The speakers I'm working on have the bass range split into sub and bass both covering the other but crossed at difference points and interact with the room from different points. Each speaker consists of a dual MLTL sub, dual acoustic suspension Q0.62 and an upper MTM. All drivers other than the AMT are the same. The MTM will be high passed around 100. The AS bass will be crossed low pass below 330 and the MLTL below 150. Both the MLTL subs and AS drivers are high passed above 27. This is active and allows for four points of subbass loading of the room with a stereo pair. Height adjustable so floor coupling can be taken advantage of if needed.
As I've said before, room modes, floor bounce et al be damned, we'll get'm all one day! 😀
As I've said before, room modes, floor bounce et al be damned, we'll get'm all one day! 😀
And yet I did hear it with monopoles when experimenting with toe-in and controlled reflections. The same speaker, the same amount of diffraction.
As I've said before, use two subs in the near field driven by a stereo signal and you'll have it.
😛
Well for one I don't have any stereo system set up where we've been living for the last 6 months (its all packed away in storage) to suck up time listening to and tinkering with speakers...so posting on the forum helps keep me thinking about audio and keep things fresh in my mind...(use it or lose it!)
The other is that I'm a very fast typist. 😉 What looks like a humongous message to you probably only took me 5-10 minutes to type 😀 It takes me longer to think about what I might say than actually put it into words and type it.
Some of the ideas I'm trying to get across don't necessarily lend themselves to abbreviation either, especially when there is a large variation in knowledge and experience among those reading. To get my point across I can't assume everyone already knows and understands everything that leads up to a point I'm trying to make...one of my failings is I try to explain ideas in the most minute detail.
No its not common especially with typical cone and dome systems. A typical wide dispersion dome tweeter mounted on a typical flat baffle is a diffraction catastrophe even if it is offset.
I think some degree of directivity control of midrange and treble is necessary to fully pull it off - whether that's because it helps to reduce early side wall reflections, or whether the directivity is more important in reducing diffraction from the baffle edge I'm not sure, but I'm leaning more and more towards the latter. That it helps with both is a good thing.
Last edited:
With regard to loudspeakers disappearing I’d really discuss in-room-reflections and baffle edge diffraction as separate issues:
The perceived distance between sound source and sound stage is a function of the several perceived reflections in the room. If the earliest and/or strongest reflections come mainly from the front and (forward area of the) side walls, we perceive the sound stage behind the drivers. This applies to dipoles and loudspeakers with small baffles and/or generously rounded baffle edges. If the reflections come mainly from the back wall or rear area of the side walls, as happens with horns for instance, we perceive the sound stage in front of the drivers. If the directivity of the loudspeakers is sufficiently homogenous and the reflection pattern from the walls is left-right symmetric and free from sudden peaks and holes in the reflection coefficient, we should have a convincing stereo stage that doesn’t appear to originate from particular speakers.
I have heard large planars/ESLs which completely disappeared even though their diffraction pattern had to be a total mess. The “bad diffraction” mark of those planars was the diffuse rendition of the virtual sources on the stereo stage – singers with their mouth reaching from one speaker to the other etc.
I’m well aware that a well controlled dispersion pattern of a loudspeaker implies a controlled diffraction pattern. But even the speaker with the best controlled diffraction will have the sound glued to it, if the reflection pattern which the room returns to our ears is a total mess which can’t be divided out by our brain.
Rudolf
The perceived distance between sound source and sound stage is a function of the several perceived reflections in the room. If the earliest and/or strongest reflections come mainly from the front and (forward area of the) side walls, we perceive the sound stage behind the drivers. This applies to dipoles and loudspeakers with small baffles and/or generously rounded baffle edges. If the reflections come mainly from the back wall or rear area of the side walls, as happens with horns for instance, we perceive the sound stage in front of the drivers. If the directivity of the loudspeakers is sufficiently homogenous and the reflection pattern from the walls is left-right symmetric and free from sudden peaks and holes in the reflection coefficient, we should have a convincing stereo stage that doesn’t appear to originate from particular speakers.
I have heard large planars/ESLs which completely disappeared even though their diffraction pattern had to be a total mess. The “bad diffraction” mark of those planars was the diffuse rendition of the virtual sources on the stereo stage – singers with their mouth reaching from one speaker to the other etc.
I’m well aware that a well controlled dispersion pattern of a loudspeaker implies a controlled diffraction pattern. But even the speaker with the best controlled diffraction will have the sound glued to it, if the reflection pattern which the room returns to our ears is a total mess which can’t be divided out by our brain.
Rudolf
Edge diffraction: Mitgation of effects using asymmetrically mounted tweeter
That was my starting point: The frequency response deviations between both speaker will not track with
listener's offset. That is why frequency response deviation due to angle should be minimised.
I provided an example simulation using a 20mm tweeter and a conventional sharp edged baffle, which shows,
that deviations with angle can be minimised by using asymmetric mounting of the tweeter.
(See the three pictures at bottom of post ...)
http://www.diyaudio.com/forums/mult...-pattern-stereo-speakers-237.html#post3764121
If you use that technique, also frequency response deviations between both speaker will be mitigated,
otherwise they will not.
This holds, as long as you listen both speakers along the "short path" from tweeter to inner baffle edge.
The exeption mentioned occurs if you toe in both speakers to such extent, that a side shift of the LP will
cause the horizontal angle to be on the short side for one speaker while on the long side for the
other:
In that case a symmetric mounting might have (a vague chance for) less differences between L+R,
compared to an optimized asymmetric one (even that will not hold for small off axis angles, because
the asymmetry smoothing effect by far dominates differences over angles in a wide range of angles).
Furthermore such exceptional LP's are outside the listening zone in home environments usually:
The argument only works for setups listened near axis (00-degrees), where the detrimental effects of
edge diffraction are worst , not in terms of L-R asymmetry maybe, but in overall FR-ripple.
Thus a sarcastic interpretation of your proposal would be:
"Make your system as worse as possible, then you may have more symmetry due to LP-shift in some special
cases, which are irrelevant in practice." (No offence meant!)
(Possibly relevant just for "00-degrees straight on axis toe-in" or "RLX" setups and even for those i cannot
see any proof of advantage for the symmetric setup, at least for small off axis angles.)
You would have to provide at least some data to support your point of view:
I did provide an example simulation at least comparing angle vs. asymmetry effects.
We both can agree easily - i am pretty sure - in that it is necessary (also) to use
alternatives ( absortion, roundoff etc.) to mitigate the edge diffraction problem.
If diffraction is mitigated that way to a sufficient degree, then you can afford tweeters which are
mounted symmetrically.
On a conventional baffle with unrounded edges, a symmetric tweeter is a misconstruction to me.
This is my point of view, until one provides data that says otherwise ....
Your arguments are kind of "estethical" to me and also have some "inherent logic", but IMO simply
miss weighing magnitudes of the different effects to deal with.
Many to most speakers come with an (often invisible) sticker saying:
"Caution, do not listen on axis".
Which is ignored mostly due to the well known argument:
"But why ? It can't be wrong 'cause that's how we did it all the time."
Kind Regards
That was my starting point: The frequency response deviations between both speaker will not track with
listener's offset. That is why frequency response deviation due to angle should be minimised.
I provided an example simulation using a 20mm tweeter and a conventional sharp edged baffle, which shows,
that deviations with angle can be minimised by using asymmetric mounting of the tweeter.
(See the three pictures at bottom of post ...)
http://www.diyaudio.com/forums/mult...-pattern-stereo-speakers-237.html#post3764121
If you use that technique, also frequency response deviations between both speaker will be mitigated,
otherwise they will not.
This holds, as long as you listen both speakers along the "short path" from tweeter to inner baffle edge.
The exeption mentioned occurs if you toe in both speakers to such extent, that a side shift of the LP will
cause the horizontal angle to be on the short side for one speaker while on the long side for the
other:
In that case a symmetric mounting might have (a vague chance for) less differences between L+R,
compared to an optimized asymmetric one (even that will not hold for small off axis angles, because
the asymmetry smoothing effect by far dominates differences over angles in a wide range of angles).
Furthermore such exceptional LP's are outside the listening zone in home environments usually:
The argument only works for setups listened near axis (00-degrees), where the detrimental effects of
edge diffraction are worst , not in terms of L-R asymmetry maybe, but in overall FR-ripple.
Thus a sarcastic interpretation of your proposal would be:
"Make your system as worse as possible, then you may have more symmetry due to LP-shift in some special
cases, which are irrelevant in practice." (No offence meant!)
(Possibly relevant just for "00-degrees straight on axis toe-in" or "RLX" setups and even for those i cannot
see any proof of advantage for the symmetric setup, at least for small off axis angles.)
You would have to provide at least some data to support your point of view:
I did provide an example simulation at least comparing angle vs. asymmetry effects.
We both can agree easily - i am pretty sure - in that it is necessary (also) to use
alternatives ( absortion, roundoff etc.) to mitigate the edge diffraction problem.
If diffraction is mitigated that way to a sufficient degree, then you can afford tweeters which are
mounted symmetrically.
On a conventional baffle with unrounded edges, a symmetric tweeter is a misconstruction to me.
This is my point of view, until one provides data that says otherwise ....
Your arguments are kind of "estethical" to me and also have some "inherent logic", but IMO simply
miss weighing magnitudes of the different effects to deal with.
Many to most speakers come with an (often invisible) sticker saying:
"Caution, do not listen on axis".
Which is ignored mostly due to the well known argument:
"But why ? It can't be wrong 'cause that's how we did it all the time."
Kind Regards
Last edited:
Hi Rudolf,
can you please give me some hints, what in your personal view would make
up a "reflection pattern" that can be "divided out by our brain" easily ?
(One that is not a "mess" but the contrary ...)
Kind Regards
Last edited:
This is the type of discussion that has gone on since the first Western Electric amplified system was installed in a theater in the late 1920s. It will continue far into the future and is the one factor that drives our interesting and rewarding hobby. Keep it up guys.
It may well be both . . . good imaging (and good sound in general) is the sum of a whole lot of little things done right. That uniform off-axis response is a good thing is so intuitively obvious that I'm tempted to doubt it 😉 . . . that the "original" source of most music is itself a diffraction engine leads me to question its importance despite thinking (in some cases I'd say "knowing") that I can hear it. Early reflections, though, simply must be managed.
My listening room has very little of "early sidewall" . . . there's the dipole null, of course, but also a large opening to the dining room on one side and a large heavily draped window on the other, both right where first reflections would form. But there is also, on the left side, a brick fireplace that juts into the room over essentially the entire front-to-back dimension of the listening area that was for ever so long the bane of my existence. Apart from what it does to "imaging" the fireplace produces a left-channel-only "distortion" that sounds very much like aliasing . . . and I'd be embarrassed to recount the number of times over the years that I swapped amplifiers or speakers or both trying to figure out what was "wrong" with the left channel. Slow learner, I guess . . . 😱
A drape of movers blanket hung from the mantle pretty much cures it, and it doesn't have to cover the whole fireplace, either . . . just one side and corner.
It's the little things . . .
Reflections from the diffusive front wall don't cause any such problem, probably because of the much greater delay. There's no negative impact on "localization" of specific sources, while the sense of spaciousness and "depth" is substantially reduced if absorption (rigid fiberglass or more mover's blankets) replaces diffusion along that wall. Those reflections seem to have a beneficial effect similar to putting a shell behind the performers on an otherwise too-large stage. The effect is all but absent with boxes placed where the dipoles usually sit, and the illusion of "reality" is much diminished.
The room, and pattern control to "manage" the room, makes a substantial difference. One cannot ignore any of it . . .
It's all in Toole's book, basicly. 🙂
Make sure all reflections are down -15 dB in the first 15 ms. -20 dB would be even better, but usually much harder to achieve.
Make sure that the room/wall situation to the front, left and right of your stereo triangle is as symmetric as possible. If you don't have much control over the room, you better reduce the size of the stereo triangle radically.
Sit in a position that ensures equal distance to both speakers.
Rudolf
But of course it is . . . practical . . . that's what DIY is all about.
As a "practical" matter I doubt that any of us can build a significantly "better" speaker than the best of the commercial offerings . . . what we can do (that the commercial offerings rarely accomplish) is build (or at least select) exactly the right speaker for our listening environment.
... any preference on absorbtion or diffusion to achieve that ?
e.g. due to early (ipsi) lateral side reflections ...
tolerate them, absorb them partially , diffuse them, avoid them by directivity pattern ?
Alltogether ?
Last edited:
Floor and ceiling are by far the hardest to manage . . . I suspect that the "line array" aspect of planar dipoles is an advantage over point-source dipoles in that regard.
I heavily rely on the directivity pattern of my dipoles for the front and ipsilateral reflections. I use 5 cm thick Basotect patches (50x50 cm) at the front reflection point for last refinement. I use 5cm thick Basotect sheets as edge absorbers in the front room corners. Those are mandatory. If they are not in place, the stereo stage will split into a central 30° area and another 30° area in the left and right room corner. Spectacular panorama, but far from hifi.
I yet have to test, if and how diffusion would work instead of absorption.
I try to keep to the "stereo" aspects in this discussion 🙂
Rudolf
My experience suggests that diffusion along one wall very much reduces the corner reflection problem. Without diffusion (or absorption) the corners become very pronounced secondary "point sources" . . . (I absorb the front wall/ceiling corner, which is not diffused by the wall-to-wall bookcase that does not reach to the ceiling).
I'm not at all convinced that the ear does all that good a job of discriminating the direction of those early reflections. The ceiling bounce, with just a few milliseconds delay, can mask a whole lot of desirable "stereo" cues . . .
Last edited:
But this is not at all what Floyd believes. He believes in wide directivity for more early reflections. This is precisely where I disagree with him.
IMO opinion the ceiling bounce does nothing for or against localization cues ('"stereo" cues') because they arrive equally at each ear. They can be a large source of coloration but not much effect on imaging. Greisinger states that our hearing simply ignores vertical reflections.
Ripple due to edge diffraction ...
I am not sure, if one is allowed to think any ripple
(that is also accompanied by non smooth group delay for free)
that is "symmetric" on both channels is not detrimental
to imageing.
Some assumptions/speculations i was thinking about lately:
- following a single instrument or voice in an auditory scene,
..(that is reproduced by a stereo system if you like..)
..may not be a different task from "localizing" it:
..Localization and "extraction/following"
..(of a souce in an auditory scene)
..might be tasks, which depend/rely on each other
..and thus go together.
..Having "picked something", that something might be
..localizeable, but that may hold vice versa too.
- That would mean localization relies on sound sources, which have
..known properties e.g. spectral (long term) , transient (narrowband
..correlation in onsets of musical instruments and spoken consonants)
..etc.
- if spectral and temporal aspects are distorted
..(e.g. by a sharp 8Khz peak) that may affect extracting the source
..as well as localizing it, because we have to deal with a
..deviation of the properties of that type of source, as we are used
..to perceive it.
- additionally the example 8Khz peak may cause a perceived
..elevation of a phantom image, because of the peak coinciding with
..a directional band, which also might contribute to "smear" the phantom
..image in space (dependent on when, how the 8Khz peak is
.."hit" by the spectrum of our source under question).
Although "symmetry" is beneficial for stereo as such to work
(the system depends on summing localization), one cannot say
that any FR ripple which is (seems!) symmetrical should be
"harmless" due to imageing:
1) I would in no way rely on that.
2) I believe that most irregularities concerning
..a) diffraction at the speaker's edges and
..b) (early) room reflections correlated with the direct sound
inherently tend to also develop (fluctuating) asymmetries due to a
practical (real) listening position, also because the listener is not
"clamped".
Those asymmetries add to the ones inherent to stereo ...
To me "stereo imageing" is mainly a "cue stabilization task" from a
technical point of view.
Kind Regards
I am not sure, if one is allowed to think any ripple
(that is also accompanied by non smooth group delay for free)
that is "symmetric" on both channels is not detrimental
to imageing.
Some assumptions/speculations i was thinking about lately:
- following a single instrument or voice in an auditory scene,
..(that is reproduced by a stereo system if you like..)
..may not be a different task from "localizing" it:
..Localization and "extraction/following"
..(of a souce in an auditory scene)
..might be tasks, which depend/rely on each other
..and thus go together.
..Having "picked something", that something might be
..localizeable, but that may hold vice versa too.
- That would mean localization relies on sound sources, which have
..known properties e.g. spectral (long term) , transient (narrowband
..correlation in onsets of musical instruments and spoken consonants)
..etc.
- if spectral and temporal aspects are distorted
..(e.g. by a sharp 8Khz peak) that may affect extracting the source
..as well as localizing it, because we have to deal with a
..deviation of the properties of that type of source, as we are used
..to perceive it.
- additionally the example 8Khz peak may cause a perceived
..elevation of a phantom image, because of the peak coinciding with
..a directional band, which also might contribute to "smear" the phantom
..image in space (dependent on when, how the 8Khz peak is
.."hit" by the spectrum of our source under question).
Although "symmetry" is beneficial for stereo as such to work
(the system depends on summing localization), one cannot say
that any FR ripple which is (seems!) symmetrical should be
"harmless" due to imageing:
1) I would in no way rely on that.
2) I believe that most irregularities concerning
..a) diffraction at the speaker's edges and
..b) (early) room reflections correlated with the direct sound
inherently tend to also develop (fluctuating) asymmetries due to a
practical (real) listening position, also because the listener is not
"clamped".
Those asymmetries add to the ones inherent to stereo ...
To me "stereo imageing" is mainly a "cue stabilization task" from a
technical point of view.
Kind Regards
Last edited:
But they don't . . . they have the same inter-aural delay as any other sound from the originating loudspeaker. That wouldn't effect localization of the loudspeaker, but it might mask cues embedded in the signal. I'm not convinced either way . . .
They'd have to be identified as vertical reflections to be ignored as vertical reflections. Not sure how the ear would accomplish that . . .
It has to be room dependent. In a small room the early reflections inevitably say "small room" to our perception . . . in a larger room the longer delay may permit them to blend with the room signature in the recording and simply enhance the sense of "spaciousness". In a small room I have to listen near field to ignore the room . . .
- Status
- Not open for further replies.