with your treatment on the cones, but just haven't gotten around to painting them on. Doing it on a baffle is past my understanding as to how it could be benefical. On wood do you recommend the same acrylic paint as for cones? If you told me to carve or glue blocks of foam or wood in the pattern at the edge, I'd probably be gung-ho about doing it instead of reserved.
I have 3 pairs that I consider useless at this point, so I might as well just give it a shot. If it works, it would be quite a testament to your procedure, and I'd probably start painting them on everything that makes sound like my car, lawnmower and noisy generator.
Re: SL: "Much is hypothesized, little is proven and much is overrated when it comes to di
Hi again Michael,
Yes, I guess that'll be even quicker.
The EDGE is fairly accurate but not perfect. I took data from SEAS' anechoic chamber, on IEC baffle, and in box, and looked at the predictive error for the EDGE and the BDS based on the baffle and box dimensions along with the driver mounting. Attached is what I found.
http://www3.sympatico.ca/dalfarra/Diffraction_model_errors.gif
The BDS, in this example, converged at 5.5 dB baffle gain, whereas the EDGE converged at 6 dB. I believe the BDS is correct, the EDGE in error. The EDGE doesn't factor in the impact on level of the longer path difference to the baffle edge, relative to the direct on axis. This is really only important when trying to determine baffle diffraction compensation, but its worth noting, as the EDGE's error will be even greater for a large baffle, such as that in an OB. At longer listening distances, the error is probably in the noise.
These results were taken with a tweeter so some of the low end error may be due to the reduced level there, and the difficulty getting consistent results (chamber noise etc). For the higher end, I'm not sure what's happening. If there was a woofer colacated in the box, (there would only be one driver in the IEC baffle), this could be reflection off the woofer surround, or diffraction from the woofer cone itself, but I don't know. I doubt its the mic stand, SEAS wouldn't make that mistake.
Its not clear at all why SL would consider edge treatment merely cosmetic. Certainly the power response is important, but less so than the direct sound.
mige0 said:
Hi again Michael,
Yes, I guess that'll be even quicker.
The EDGE is fairly accurate but not perfect. I took data from SEAS' anechoic chamber, on IEC baffle, and in box, and looked at the predictive error for the EDGE and the BDS based on the baffle and box dimensions along with the driver mounting. Attached is what I found.
http://www3.sympatico.ca/dalfarra/Diffraction_model_errors.gif
The BDS, in this example, converged at 5.5 dB baffle gain, whereas the EDGE converged at 6 dB. I believe the BDS is correct, the EDGE in error. The EDGE doesn't factor in the impact on level of the longer path difference to the baffle edge, relative to the direct on axis. This is really only important when trying to determine baffle diffraction compensation, but its worth noting, as the EDGE's error will be even greater for a large baffle, such as that in an OB. At longer listening distances, the error is probably in the noise.
These results were taken with a tweeter so some of the low end error may be due to the reduced level there, and the difficulty getting consistent results (chamber noise etc). For the higher end, I'm not sure what's happening. If there was a woofer colacated in the box, (there would only be one driver in the IEC baffle), this could be reflection off the woofer surround, or diffraction from the woofer cone itself, but I don't know. I doubt its the mic stand, SEAS wouldn't make that mistake.
Its not clear at all why SL would consider edge treatment merely cosmetic. Certainly the power response is important, but less so than the direct sound.
I don't think the discussion of what edge treatment is best is going to lead anywhere because I don't think there is agreement on what is meant by dipole response. From my point of view a driver operation on an open baffle produces a dipole response only when the front and rear responses are symmetrical (or very close to it). If they aren't symmetrical it's not a dipole. It's just an open baffle response.
When I say that an zero edge thickness eliminates diffraction I am referring to the region of the response where there is front to rear symmetry. When you speak of the peaks and nulls above the dipole peak, and the smoothing that comes with rectangular or odd shaped baffles this is still a case where front and rear are symmetric and the smoothing is a result of the variation in path lengths to the baffle edge so that cancelation and summation doesn occur at the same frequecy all around the baffle edge. It is not diffraction, as JohninCR pointed out.
Diffraction effects in a flat baffle, open baffle (note open baffle as opposed to dipole) speaker only occur as a result of asymmetries in the front and rear response of because of finite baffle thickness. Wings and rearward extentions are a source of asymmetry. At lower frequencies things like diffraction from the cone surround, motor blockage, etc are usually not significant. Wings and rearward extensions may have an effect at low frequency, depending on their angle and length. At higher frequencies all these effects introduce asymmetries which result in loss of true dipole behavior and introduce diffraction effects into the net response.
Also as JohninCR states, the baffle step in a conventional speaker is a diffraction effect. But the dipole roll off is not. It is cancelation between the front and rear primary radiation. One thing that has to be remembered is that the rear response that wraps around the baffle to sum to the front direct sound is the 90 degree off axis response of the rear source. Obviously in the real driver case, this will only be symmetric up to the point were the driver becomes directional even before things like motor blockage, etc are considered.
If you concern is using an open baffle mounted driver into the frequency range well above the point where dirver directionality and front to rear asymmetry is in play then you are really dealing with, more or less, conventional diffraction effects. These can be better or worse than for a conventional speaker and will depend on all ypes of things including baffle shape, edge treatment, driver poition, etc. But as I have said, this is beyond the point where the driver is acting as a dipole.
When I say that an zero edge thickness eliminates diffraction I am referring to the region of the response where there is front to rear symmetry. When you speak of the peaks and nulls above the dipole peak, and the smoothing that comes with rectangular or odd shaped baffles this is still a case where front and rear are symmetric and the smoothing is a result of the variation in path lengths to the baffle edge so that cancelation and summation doesn occur at the same frequecy all around the baffle edge. It is not diffraction, as JohninCR pointed out.
Diffraction effects in a flat baffle, open baffle (note open baffle as opposed to dipole) speaker only occur as a result of asymmetries in the front and rear response of because of finite baffle thickness. Wings and rearward extentions are a source of asymmetry. At lower frequencies things like diffraction from the cone surround, motor blockage, etc are usually not significant. Wings and rearward extensions may have an effect at low frequency, depending on their angle and length. At higher frequencies all these effects introduce asymmetries which result in loss of true dipole behavior and introduce diffraction effects into the net response.
Also as JohninCR states, the baffle step in a conventional speaker is a diffraction effect. But the dipole roll off is not. It is cancelation between the front and rear primary radiation. One thing that has to be remembered is that the rear response that wraps around the baffle to sum to the front direct sound is the 90 degree off axis response of the rear source. Obviously in the real driver case, this will only be symmetric up to the point were the driver becomes directional even before things like motor blockage, etc are considered.
If you concern is using an open baffle mounted driver into the frequency range well above the point where dirver directionality and front to rear asymmetry is in play then you are really dealing with, more or less, conventional diffraction effects. These can be better or worse than for a conventional speaker and will depend on all ypes of things including baffle shape, edge treatment, driver poition, etc. But as I have said, this is beyond the point where the driver is acting as a dipole.
Re: Re: SL: "Much is hypothesized, little is proven and much is overrated when it comes to di
Dave
I have not followed this whole discussion (edge diffraction), but much of what I have read is incorrect - or at the very least the terminology is wrong.
In the test you talk about, did you take the data? You compared the same speaker in a closed box of the same baffle size as an OB? Do the programs that you quote handle baffles with free edges as well as box edges? These two things are quite different and very tricky to do correctly. I did a lot of work on edge diffraction modeling about 20 years ago.
I would disgree about the relative importance of direct sound and power response - at the very least I would rate them as equal, but I might even lean to the power/polar response as being more important depending on the room. Clearly the room and the speaker setup are parmount in making this comparison. The line of sight direct field is all that matters in an anechoic chamber and the reverb field is about all you hear in a reverberation chamber. Real rooms are somewhere in between.
DDF said:
Dave
I have not followed this whole discussion (edge diffraction), but much of what I have read is incorrect - or at the very least the terminology is wrong.
In the test you talk about, did you take the data? You compared the same speaker in a closed box of the same baffle size as an OB? Do the programs that you quote handle baffles with free edges as well as box edges? These two things are quite different and very tricky to do correctly. I did a lot of work on edge diffraction modeling about 20 years ago.
I would disgree about the relative importance of direct sound and power response - at the very least I would rate them as equal, but I might even lean to the power/polar response as being more important depending on the room. Clearly the room and the speaker setup are parmount in making this comparison. The line of sight direct field is all that matters in an anechoic chamber and the reverb field is about all you hear in a reverberation chamber. Real rooms are somewhere in between.
JohninCR,
I honestly do not know to what SPL the paint applied in an EnABL pattern will control baffle edge diffraction. In the closed box I own, it is useful to 107 dB at 100 Hz and up. The box is really too small to use as a reliable tool for lower frequencies than this and even 100
hz is pushing it. Through the frequencies I do have some trust in, the square cut edge on the baffle does not exhibit any diffraction that is audible and neither do the various drivers mounted on the surface.
What this means for OB control I cannot say anything about, below 100 Hz. Above that point I am certain it will control the diffraction of a hard edge or a rounded edge. I do not know how the pattern will deal with the holes at the edge, or cork, or felt, other than to keep those different surfaces and attendant edges from creating standing waves back across the OB surface.
I only offer the pattern as a way to eliminate two of the formidable number of problems with speakers mounted on a finite surface, edge diffraction and transient standing wave reflections.
Both of these can be addressed in the boundary layer with some very small tools. This is not intuitive at all, until you begin to think about all of the other small scale disruptions that are applied to a surface, in the boundary layer, to effect a variety of damping effects.
This is just a different use of that boundary layer. I certainly do not think it is a panacea, though, oddly, it does work rather nicely on a wall of windows, an orchestra half shell, turntables and likely any other surface that has edges and can develop standing waves on it's boundary layer with air.
So, try it out and tell us what it will not do and what it seems to help with in your situation, with a plethora of OB shapes just laying around. After all, if it proves useless to you you can sand it off and tell me I am just confused...... won't hurt my feelings to be wrong.
Bud
I honestly do not know to what SPL the paint applied in an EnABL pattern will control baffle edge diffraction. In the closed box I own, it is useful to 107 dB at 100 Hz and up. The box is really too small to use as a reliable tool for lower frequencies than this and even 100
hz is pushing it. Through the frequencies I do have some trust in, the square cut edge on the baffle does not exhibit any diffraction that is audible and neither do the various drivers mounted on the surface.
What this means for OB control I cannot say anything about, below 100 Hz. Above that point I am certain it will control the diffraction of a hard edge or a rounded edge. I do not know how the pattern will deal with the holes at the edge, or cork, or felt, other than to keep those different surfaces and attendant edges from creating standing waves back across the OB surface.
I only offer the pattern as a way to eliminate two of the formidable number of problems with speakers mounted on a finite surface, edge diffraction and transient standing wave reflections.
Both of these can be addressed in the boundary layer with some very small tools. This is not intuitive at all, until you begin to think about all of the other small scale disruptions that are applied to a surface, in the boundary layer, to effect a variety of damping effects.
This is just a different use of that boundary layer. I certainly do not think it is a panacea, though, oddly, it does work rather nicely on a wall of windows, an orchestra half shell, turntables and likely any other surface that has edges and can develop standing waves on it's boundary layer with air.
So, try it out and tell us what it will not do and what it seems to help with in your situation, with a plethora of OB shapes just laying around. After all, if it proves useless to you you can sand it off and tell me I am just confused...... won't hurt my feelings to be wrong.
Bud