So your conclusion is, after extensive tests of your own 'bending wave' speaker, an array of multible small drivers is better ?
I suppose there may be some truth. Array is at least more DIY friendly and thus more interesting generally.
The whole thing comes to finding optimum array design.
How about 2 dimensional Golomb ruler ?
(lets pretend such exists in theory)Elias can you say something about the listening room used, when Toole
came up with these recommendations ?
Let me guess: It was not a usual living room, but a room with side
reflections somewhat diffused.
If i am guessing right, we cannot apply the recommendations to
any usual living room conditions without at least some restrictions and
refinements.
came up with these recommendations ?
Let me guess: It was not a usual living room, but a room with side
reflections somewhat diffused.
If i am guessing right, we cannot apply the recommendations to
any usual living room conditions without at least some restrictions and
refinements.
I was talking about "small disc shaped" bending wave designs ... my
current design is a multipanel design with about 0.5 squaremeters
of area.
"Modal balancing" like used in BMR is not needed in my particular
design, because modal density is by >2 orders of magnitude higher
than in those toy designs we were talking here. Thus the frequency
response is much smoother even in the Khz region.
But if you want a balanced frequency response on axis plus decorrelated
radiation to larger off axis angles without spending the effort in making
an "acoustically larger" bending wave design including all the
weird technical challenges , it seems to yield better results working
with arrays.
I can hardly see any benefit in "toy sized" bending wave designs.
What sounds different in (some) bending wave designs is the kind of
radiation and room interaction.
It is not the transducer working according to a different principle, that
causes "different sonic impression".
The BMR I tested is far beyond the usual coaxial ceiling speakers one can hear in stores in the subjective far-off-axis sound quality.
You are right, if done well BMRs can deliver more flat power response
and wider dispersion of the uppermost octaves especially in comparison
to mid and larger sized conventional "fullrangers".
It is an attractive and modern way of making a fullranger, especially in
cases where a single driver is preferred to do the job (e.g. for place- or
cost reasons).
I was more referring to the reqirements
- "wide dispersion" and
- "decorrelation above 2Khz" and
- "smooth on axis response"
where i find the concept not very attractive, especially when using
dedicated woofers.
But even in a more conservative FAST concept aiming at coherent
radiation, a smallish "old school optimized" widebander of high quality
is hard to beat using a BMR.
Nevertheless i can imagine them replacing conventional fullrangers
in many typical applications, like e.g. said ceiling speakers in
supermarkets or even line arrays in churches etc. ...
Last edited:
Just received a HIBM65C20F-4 and did some quick and dirty measurements. Did you also get that huge spike around 12kHz off-axis?
Sorry, I don't remember and didn't store the measurements.
Here are the measurements of the current version of my own driver:
http://www.diyaudio.com/forums/full-range/238089-building-ones-own-diaphragm.html
Stereo triangle smears the time coherence
Stereo triangle adds something we dont want: comb filtering.
I do not always adhere to the stereo triangle setup and prefer to have them oriented on the wall flats instead. For the life of me cannot remember what this alternative setup is called but have been using it since the mid '80's.
The speakers I'm working on will be setup in this arrangement. If I want the tightest imaging switching is not an issue, but is rather confining IMHO for normal listening. Most of the living rooms I deal with (including mine) are not symetrical with respect to where the furniture (seating) is placed. I've found many prefer this over the norm as the difference can be somewhat profound. Fill the room with sound vs X marks the spot. An exceptionally broad dispersion / even polar is an absolute necessity. Controlling how wide in the upper octaves above 2kHz is not difficult to pull off to achieve best results.
This comb filtering is not really an issue. The last speakers I built for personal use (~22 years ago) were a short line array MMTMMM-WW setup. Drivers were time aligned with gradient shading (if that's what they want to call it now, OK). Listening distance was to be 3-4m for the time alignment to be maintained. Comb filtering was never an issue and on/off axis was impeccable. Stellar imagining IMHO.
Could be interestring observations for those who search in 3 channel direction:
Moulton Laboratories :: The Pitfalls and Promises of the Center Channel
and second part
Moulton Laboratories :: A Happy Accident: A Better Way to Play Back Stereo?
Moulton Laboratories :: The Pitfalls and Promises of the Center Channel
and second part
Moulton Laboratories :: A Happy Accident: A Better Way to Play Back Stereo?
After reading those it looks like, unfortunately, that Moulton considers the matrix as only a summation of the center channel. He does not seem to enhance the side speaker signals. Thus he is not experiencing the true finesse of matrix reproduction. A pity, actually.
The best 3 speaker matrixes are having the form:
Ls = a*L - b*R
Cs = c*(L + R)
Rs = a*R - b*L
where Ls, Cs and Rs are the speaker signals after the matrix, a, b and c are coefficients (which may or may not be frequency dependent), and L and R are the original stereo signals.
The most important term after the center channel summation is the difference term in the side speaker signals.
A lot has been written in literature about the 'optimal matrixes' for stereo reproduction during the past decades. I have also done my share
Elias Pekonen Home Page - New Optimal Linear Stereo Matrix for 3 Speakers
.
He does, however, recognize the necessity of increasing the angle to the L and R speakers, since the matrix (any matrix) moves apparent L and R to virtual images between the "new" center and the actual L and R sources.
pretty much nails it, but to which I would add two additional consideration:
1) the ceiling (and to a lesser extent, the floor). The vertical radiation pattern of multi-driver loudspeakers is too often ignored, with the result being that the ceiling reflection is often wildly colored as frequency dependent lobes shift across the first (specular) reflection point. It is a “room dependency” that can cause the same speaker to sound quite different in rooms of differing ceiling height or at differing listening distances.
2) the “center channel”, which has at least the potential of resolving many of the issues that arise from having center stage performers presented as ill defined “phantom images”. The center does not have to be provided as an independent “third channel” in the delivery system (although that would be best), but can often be derived (matrix) from existing two channel recordings.
“Surround” channels are unnecessary for music reproduction . . . there is already more than enough “reverberant field” in the typical listening room, and that, when combined with hall reverberation captured in the recording, is enough to give an adequate sense of “space”. Additional rear sound sources allow interesting “effects” in movie presentations, but those effects are just gimmicks which seldom enhance “realism” when reproducing music.
To me it's a very different perception when enveloping sounds from the back are missing.
Moulton moved the side speakers in wider angle because his matrix was not optimal. Matrix stereo is not a pair wise panning between two speakers, no. All the speakers contribute to the imaging. You may need to draw some vectors to realise that
Start with a matrix with cross feed coefficient of 0.5, for example.- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.