The reflector itself shall emit its own sound when being excited, and/or absorb some sound in the mean time, no? I'm not sure how significant it'd be, though...
Ideally, used for reflecting sound wave, it should be dense, heavy, inert with very smooth surface (flat or curvy) and rigidly secured.
In this regard, corrugated cardboard and "loudspeaker design cookbook" are not ideal refecters. They shall put their own sounds into the overall performance.
(Sorry if this has been brought up before. I searched a bit, to several pages back and didn't find it. )
Ideally, used for reflecting sound wave, it should be dense, heavy, inert with very smooth surface (flat or curvy) and rigidly secured.
In this regard, corrugated cardboard and "loudspeaker design cookbook" are not ideal refecters. They shall put their own sounds into the overall performance.
(Sorry if this has been brought up before. I searched a bit, to several pages back and didn't find it. )
Yes, the reflector should be as stiff as possible. The "Loudspeaker Design Cookbook" should be rigid enough. At least the German edition is 🙂
I had to coat my paper cone on the indside with a mix of glue and risotto. For measurements it was also fixed with a rubber band to the enclosure.
I had to coat my paper cone on the indside with a mix of glue and risotto. For measurements it was also fixed with a rubber band to the enclosure.
Let me rewind this thread a bit because there are some essential points here already:
Anyway, I would agree. Although more imprecise, I have to say that I like Griesinger's term "early spatial impression" a bit more.
Toole's comment on this from chapter 20 is "a flawed "feeling of space" is preferable to one that is insufficient". While that is true, "some ASW" does so much for realistic sound reproduction. IMO it weighs
so much more than e.g. absence of distortion.
explain as to why they create their sensation.
We have many recordings that benefit from it. All of those are either live recordings or recordings with sufficient artificial or better natural spatial cues.
There is also a good amount of recordings that is neutral.
We probably have an average sized collection of about 600 CDs containing virtually every genre, recorded over a period of the last 5 decades. Only recordings of choirs are underrepresented.
Under the circumstances that delays >= 20ms is wishful thinking in regard to regular living rooms, I think what is required and achievable, contains but is probably not limited to:
- Sufficient "early spatial impressions" / ASW, which is a result of wide dispersion; wider than +-45°. See also the 3dB…5dB “rule” that Klippel found (e.g. Toole chapter 20).
- controlled directivity if not constant
- delay of reflections > 6ms
- symmetrical reflections
But that already raises the next questions:
- Is constant directivity optimum at all ?
- How wide should dispersion be ? See the 3..5dB “rule” again ? And how wide in what frequency range if directivity is not constant but only “controlled” ?
- and if the 3..5dB "rule" applies, does it really matter from where exactly (front wall and side wall vs. side wall only) around one speaker the reflections come ?
Thoughts, experiences ?
Agreed.
Well, at least in chapter 7 of his book he mainly distinguishes between reflections < 80ms and >80ms but as we know even 20ms are not really feasible in regular living rooms.
I think your question was directed mainly to Earl ?
Anyway, I would agree. Although more imprecise, I have to say that I like Griesinger's term "early spatial impression" a bit more.
Toole's comment on this from chapter 20 is "a flawed "feeling of space" is preferable to one that is insufficient". While that is true, "some ASW" does so much for realistic sound reproduction. IMO it weighs
so much more than e.g. absence of distortion.
Yet, dipoles are not quite straight forward in this context because of the relative lack of lateral reflections. You toe them in and they produce reduced IACC (increased directivity) but this does not fully
explain as to why they create their sensation.
SL's take on it: Reflection delay >6ms (>1m from boundaries).
Let me state my experience in another way: There is NO recording our collection that suffers from increased ASW.
We have many recordings that benefit from it. All of those are either live recordings or recordings with sufficient artificial or better natural spatial cues.
There is also a good amount of recordings that is neutral.
We probably have an average sized collection of about 600 CDs containing virtually every genre, recorded over a period of the last 5 decades. Only recordings of choirs are underrepresented.
As a result of 2, I think that is not required.
Under the circumstances that delays >= 20ms is wishful thinking in regard to regular living rooms, I think what is required and achievable, contains but is probably not limited to:
- Sufficient "early spatial impressions" / ASW, which is a result of wide dispersion; wider than +-45°. See also the 3dB…5dB “rule” that Klippel found (e.g. Toole chapter 20).
- controlled directivity if not constant
- delay of reflections > 6ms
- symmetrical reflections
But that already raises the next questions:
- Is constant directivity optimum at all ?
- How wide should dispersion be ? See the 3..5dB “rule” again ? And how wide in what frequency range if directivity is not constant but only “controlled” ?
- and if the 3..5dB "rule" applies, does it really matter from where exactly (front wall and side wall vs. side wall only) around one speaker the reflections come ?
Thoughts, experiences ?
Last edited:
From my experience there have to be reflections coming from the sides and from an elevated angle. As long as the level, delay and spectrum of these reflections amongst themselves and amongs the direct sound have a certain relationship, they will create this distinct spatial impression where sounds perceptually fill a space without creating unnatural sounding effects.
While experimenting with cone reflectors I got a wide but still unnatural spatial representation. Height was missing. On the other hand, without any reflector a high directivity fullrange speaker generates strong elevated reflections when pointed to the ceiling. Some sounds like voices in the center came from unnatural high locations.
I think smooth polars is the greater goal. Constant directivity can be achieved a a variety of ways, some of no value. When studying horns Keele showed that multicell horns had constant directivity and very crappy polars. Abberations in one plane would compensate for those in the other.
Nobody has shown that flat power response is necessary or even good. Linkwitz and Vanderkooy show that holes in the power response are benign. The better approach is to aim for exceedingly smooth response over a broad listening window. i.e. well behaved polar curves with a consistant front lobe.
How much directivity? Toole tests showed people not liking dipoles as much as systems with wider dispersion. In mono tests they were severely downgraded, in stereo tests not as much. If we like pinpoint imaging then we will give up spaciousness, but that can be a personal choice. I do think enough studies show that vertical reflections (floor and ceiling) are hard to seperate from the direct sound and end up perceived as resonse errors, so lateral dispersion is preferable to vertical dispersion.
Later arrival of lateral reflections is a good thing. I'm not sure that 20ms is an insurmountable barrier. We could certainly devise a system that gives a more ideal arrival pattern. It could be done electronically as Kantor did, or simply with directional secondary speakers and side wall diffusers. 5.1 can easily achieve this also, and will allow us to have more directional fronts without the overly dry sound they may givee in 2 channel.
David S.
After leaning different ways over the years I've finally settled on the approach of "controlled directivity" above a certain frequency being the optimal for most typical living rooms, at least for larger speakers intended to be listened to at 2.5m+, and whose imaging concious listening is done primarily near on axis.
Controlled directivity meaning "not too directional nor too wide, over as wide a frequency range as possible", (yes, a little bit vague on numbers) constant directivity being one example but not the only possible approach.
Whilst being very directional, or very directional only at high frequencies (large full range drivers etc) is generally a bad thing, excessively wide dispersion comes with it's own host of problems, the biggest one being baffle diffraction effects, particularly through the critical presence region.
The usual approach to combat this is very narrow baffles with rounded edges, everything flush mounted etc, but is this really necessary, or the best approach ? Do we really need 180 degree dispersion at midrange and treble frequencies ? I really don't think so, and it cant be achieved all the way up to 20Khz anyway, practically speaking.
Constrain the directivity with a device like a CD wave-guide, and the bad effects of baffle edge diffraction more or less go away - even an ugly looking baffle has little effect on the response, at least at the important higher frequencies, and imaging is greatly improved.
As you say, CD can be achieved in some ways that don't give good polar responses, and I'm not sure that CD is the optimal response (when you still have to blend it into omni-directional response at some cut-off frequency) but I think it's a big step in the right direction.
I have to agree. I've never found small dips in the power response to be a problem or even that noticeable, unless you're at the far end of a very reverberant room, and if you are you have more problems to worry about than a power response dip.
If you're within the boundaries of a normal listening set up where the direct field is dominant over the reverberant field, the listener axis frequency response really does seem to dominate how it sounds, with a gentle dip even of a few dB in the power response mattering not at all.
For this reason I'm far more likely to favour distortion and dynamic range considerations and lean towards a higher crossover frequency for a mid-treble crossover, even though on paper the dip in the power response might look bad.
The reality is that provided that the response is uniform over a reasonable listening angle window, a power response dip doesn't sound bad in any reasonable listening set-up, and the benefits of reduced tweeter distortion outweigh it by far. If the tweeter itself has controlled directivity like a wave-guide, this is even more successful because the shift in directivity and polar pattern at crossover frequency is reduced.
What I think is sometimes forgotten by those that eschew higher crossover points for power response reasons, is that unless the drivers are exceptionally close together, any even order crossover is going to introduce a 3dB dip in the power response independent of any loss of power response from the individual driver responses anyway. If it was such a big problem we wouldn't be able to use an even order crossover in most situations.
I'm not sure that a dipole is a good way to compare the effects of dispersion on listening preferences as a dipole is primarily a way of achieving directivity at low frequencies, but in the process is introducing an interfering back wave. I would say this interference from the back wave is the main reason why his listeners didn't like the results.
In the context of a move to constant directivity designs, most such designs don't make any attempt to introduce directivity at low frequencies where a dipole would be directional, and they don't introduce an interfering back wave either.
They focus on achieving a CD response from somewhere in the midrange upwards. I'm not sure that there is any benefit in trying to acheive any more directionality at low frequencies than what you could get from a moderately large baffle. (Eg below ~300Hz)
I think there is definitely something to be said for having a narrower directivity in the vertical plane than the horizontal plane, especially above 2Khz. There is nothing to say that a CD design has to have the same directivity in both axes, and some of the better designs are 90 degrees horizontally but considerably less vertically, on purpose.
I'm a bit of a fan of wave-guide loaded ribbon tweeters, and I think the different directivity in horizontal and vertical planes is a big part of this.
Horizontally many are 90 degree constant directivity over nearly all of their operating range (my AC G2 are) - wide enough to provide some illumination of the side walls, (for spaciousness) and with a relatively flat off axis spectral response, but still narrow enough to avoid baffle edge diffractions and also keep the direct to reflected ratio in the room significantly higher than typical, thus maintaining imaging further out into the room.
In the vertical plane they're a lot more directional with a gradually narrowing beam that reaches about 40 degrees at 10Khz and 20 degrees at 20Khz. Practically speaking you have to go about 20 degrees off the vertical axis before treble noticeably drops, which is a more than wide enough listening window if they're mounted at seated ear level. Meanwhile the bad effects of treble reflections off the ceiling and floor are dramatically reduced.
As you suggest, there is nothing beneficial about ceiling and floor reflections at high frequencies - IMHO they just serve to reduce the direct to reflected ratio and "stretch" the vertical image unnaturally. With reduced floor/ceiling reflections I find I can still get good imaging a lot further into a long room with a typical 2.4 metre ceiling than a more conventional tweeter whose excessive vertical off axis radiation tends to stretch the image into a large "wall of sound" after some distance due to the excessive floor and ceiling reflection. (Primarily ceiling)
If I had to guess at the optimal radiation pattern at high frequencies for home listening I would say 90 degrees constant directivity horizontally and 40-60 degrees "near" constant directivity vertically, with a little bit of deliberate high frequency off axis roll off in the vertical plane only. (More than a pure CD response, but less than what you'd typically get with a vertical ribbon)
Last edited:
I think the floor bounce region of interest is more in the middle hundreds.
I read everything I can find on the correlation between measuring systems in a live room and their perceived frequency response. The most "enlightened" papers point to an effective audible time window that varies with frequency, long at low frequencies and short at high frequency (see Kates and Salmi, for example). In-room curves taken with gating approaches to mimic the hearing process very typically show the floor bounce as a big abberation. The usual measurement shows response errors primarily in the 150 to 500 Hz range. The same papers suggest that in the treble range the ear pretty much just hears direct sound (or any cabinet reflection related issues) only.
That said, I frequently hear comb filtering when doing "deep knee bends" in front of a speaker. This would be treble floor and ceiling reflections. Even carpet doesn't prevent their audibility.
David S.
I believe they were commenting on "dryness" or lack of a realistic image, rather than anything that might be related to the rear wave.
I agree that we need to be careful about any generalizations about a particular pattern. The placement in the room will change the practical effect. A dipole with a large amount of toe in can give a nicely delayed strong lateral reflection. We might like the effect. Yet at the typical slight toe-in the side null might be just in line with the wall reflection point and we get nothing.
David S.
yes, I agree, absolutely
perhaps - but why? because the "WHY?" part of my question remains unaswered
same here but perhaps Markus has some or at least knows of such recordings
I see ! Search here http://www.linkwitzlab.com/AES'07/AES123-final2.pdf for "6 ms" with space in between and "whole words" only.
Is has been found "empicrically" with some reference to "Green" (you might as well search for those words).
If you go way below (e.g. 1 ms) the precedence effect does not operate.
Two things about the paper:
1. It is an excellent re-read in this light.
2. If you can cater for 10 ms, take 10. I think 6 ms is considered the bare minimum.
it is said on the page 13:
no further references as to who, when and how found that empirically
I don't buy it as it contradicts my experience and also experience of other diyaudio users who tried a flooder or a Carlsson-type short omni in case of both of which there is a lot of reflection also < 6 ms yet the quality of spatial reproduction can be described even as godlike as one of those users put it above in this thread
my "why?" remains unaswered
Since those 6-10 ms are the result of experiments with clicks and noise in anechoic chambers (or headphone experiments) it certainly can't contradict your experience ... you simply don't have it. 😛
Figure 2, 5, 6
Fig. 2.5, page 36
If what you have achieved is godlike you certainly don't need advice from "earthlings" any longer. 😉 But if widening the ASW is your goal, then filling the 1.5 - 10 ms gap with reflections is a very effective measure.
yes, and as such these experiments are completely irrelevant from perspective of high fidelity music reproduction at home
yes, or at least - so it seems
but are there any relevant peer-revied 😉 experiments or are You referring just to Your own experience?
and what about <1.5 ms? are they detrimental/beneficial? can they? in what way? under what circumstances?
If you ignore the first principles, how do you feel able to build upon them? 😱
I try to avoid added ASW. I'm into pin-point-image. 😀
Just google for "Apparent source width". Plenty of relevant results on the first two pages ...
Reflections below 1.5 ms might interfere with stereo ITD ...
if You equate "clicks and noises in anechoic chambers" with "musical sounds in small rooms" then how can we meaningfully discuss the matter further? 🙄
suit Yourself but don't try to make it a point of reference for others
😉 good joke
...
Last edited:
I don't equate. If you don't understand Galilei's experiments with the inclined plane, you can't develop a spaceship. Same with "click and noise" experiments and constructing godlike speakers. We need to know the basic experiments to understand where it is safe to ignore them.
we need to have a basic understanding of how our senses work, including sense of hearing
in the light of what we know the analogy You have proposed is false because those situations are different not only quantitatively bu also qualitatively
our hearing sense doesn't only measure, it interprets as well, hearing is not only quantitative process but also qualitative
this paper looks more interesting:
http://www.acoustics.hut.fi/research/cat/psychoac/papers/merimaaaes117.pdf
but they were not using just clicks/noises and just anechoic spaces/headphones
therefore this particular study is more in touch with reality
http://www.acoustics.hut.fi/research/cat/psychoac/papers/merimaaaes117.pdf
but they were not using just clicks/noises and just anechoic spaces/headphones
therefore this particular study is more in touch with reality
Let me elaborate a bit on my comment on floor/ceiling reflections of treble versus the effects at lower midrange frequencies, as I think there are two different psycho acoustic effects going on here due to the very different wavelengths, despite the ultimate cause of both being reflections.
I agree that the floor bounce effect is a big factor in the lower midrange upper bass region, something we've both talked about at length in other threads, and that its effect below the Schroeder frequency of the room, (< ~200-300Hz) is primarily a shift in tonal balance, as the ear starts to become unable to separate the direct and reflected signals, fusing them together into one comb filtered sum.
In the treble though the effect is very different. Treble reflects well off smooth surfaces like a ceiling or bare floor, but because the wavelength is so short the comb filtering is so tightly packed (per 1/3 octave) that we don't hear broadband frequency response anomalies like we do in the lower midrange/upper bass, either at a fixed position or while moving about, and the time delay is usually great enough so that the reflection is windowed out by our ear/brain processing - we hear the direct sound as your reference says.
However I think it would be fair to say that the direct sound only dominates our perception at high frequencies when there is some minimum direct to reflected ratio, and that once you're far enough into the reverberant field of a large fairly reflective room so that the direct sound is less than the reverberant field, this effect goes away and we start to hear the total power response of the room, including reflections from the floor and ceiling, and that the apparent image locations become a fused conglomerate of the various sources. I'm not sure what the actual figures are for this threshold, but the transition is certainly there.
As you get further away from the speakers in a large room with a low ceiling, (one of my previous listening rooms was 8m long by 4m wide with a typical 2.4m ceiling) a number of factors all conspire together to reduce the ability to separate the direct treble and the reflection off the floor and ceiling.
At a "normal" listening distance there is a lot of delay on the ceiling reflection, (less so on the floor reflection) which helps the ear/brain window it out, but this delay reduces a lot as the reflection angle becomes shallower with increased distance. At some distance the delay in ms will be insufficient for the ear/brain to efficiently window out the reflection based solely on time delay.
Then you have the change in direct to reflected ratio with distance. As well as a general loss of direct to reflected ratio with distance due to the reverberant field there is also specifically a large change in direct to floor/ceiling reflection ratio with distance.
When you're closer to the speaker the angle of reflection off the floor and ceiling is rather steep, so the off axis fall-off of the tweeter will help improve the direct to reflected ratio more than it will at a greater distance where the reflection angle is shallow.
The steep angle also means that the two reflections arrive at the ear from greatly different vertical angles than the direct signal. I don't know how much the ear can discern height when trying to separate reflections, but there is probably at least some effect, if not as good as lateral reflections. At a far enough distance the vertical angle the reflections arrive at aren't that different from the direct signal, removing one of the possible ways of differentiating it.
So by the time you get far enough away you (a) don't have much time delay between the direct signal and floor/ceiling reflections, (b) don't have much difference in vertical arrival angle at the ear (c) don't have much off axis attenuation from the tweeter in the reflections, (d) don't have much direct to reflected ratio in general.
All of these seem to conspire to fuse the direct and reflected signals together and create the "wall of sound" effect that I alluded to, where the high frequency image seems to come from the entire speaker wall, with a lot of vertical stretch to it compared to closer listening.
Making the tweeter more directional is obviously going to improve the direct to reflected ratio, thus extending the maximum listening distance before this happens, but too much directivity horizontally is going to reduce the "envelopment".
Vertically displaced reflections don't help envelopment, so if you deliberately make a tweeter that is a lot more directional vertically than horizontally I think you get the best compromise - you still get a worthwhile improvement in direct to reflected ratio in the room, you keep some side wall reflection (and what side wall reflection you do get is close to spectrally balanced if it's CD in the horizontal plane) and you minimize the unwanted floor and ceiling reflection dramatically, even out to moderately far listening distances.
Viewed from further down a long room you have a sound field that is wide horizontally (including the side wall reflections) but remains narrow (almost point source) vertically as the ceiling isn't illuminated much if at all at high frequencies at the necessary angle of incidence, so the vertical image remains tightly focused on the speakers. The practical upshot is that the speakers can extend their "up close" imaging characteristic much deeper into the room, despite the proximity of the floor and ceiling.
On music ? Can you be sure you're not just hearing changes in the vertical off axis response of the speakers, especially at the crossover frequencies ?
Last edited:
- Status
- Not open for further replies.