check out the Beveridge which is in a sense exactly the opposite of this bipolar approach that Markus advocates
Do you think if you promote it enough times it will eventually wear me down to the point where I give in and decide I like it ?
Do you think if you promote it enough times it will eventually wear me down to the point where I give in and decide I like it ?
it is not question of like it or not but of facts
I expect You to take an unbiased look at it in good faith and true spirit of sincere discussion and to do the math - math doesn't lie
am I expecting too much?
do You simply reject facts You don't like?
What facts ? What problem is the Beveridge placement solving ?
You imply that it is eliminating early reflections, but you seem to ignore the reflection off the wall behind the speaker. (Which conveniently doesn't have an arrow in your diagram, as if it somehow doesn't exist) Unless the speaker is flush mounted inside the wall there will still be both diffraction from the cabinet edge and reflection from the wall behind it. Something which has already been pointed out to you in this thread. (By Dave S, I think)
This is no different than pushing a conventionally placed speaker right up against the front-wall of the room and somehow expecting the front wall to no longer exhibit reflections.
Granted that the directivity of the speaker will reduce the wall reflection somewhat at higher frequencies, but for proper function a Beveridge speaker would have to have extremely wide dispersion, thus greatly reducing this mitigating effect.
You then introduce the serious problem that the listener is now about 70 degrees off axis from the speaker. This by necessity is going to mean an extremely wide dispersion design is required, which means narrow baffle and small drivers, which is going to provide heavy illumination of both the front wall and contralateral walls. What for ? Most rooms already have too much illumination, a wide dispersion Beveridge placement is only going to increase that further, thus reducing direct to reflected ratio.
What practical design of Beveridge speaker can you come up with that would have the same performance level as a large 3 way system but is flat enough to hardly or not at all protrude from the side wall ?
Its easy to overlook dynamic performance of speakers when thinking only in terms of polar patterns and directivity.
So in summary, Beveridge placement attempts to solve early wall reflections, but unless it's actually mounted in the wall it doesn't in fact solve this problem. In the process we are now forced to listen 70 degrees off axis from the speakers, and get a far lower direct to reflected ratio. No thank you.
CBT = you could place omni at the floor to create identical circular wavefront. Could even use 'flooder' lol
- Elias
CBT = you could place omni at the floor to create identical circular wavefront. Could even use 'flooder' lol
- Elias
Where can I buy working "omni" or "flodder"?
http://www.xlrtechs.com/dbkeele.com/CBT.php - that's a different animal.
If you're talking about equalizing a room with any variation of electronic EQ, it's not as trivial as it sounds on paper. You can attenuate any resonant areas of frequency successfully (no bad side effects that I know of - although there's the compromise between how a room resonance affects transients vs. steady state acoustic signals). But you can NOT force up any cancellation areas of frequency without creating peaks at other locations in the room, and in extreme cases you can cause premature driver blowout. As far as I know, cancellations, have to be dealt with in the acoustic realm (since they are different for each physical location in the room).
One of the places Linkwitz and I disagree is where he says the weakest link in reproduction, is the speaker. I say it's more often how the speaker interacts with the listening room acoustics.
If the listening room is the right size and several other things about it are good (very minimal reflections less than 6mS or more than 35mS so time response is not a big detriment), then Linkwitz might be right. In the many rooms I've lived in (I'm not as rich so my rooms are a little smaller), I've found the room to be more likely the biggest problem. I noticed that some $300 Radio Shack speakers in a friends apartment sounded substantially better than some $5K high end speakers down at the woopi-doo Hi-Fi store showroom. I believe it is because of how they interacted with the acoustics of the room.
In the bass frequencies (20HZ - 100HZ), the typical problem is boominess, from uneven response, which is usually from room boundary reflections creating cancellations that don't get filled in by other reflections as well as higher frequencies typically do, due to the scarcity of effective reflections in that frequency range, which is due to the size of the listening room. Lower mid frequencies (80HZ - 200HZ), same argument.
100HZ - 800HZ clarity of timing cue info = clarity of embedded imaging cues. Room reflections can create a sense of spaciousness, at the expense of clarity of embedded timing imaging cues. If you're not doing something about inter-aural crosstalk in this frequency range, then do not pass GO and do not collect $200. Just worry about getting your cancellations filled in however. More reflections are probably better.
In the upper mid freqs., anything that creates an amplitude difference over frequency of left vs right signal, will damage imaging clarity. Sidewall reflections may artificially create a sense of spaciousness in both the 100HZ - 800HZ range and here in the 2kHZ - 6kHZ range, but likely at the expense of good imaging of embedded imaging cues. This is because reflection cancellations happening at different frequencies on one side vs. the other, will cause significant amplitude variations over frequency left to right.
Above about 6kHZ, we apparently perceive a sense of space upward. In that range anything that fills in the inevitable cancellations (due to the size of the wavelengths involved) probably sounds the least colored. Lately I'm thinking multiple tweeters. I'm putting a five tweeter array on my center channel speaker right now. There will many cancellations, but because they will all be at different frequencies, due to the physical variations, the end result is that the cancellations will be largely filled in by each other at the listening position.
Unless the speaker is pretty bad, I see the room as being the bigger challenge. Electronic EQ alone won't get you as far as you might initially expect.
Hum, that's a nice post, my 2 cents on it :
>>> EQ is good for large corrections, but when it comes to cure a resonance, it's likely that it's only at the microphone point. A few centimeters further and it's perturbed again and this adds a permanent coloration. Other solutions are required as you say.
>>> I agree totally with the pertinence of the 100/800 Hz zone for a XTC, even of moderate amplitude (-6 dB is already good)
>>> And of course with your conclusion that gives me this idea:
One of these days I have a project of building an other listening room in the "garden". Maybe a square structure of 12 x 15 meters, but with modular walls, e.g. selective reflective panels that can be put in any position or even removed. I see very well a cathedral ceiling, it has to be around 6 meters above the level 0. Level 0 because no floor, only the earth and grass (not very reflective) 2 meters below, the room being on pillars. The speakers have to be suspended and the listener on a kind of tennis umpire chair.
Utopy ? Not really, it's only for me an affair of 3000 $ maximum, less than a few lengths of magic cables, but actually I don't have time for this.
>>> EQ is good for large corrections, but when it comes to cure a resonance, it's likely that it's only at the microphone point. A few centimeters further and it's perturbed again and this adds a permanent coloration. Other solutions are required as you say.
>>> I agree totally with the pertinence of the 100/800 Hz zone for a XTC, even of moderate amplitude (-6 dB is already good)
>>> And of course with your conclusion that gives me this idea:
One of these days I have a project of building an other listening room in the "garden". Maybe a square structure of 12 x 15 meters, but with modular walls, e.g. selective reflective panels that can be put in any position or even removed. I see very well a cathedral ceiling, it has to be around 6 meters above the level 0. Level 0 because no floor, only the earth and grass (not very reflective) 2 meters below, the room being on pillars. The speakers have to be suspended and the listener on a kind of tennis umpire chair.
Utopy ? Not really, it's only for me an affair of 3000 $ maximum, less than a few lengths of magic cables, but actually I don't have time for this.
CBT = you could place omni at the floor to create identical circular wavefront. Could even use 'flooder' lol
- Elias
So, how does one build a pair of Keele's CBT loudspeakers? Do you use a bunch of full range drivers or what? Is there a website about this besides his own:
Don Keele's CBT (Constant Beamwidth Transducer) Page
Hi Humdinger,
Here you said: http://www.diyaudio.com/forums/mult...-pattern-stereo-speakers-153.html#post2736167
And it seems you missed my question following that.
And then, you brought something might be related here: http://www.diyaudio.com/forums/mult...-pattern-stereo-speakers-160.html#post2738221
Would you please share more about how you made your channel distributions?
Thanks a lot,
CLS
Here you said: http://www.diyaudio.com/forums/mult...-pattern-stereo-speakers-153.html#post2736167
And it seems you missed my question following that.
And then, you brought something might be related here: http://www.diyaudio.com/forums/mult...-pattern-stereo-speakers-160.html#post2738221
Would you please share more about how you made your channel distributions?
Thanks a lot,
CLS
and now,
People describe L-XR various ways. X = attenuation factor. So it's Left input minus a certain percentage of the Right input signal, and vice versa. No sweat with four opamps and a dual section pot. When I go to upload an image, diyaudio only gives me the option of an http address. Each input buffer opamp drives a pot, the wiper of which is crosscoupled over to the other channel's output mixer opamp, where it is mixed with the regular signal of that channel, and vice versa. As you rotate the pot, the center image (L+R) gets attenuated, but the ambience (L-R) doesn't. About 2/3 the way along in pot rotation, when the L+R is greatly attenuated but still audible, the sense of stereo collapses. It gets too close to straight L-R. I've been playing with that circuit since the 1970's. Someone probably patented it back then (?). Maybe Jim Fosgate. I never did get any deep info on what he was doing circuit wise.
So I plan to use the L-XR on the front pair and the side pair, independantly. Nothing else will affect those outputs (except for the four tone controls and master output level control on all outputs), which can be run dry into the speakers or run through any reverb. I'll be using the Lexicon MX-400.
The center speaker signal will have steering of about 3dB (a VCA from Analog Devices). Enough to make a difference, but hopefully not enough such that the steering itself is audible (50uS attack time, 50mS decay tenatively). Further, the steering signal driving the VCA will cause more gain when program content is rich in upper mid frequency energy, since that's where a boost in the center speaker will be most helpful, due to how we sense image location. The gain of the center channel circuit will be the difference in L+R, rectified and peak detected, before and after the L-XR circuit that drives the front L&R speakers. Bass below 100HZ goes through none of this. It's just an L+R. Did this answer your question?
Last edited:
Interesting find !
Localisation of sound sources using coincident microphone techniques IOA_Proceedings_Vol29-7
http://eprints.hud.ac.uk/3581/1/IOA_Proceedings_Vol29-7_RS23-2007.pdf
Some quotes of their findings:
"The results obtained for listener detection of phantom images show that the systems tested have not provided the right cues to ensure reliable measurement of apparent source angles. Although listeners appear to be able to distinguish the general direction of a phantom image, the exact match between real recorded angle and its reproduced counterpart has not been demonstrated."
What, phantom imaging does not work ?!
"The accuracy of these techniques in transferring a representation of the real stage into a realistic reproduction of it has not been proven."
Wow, no realism also !?
"None of the various microphone and speaker techniques tested have led to significantly accurate detection of phantom images in an reproduction environment."
No accuracy either !?
".. there does not appear to be any significant benefit of decoding signals for 3 speaker reproduction."
More is not better !?
- Elias
Localisation of sound sources using coincident microphone techniques IOA_Proceedings_Vol29-7
http://eprints.hud.ac.uk/3581/1/IOA_Proceedings_Vol29-7_RS23-2007.pdf
Some quotes of their findings:
"The results obtained for listener detection of phantom images show that the systems tested have not provided the right cues to ensure reliable measurement of apparent source angles. Although listeners appear to be able to distinguish the general direction of a phantom image, the exact match between real recorded angle and its reproduced counterpart has not been demonstrated."
What, phantom imaging does not work ?!
"The accuracy of these techniques in transferring a representation of the real stage into a realistic reproduction of it has not been proven."
Wow, no realism also !?
"None of the various microphone and speaker techniques tested have led to significantly accurate detection of phantom images in an reproduction environment."
No accuracy either !?
".. there does not appear to be any significant benefit of decoding signals for 3 speaker reproduction."
More is not better !?
- Elias
Last edited:
I realized that my above described technique for modulating the amplitude of the center channel has a flaw... Off set and gain settings on the VCA will only be optimal for one input level. the amount of steering will change as the input level changes, because the difference of signals before and after the front L-XR circuit will change as input signal level changes. It's no good.
What I'll do instead is the original idea I had, which is to rectify and peak detect both L+R and L-R independently (from the input signal), and then take the differential of those two signals. So the degree to which L+R is the "center signal" will be determined by comparison to the amplitude of the L-R signal. You can compare the rectified L+R signal to Gnd. (no signal), and that will work to some degree, or you can compare it to anything else that necessarily accurately tracks the amplitude of the L+R signal, such as the L-R rectified signal. I think that will give be the best differentiation for the center channel gain modulation, without making the circuit substantially more complex. I'll also have the option of using an interaural cancellation circuit (a variation of the Bob Carver "Holographic Generator") on the front left and right signals, maybe after the Lexicon MX400 quad digital reverb(?).
Just when you think you've got everything figured out, you think of yet another flea in the ointment... I've got the boards half built, I've got the chassis fabricated, painted and lettered, and I'm still making changes in the circuit... I also added a "side return" stereo input, in case I want to separately delay or put some reverb on the center signal and return it to only the side channels, for better integration into the largely reverberant soundfield.
Criticisms, corrections and comments are encouraged.
What I'll do instead is the original idea I had, which is to rectify and peak detect both L+R and L-R independently (from the input signal), and then take the differential of those two signals. So the degree to which L+R is the "center signal" will be determined by comparison to the amplitude of the L-R signal. You can compare the rectified L+R signal to Gnd. (no signal), and that will work to some degree, or you can compare it to anything else that necessarily accurately tracks the amplitude of the L+R signal, such as the L-R rectified signal. I think that will give be the best differentiation for the center channel gain modulation, without making the circuit substantially more complex. I'll also have the option of using an interaural cancellation circuit (a variation of the Bob Carver "Holographic Generator") on the front left and right signals, maybe after the Lexicon MX400 quad digital reverb(?).
Just when you think you've got everything figured out, you think of yet another flea in the ointment... I've got the boards half built, I've got the chassis fabricated, painted and lettered, and I'm still making changes in the circuit... I also added a "side return" stereo input, in case I want to separately delay or put some reverb on the center signal and return it to only the side channels, for better integration into the largely reverberant soundfield.
Criticisms, corrections and comments are encouraged.
so when scientifically tested under controlled conditions the trinaural-like solutions are not better than conventional stereo?
I am not surprised - 2-channel or 3-channel, no difference - inconsistent and conflicting cues remain
The study Elias does not look to be very professional, IHMO
They didn't get that localisation from XY (amplitude difference only) undergoes two transfomations :
1) incident angle into level differences at the mic outputs, this is the microphone equation which is analytical (assuming perfect mics, not partly crippled solutions as used by the researchers, eg C414's are NOT usable mics for this).
2) level difference into percieved localisation (given from -100% -- full left to +100% full right) with the standard stereo triangle. This is truly empirical measure -- data typically given in tables or graphs, with tolerances between the findings of different researchers -- and it is found that it is frequency dependant, too, there even is a L/R asymmetry in our hearing for most people. (See http://www.sengpielaudio.com/HoerereignRichtungDL.pdf and http://www.sengpielaudio.com/InterchannelLevelDifferencesAndInterchannelTimeDifferences1.pdf). An empirical equation can be given, too, using a polynominal fit.
An online calculator for the same issue can be found here :
Lokalisationskurven Lagrange Interpolation Hörereignisrichtung berechnen Rechner Berechnung Pegeldifferenz Pegelunterschied Laufzeitdifferenz Laufzeitunterschied Interchannel Stereofonie Stereo - dB Umrechnung umrechnen Dämpfung Verstärkung Bezug Bez
Further is a JAVA Applet which calculates and visualizes the apparent source direction based on 1) and 2) (the latter by using the data from Sengpiel), for MS dial in two cardiods at +-90deg (which is equivalent to MS with an omni-M and fig8-S), then widen orchestra angle to +-90deg and you'll see the "angle distortion".
http://www.sengpielaudio.com/Bosem.jar
A similar calculator web-based and lower resolution :
XY Niere 90° Mikrofonanordnung - Visualisation des Stereosystems - Aufnahmebereich Pegeldifferenz Richtcharakteristik Acht - sengpielaudio Sengpiel Berlin
(change "Angle betw. mics" to 180deg, orchestra to +-90deg)
So that is really old and know stuff, modern well educated sound engineers use more modern/flexible microphone techniques to get the localization right.
They didn't get that localisation from XY (amplitude difference only) undergoes two transfomations :
1) incident angle into level differences at the mic outputs, this is the microphone equation which is analytical (assuming perfect mics, not partly crippled solutions as used by the researchers, eg C414's are NOT usable mics for this).
2) level difference into percieved localisation (given from -100% -- full left to +100% full right) with the standard stereo triangle. This is truly empirical measure -- data typically given in tables or graphs, with tolerances between the findings of different researchers -- and it is found that it is frequency dependant, too, there even is a L/R asymmetry in our hearing for most people. (See http://www.sengpielaudio.com/HoerereignRichtungDL.pdf and http://www.sengpielaudio.com/InterchannelLevelDifferencesAndInterchannelTimeDifferences1.pdf). An empirical equation can be given, too, using a polynominal fit.
An online calculator for the same issue can be found here :
Lokalisationskurven Lagrange Interpolation Hörereignisrichtung berechnen Rechner Berechnung Pegeldifferenz Pegelunterschied Laufzeitdifferenz Laufzeitunterschied Interchannel Stereofonie Stereo - dB Umrechnung umrechnen Dämpfung Verstärkung Bezug Bez
Further is a JAVA Applet which calculates and visualizes the apparent source direction based on 1) and 2) (the latter by using the data from Sengpiel), for MS dial in two cardiods at +-90deg (which is equivalent to MS with an omni-M and fig8-S), then widen orchestra angle to +-90deg and you'll see the "angle distortion".
http://www.sengpielaudio.com/Bosem.jar
A similar calculator web-based and lower resolution :
XY Niere 90° Mikrofonanordnung - Visualisation des Stereosystems - Aufnahmebereich Pegeldifferenz Richtcharakteristik Acht - sengpielaudio Sengpiel Berlin
(change "Angle betw. mics" to 180deg, orchestra to +-90deg)
So that is really old and know stuff, modern well educated sound engineers use more modern/flexible microphone techniques to get the localization right.
One thing a center speaker can do is stabilize the center image for horizontally off axis listening. Another thing it can do is help clarify the solidity of the "phasey" sounding center image when an inter-aural cancellation circuit is engaged (which I find works pretty well, especially with head mic recordings). Since it's physically displaced from either side speaker, it can also contribute to the filling in of comb filter cancellations from room reflections. With room reflections as they are, creating comb filter cancellations and imaging confusion, there's only so much you can do from the point of view of "fidelity". Once at a point of maximum technical "fidelity", I am glad to embrace whatever make it sound "better", real or not. I may be wrong, but I feel driven to experiment/explore with my analog surround extractor circuit described above, and the option of the Lexicon reverb algorithms. For music, I'm thinking it will be nice.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.