"The Clue" - ny amerikansk högtalare

[celef]

...

...

thank you for the post, i need to re-read it a few times. i have always thought about the loudspeaker as a mono-source, then interacting with the room or parts of the room, the interaction in a stereo setup is all new to me, i sure would have liked to discuss this in swedish

Celef,

I'm sorry that I can't deliver the information in Swedish. Maybe someone in the forum can help us with translations.

I think it is important in two-channel loudspeaker design to always consider both loudspeakers to be an integrated system.

In fact, one must consider the pair of loudspeakers, the environment/room, and the listener all as an integrated, inter-dependent system. The loudspeakers are coupled locally (mutual coupling) and also coupled globally (listener and boundaries).

(In a perfect world, the program source and method of recording would be included, but not in today's discussion.)

Mono-source evaluation of a loudspeaker is certainly useful, but it is only the starting point, not a complete characterization.

There are more issues than I can cover in this post, but, as one example, even the interference and resulting frequency response ripple due to cross-talk of the two channels at the listener's head must be considered when balancing the loudspeaker 'system', and this can only be observed when both loudspeakers operating at the same time, and interaction at the listener's body, is considered.

With two-channel reproduction, there is far greater complexity of all the interactive effects, but also more possibilities to improve, not only the spatial aspects, but also the tonal effects.

Then, of course, there is another set of issues relative to loudspeaker design for more than two channels...

Cheers,

- James

how is this done, if i look at it at different angles i get different results for each angle!? how important is this "compensation" for the average listener? do we all have different sensitivities for errors in this regards?

[Jacro]

...

...

thank you for the post, i need to re-read it a few times. i have always thought about the loudspeaker as a mono-source, then interacting with the room or parts of the room, the interaction in a stereo setup is all new to me, i sure would have liked to discuss this in swedish

Celef,

I'm sorry that I can't deliver the information in Swedish. Maybe someone in the forum can help us with translations.

I think it is important in two-channel loudspeaker design to always consider both loudspeakers to be an integrated system.

In fact, one must consider the pair of loudspeakers, the environment/room, and the listener all as an integrated, inter-dependent system. The loudspeakers are coupled locally (mutual coupling) and also coupled globally (listener and boundaries).

(In a perfect world, the program source and method of recording would be included, but not in today's discussion.)

Mono-source evaluation of a loudspeaker is certainly useful, but it is only the starting point, not a complete characterization.

There are more issues than I can cover in this post, but, as one example, even the interference and resulting frequency response ripple due to cross-talk of the two channels at the listener's head must be considered when balancing the loudspeaker 'system', and this can only be observed when both loudspeakers operating at the same time, and interaction at the listener's body, is considered.

With two-channel reproduction, there is far greater complexity of all the interactive effects, but also more possibilities to improve, not only the spatial aspects, but also the tonal effects.

Then, of course, there is another set of issues relative to loudspeaker design for more than two channels...

Cheers,

- James

how is this done, if i look at it at different angles i get different results for each angle!? how important is this "compensation" for the average listener? do we all have different sensitivities for errors in this regards?

Celef,

I've put your questions below with answers after...


HOW IS THIS DONE?

There are a number of techniques.

One way is to simply calculate the differential in arrival distance/timing between the left and right channel to one ear. The half-wavelength (and each odd half wavelength) frequency the length differential corresponds to determines the cancellation or dip frequencies from which one can plot the ripple in the response.

Another way is to measure the differential error with a dummy head and in-ear microphones.

Another step is to compare the sound with and without a crosstalk cancelation device, using the crosstalk-cancelled mode as a reference for tonal balance.

Upon establishing the error, it is not necessary to correct all of the resulting measurable ripple. Instead, an gentle spectral correction, on a ½ to 1/3-octave basis, tends to be adequate for this type of error in that the first octave of error has a dip and peak that are a half octave apart. Above the first octave of onset, the comb filtering of the response is closely spaced and has high enough density that it doesn’t require further correction beyond the first octave, or so, of error.

The correction is a delicate one, in that if implemented inappropriately, it can create coloration in the mid-band.

As usual, this is an over simplification, and more is required to deal with the issue effectively, but hopefully this gives you the general idea.



IF I LOOK AT IT AT DIFFERENT ANGLES I GET DIFFERENT RESULTS FOR EACH ANGLE?

Yes, you are correct. The effect is different at each different listening angle.

That is one of the reasons that one must define the listening angle in relation to the loudspeaker pair very precisely.

Most loudspeaker companies only provide general positioning recommendations, hoping to satisfy a wide variety of different placement preferences of a large number of customers. Necessary for high unit volume products.

As an example of an alternative approach, with ( the clue ), we recommend a very limited use model, with precise placement, and listening angle requirements.

While this is not practical for everyone, due to the restricted use model, it allows us to better optimize the performance of the loudspeaker pair in a way that wouldn’t be possible if we recommended generalized placement.



HOW IMPORTANT IS THIS "COMPENSATION" FOR THE AVERAGE LISTENER?

I’m not sure what you mean by the ‘average listener.’

In terms of amplitude errors related to crosstalk cancellation, they tend to fall in the frequency range that the ear is most sensitive, and are also above just noticeable detection thresholds for amplitude errors, so most any listener with healthy hearing, will detect the difference.

But, your question is valid, in that even though most can discern the difference, will it be important to them?

I would expect discerning listeners to find each change of a similar magnitude to this one to be important, but I guess you would have to decide that one for yourself.



DO WE ALL HAVE DIFFERENT SENSITIVITIES FOR ERRORS IN THIS REGARD?

Interesting question.

I find that each listener comes to a listening session with a personal set of system attributes that they tend to prioritize and focus on in their listening experience and judgment.

BUT, I also find that often in a casual, uncontrolled listening session, listeners will often miss subtle system errors, but if I put that same person in a controlled, blind listening situation, I can almost always teach them to recognize the sonic error, such that after the training, they can then notice the same error in the general listening session.

Again, in the cases of crosstalk cancellation amplitude errors, they appear in the portion of the midrange that the ear is most sensitive to so they will tend to be audible to most listeners.

I hope this all makes sense.

All the best,

- James

[celef]

Celef,

I've put your questions below with answers after...


HOW IS THIS DONE?

There are a number of techniques.

One way is to simply calculate the differential in arrival distance/timing between the left and right channel to one ear. The half-wavelength (and each odd half wavelength) frequency the length differential corresponds to determines the cancellation or dip frequencies from which one can plot the ripple in the response.

Another way is to measure the differential error with a dummy head and in-ear microphones.

Another step is to compare the sound with and without a crosstalk cancelation device, using the crosstalk-cancelled mode as a reference for tonal balance.

Upon establishing the error, it is not necessary to correct all of the resulting measurable ripple. Instead, an gentle spectral correction, on a ½ to 1/3-octave basis, tends to be adequate for this type of error in that the first octave of error has a dip and peak that are a half octave apart. Above the first octave of onset, the comb filtering of the response is closely spaced and has high enough density that it doesn’t require further correction beyond the first octave, or so, of error.

The correction is a delicate one, in that if implemented inappropriately, it can create coloration in the mid-band.

As usual, this is an over simplification, and more is required to deal with the issue effectively, but hopefully this gives you the general idea.

i'm not sure if i understand this, but i sounds close to recent topics about the "bbc-dip" and "fundamental flaw in stereo" we have had at this forum, i couldn't figure out then how this correction should look like, and how to achieve a frequency response that changed dramatically att smal offaxis angles

IF I LOOK AT IT AT DIFFERENT ANGLES I GET DIFFERENT RESULTS FOR EACH ANGLE?

Yes, you are correct. The effect is different at each different listening angle.

That is one of the reasons that one must define the listening angle in relation to the loudspeaker pair very precisely.

Most loudspeaker companies only provide general positioning recommendations, hoping to satisfy a wide variety of different placement preferences of a large number of customers. Necessary for high unit volume products.

As an example of an alternative approach, with ( the clue ), we recommend a very limited use model, with precise placement, and listening angle requirements.

While this is not practical for everyone, due to the restricted use model, it allows us to better optimize the performance of the loudspeaker pair in a way that wouldn’t be possible if we recommended generalized placement.

this is very interesting, how much of the room should be treated as part of the loudspeaker and when starts the room to be just the room? for my own speakers i sometimes use a simple model by roy allison to compensate for boundary gain, even though it calculates this in a simplistic way the compensation often sounds overcompensated

HOW IMPORTANT IS THIS "COMPENSATION" FOR THE AVERAGE LISTENER?

I’m not sure what you mean by the ‘average listener.’

In terms of amplitude errors related to crosstalk cancellation, they tend to fall in the frequency range that the ear is most sensitive, and are also above just noticeable detection thresholds for amplitude errors, so most any listener with healthy hearing, will detect the difference.

But, your question is valid, in that even though most can discern the difference, will it be important to them?

I would expect discerning listeners to find each change of a similar magnitude to this one to be important, but I guess you would have to decide that one for yourself.



DO WE ALL HAVE DIFFERENT SENSITIVITIES FOR ERRORS IN THIS REGARD?

Interesting question.

I find that each listener comes to a listening session with a personal set of system attributes that they tend to prioritize and focus on in their listening experience and judgment.

BUT, I also find that often in a casual, uncontrolled listening session, listeners will often miss subtle system errors, but if I put that same person in a controlled, blind listening situation, I can almost always teach them to recognize the sonic error, such that after the training, they can then notice the same error in the general listening session.

Again, in the cases of crosstalk cancellation amplitude errors, they appear in the portion of the midrange that the ear is most sensitive to so they will tend to be audible to most listeners.

I hope this all makes sense.

All the best,

- James

when looking at measurements of varios loudspeakers at sources like stereophile or soundstage there is hard to detect if they have been compensated for any errors, or at least i have not find any, is this due to the measurement technic they are using or is this compensation rare?

[Jacro]

Celef,

I've put your questions below with answers after...


HOW IS THIS DONE?

There are a number of techniques.

One way is to simply calculate the differential in arrival distance/timing between the left and right channel to one ear. The half-wavelength (and each odd half wavelength) frequency the length differential corresponds to determines the cancellation or dip frequencies from which one can plot the ripple in the response.

Another way is to measure the differential error with a dummy head and in-ear microphones.

Another step is to compare the sound with and without a crosstalk cancelation device, using the crosstalk-cancelled mode as a reference for tonal balance.

Upon establishing the error, it is not necessary to correct all of the resulting measurable ripple. Instead, an gentle spectral correction, on a ½ to 1/3-octave basis, tends to be adequate for this type of error in that the first octave of error has a dip and peak that are a half octave apart. Above the first octave of onset, the comb filtering of the response is closely spaced and has high enough density that it doesn’t require further correction beyond the first octave, or so, of error.

The correction is a delicate one, in that if implemented inappropriately, it can create coloration in the mid-band.

As usual, this is an over simplification, and more is required to deal with the issue effectively, but hopefully this gives you the general idea.

i'm not sure if i understand this, but i sounds close to recent topics about the "bbc-dip" and "fundamental flaw in stereo" we have had at this forum, i couldn't figure out then how this correction should look like, and how to achieve a frequency response that changed dramatically att smal offaxis angles

IF I LOOK AT IT AT DIFFERENT ANGLES I GET DIFFERENT RESULTS FOR EACH ANGLE?

Yes, you are correct. The effect is different at each different listening angle.

That is one of the reasons that one must define the listening angle in relation to the loudspeaker pair very precisely.

Most loudspeaker companies only provide general positioning recommendations, hoping to satisfy a wide variety of different placement preferences of a large number of customers. Necessary for high unit volume products.

As an example of an alternative approach, with ( the clue ), we recommend a very limited use model, with precise placement, and listening angle requirements.

While this is not practical for everyone, due to the restricted use model, it allows us to better optimize the performance of the loudspeaker pair in a way that wouldn’t be possible if we recommended generalized placement.

this is very interesting, how much of the room should be treated as part of the loudspeaker and when starts the room to be just the room? for my own speakers i sometimes use a simple model by roy allison to compensate for boundary gain, even though it calculates this in a simplistic way the compensation often sounds overcompensated

HOW IMPORTANT IS THIS "COMPENSATION" FOR THE AVERAGE LISTENER?

I’m not sure what you mean by the ‘average listener.’

In terms of amplitude errors related to crosstalk cancellation, they tend to fall in the frequency range that the ear is most sensitive, and are also above just noticeable detection thresholds for amplitude errors, so most any listener with healthy hearing, will detect the difference.

But, your question is valid, in that even though most can discern the difference, will it be important to them?

I would expect discerning listeners to find each change of a similar magnitude to this one to be important, but I guess you would have to decide that one for yourself.



DO WE ALL HAVE DIFFERENT SENSITIVITIES FOR ERRORS IN THIS REGARD?

Interesting question.

I find that each listener comes to a listening session with a personal set of system attributes that they tend to prioritize and focus on in their listening experience and judgment.

BUT, I also find that often in a casual, uncontrolled listening session, listeners will often miss subtle system errors, but if I put that same person in a controlled, blind listening situation, I can almost always teach them to recognize the sonic error, such that after the training, they can then notice the same error in the general listening session.

Again, in the cases of crosstalk cancellation amplitude errors, they appear in the portion of the midrange that the ear is most sensitive to so they will tend to be audible to most listeners.

I hope this all makes sense.

All the best,

- James

when looking at measurements of varios loudspeakers at sources like stereophile or soundstage there is hard to detect if they have been compensated for any errors, or at least i have not find any, is this due to the measurement technic they are using or is this compensation rare?

Celef,

Answer to question #1:
I think that the BBC, or “Gundree”, dip is invoked to address audible coloration and inadvertently addresses the crosstalk cancellation errors. If one doesn’t analyze the source of the problem, but instead attempts to correct what is experienced as a tonal error, then one might apply a BBC type dip as a partial fix and hear a sonic preference in the result.

But, if one shapes the response correction to more closely follow the shape of the crosstalk error (not simply a dip), the tonal correction will sound more “life-like”. The BBC dip takes away the “peak” portion of the error, and softens the coloration, but also takes a bit of the “aliveness” out of the midrange.
All of this is fairly subtle, but significant in the totality of the sonic presentation.


Answer to question #2:
If I understand what you are asking, I think of the boundaries that have reflective path lengths within 10 milliseconds, and greater than -10 dB of the direct sound, to be “the loudspeaker”, and the rest to be “the room”.
***
Allison’s work was significant, in that he got people thinking about the importance of considering the effects of the first three boundaries, but it is amazing to me how few loudspeaker companies incorporate Allison’s notions into their loudspeaker design. Roy was also surprised that he had such a small influence on loudspeaker design in the industry.

I’m working on a new paper that I hope to present at AES next year that honors Allison, but shows that his basic thesis that acoustic impedance of the boundary and the change in acoustic impedance due to the distance to each boundary from acoustic matching function dominated the spectral errors. I have found that the issue is not dominated by the acoustical impedance of the boundary, but instead is based on a “specular” reflection differential relationship to the direct sound and this new way of viewing the problem provides a more accurate analysis and correction factor.



Answer to question #3:
I think this type of compensation is rare.

I am not aware of any other loudspeaker manufacturer that claims to use compensation for crosstalk amplitude errors in their loudspeakers. Some may, but I am not aware of it.

As stated above, the most similar approach may be the use of a simple BBC type dip.

Without establishing a very specific listening angle between the loudspeaker and the listener, this approach is difficult to calibrate for correct results and most loudspeaker companies don’t provide strict and specific angular use instructions.

Also, I’m not aware that very many companies design their loudspeakers with any specific design elements that optimize for two-channel usage. Most test and optimize in monophonic mode. Surprisingly, even Floyd Toole and Sean Olive’s “advanced” Harman International loudspeaker test and evaluation center uses only single loudspeaker, monophonic testing and comparative listening sessions. I’ve discussed this with them and, in private, they admit it is a shortcoming that they don’t have a solution for.

For the correction is best to use what I call a “soft correction”, which is a gentle alteration to the frequency response, which may not be easy to interpret from standard measurements. Many loudspeakers have response ripple errors that are as large as the correction, so, if they were applying this type of correction, it could be hard to see the correction hidden within their response errors in the measurements that you have observed.

All the best,

- James

[paa]

Answer to question #1:
I think that the BBC, or “Gundree”, dip is invoked to address audible coloration and inadvertently addresses the crosstalk cancellation errors. If one doesn’t analyze the source of the problem, but instead attempts to correct what is experienced as a tonal error, then one might apply a BBC type dip as a partial fix and hear a sonic preference in the result.

But, if one shapes the response correction to more closely follow the shape of the crosstalk error (not simply a dip), the tonal correction will sound more “life-like”. The BBC dip takes away the “peak” portion of the error, and softens the coloration, but also takes a bit of the “aliveness” out of the midrange.
All of this is fairly subtle, but significant in the totality of the sonic presentation.
All the best,
- James

Interesting, especially the answer to #2 about your upcoming paper.
Anyway, here is previous discussion about the BBC dip:
http://www.faktiskt.se/modules.php?name ... &p=1064224

[single_malt]

Anyway, here is previous discussion about the BBC dip:
http://www.faktiskt.se/modules.php?name ... &p=1064224

Ja, och där framgår den rimligaste förklaringen:

13.I've heard mention of 'the BBC dip' or 'the Gundry dip'. What does that mean?

There is much myth, folklore and misunderstanding about this subject.

The 'BBC dip' is (was) a shallow shelf-down in the acoustic output of some BBC-designed speaker system of the 1960s-1980s in the 1kHz to 4kHz region. The LS3/5a does not have this effect, neither in the 15 ohm nor 11 ohm, both of which are in fact slightly lifted in that region.

According to Harbeth's founder, who worked at the BBC during the time that this psychoacoustic effect was being explored, the primary benefit this little dip gave was in masking of defects in the early plastic cone drive units available in the 1960's. A spin-off benefit was that it appeared to move the sound stage backwards away from the studio manager who was sitting rather closer to the speakers in the cramped control room than he would ideally wish for. (See also Designer's Notebook Chapter 7). The depth of this depression was set by 'over-equalisation' in the crossover by about 3dB or so, which is an extreme amount for general home listening. We have never applied this selective dip but have taken care to carefully contour the response right across the frequency spectrum for a correctly balanced sound. Although as numbers, 1kHz and 4kHz sound almost adjacent in an audio spectrum of 20Hz to 20kHz, the way we perceive energy changes at 1kHz or 4kHz has a very different psychoacoustic effect: lifting the 1kHz region adds presence (this is used to good effect in the LS3/5a) to the sound, but the 4kHz region adds 'bite' - a cutting incisiveness which if over-done is very unpleasant and irritating.

You can explore this effect for yourselves by routing your audio signal through a graphic equaliser and applying a mild cut in the approx. 1kHz to 4kHz region and a gradual return to flat either side of that.


http://www.harbeth.co.uk/faq/index.php#13

Enligt detta var BBC-dipen ursprungligen en åtgärd för att hantera egenskaper hos de högtalarelement som användes. Givetvis har den trots det en psykoakustisk inverkan.

[Objektivisten]

Hi James, welcome to the forum, what do You think is the secret about tubes and that they almost always sounding more pleasant and true to the ears? And do we really need all that watts, low powered amps seems to gain in transparent sound, may be construction simplicity or cost effectiveness? Why do I think carbon sounds more real than metal as a conductor? Any clue?

[Jacro]

Answer to question #1:
I think that the BBC, or “Gundry”, dip is invoked to address audible coloration and inadvertently addresses the crosstalk cancellation errors. If one doesn’t analyze the source of the problem, but instead attempts to correct what is experienced as a tonal error, then one might apply a BBC type dip as a partial fix and hear a sonic preference in the result.

But, if one shapes the response correction to more closely follow the shape of the crosstalk error (not simply a dip), the tonal correction will sound more “life-like”. The BBC dip takes away the “peak” portion of the error, and softens the coloration, but also takes a bit of the “aliveness” out of the midrange.
All of this is fairly subtle, but significant in the totality of the sonic presentation.
All the best,
- James

Interesting, especially the answer to #2 about your upcoming paper.
Anyway, here is previous discussion about the BBC dip:
http://www.faktiskt.se/modules.php?name ... &p=1064224

Thank you for the link to your BBC dip discussions. It seems that almost every topic in audio has been discussed in your forum.
I always find enjoyable reading here.

I had some great discussions with Alan Shaw at Harbeth a few years ago. He is a wealth of information regarding the truth about the LS3/5a and BBC loudspeaker research.

It would seem that the LS3/5a is a combination of accuracy and sonic trickery. Odd for a studio monitor to be optimized for sensationalism rather than neutrality. It is a wonderfully entertaining and fun loudspeaker to listen to, but I’m not sure it qualifies as a neutral, professional monitor.

I’ve owned 8 different pairs of LS 3/5a, 15 ohm, 11 ohm, Rogers and Spendors, and still have a couple pair that I enjoy. I had to do a very extensive analysis of the LS 3/5a when I developed the Satterberg Mid-Woofer for them back in the 1980’s. It was very difficult to get a seamless match through the crossover and retain all the sonic attributes of original loudspeaker when used with the Satterbergs.
It was a fun and challenging project, in that we had to develop an unusual approach to subwoofer matching to get it to work properly.

In terms of the BBC/Gundry dip, while it was originally developed to minimize the colorations in the early generations of Bextrene cones, it has been used for many other purposes.

If I remember correctly, I believe the Spendor BC1 and BBC LS3/6 incorporated a Gundry type dip in the upper midrange.

These days this reduction of energy in the upper midrange is most often used to minimize tweeter “flare” or excess off-axis tweeter energy at the crossover frequency.

One of the things that also must be considered, when applying BBC dip type response changes in the midrange, is that they don’t just alter the perception of the tonal character, but they also impact imaging, as the amplitude changes correspond to changes in source angle, as observed in head related transfer functions.

Cheers,

- James

[LypsylateX]

This is one of the best threads in a long, long time. Thank you, all.

(And I just made it a little worse. )

[celef]

james,

thanks for your post, i need some time to think it throu. i have much to ask but one thing comes to mind; this might be against your policy but, what do you think about technologies like www.embracingsound.com

best regards

[DanNorman]

Detta är alltså den raka motsatsen till pip (och guru qm10)?
Ingvar hade problem att hitta en tillräckligt högspridande diskant till pip, medan här väljer man att stoppa den i en waveguide. Intressant.

[DanNorman]

Men givetvis liknande ideer om att integrera högtalaren med rummet.

[sprudel]

Det här är ju jätteintressant!


Citat:

"Although as numbers, 1kHz and 4kHz sound almost adjacent in an audio spectrum of 20Hz to 20kHz, the way we perceive energy changes at 1kHz or 4kHz has a very different psychoacoustic effect: lifting the 1kHz region adds presence (this is used to good effect in the LS3/5a) to the sound, but the 4kHz region adds 'bite' - a cutting incisiveness which if over-done is very unpleasant and irritating."

Så det som jag brukar kalla för överdiven kontrast, eller en analogi till en LCD-skärm med för kraftig skärpeinställning kan vara en höjning i 4kHz området? Vidare skulle en höjning av 1kHz ge en ökad närvarokänsla.

Rörsteg??????

Citat: "A spin-off benefit was that it appeared to move the sound stage backwards away from the studio manager who was sitting rather closer to the speakers in the cramped control room than he would ideally wish for."

Den falska förskjutningen av ljudbilden bakåt! Rymden i ljudbilden?

QLN Signature!

Rörsteg?????

Eller kombinationen av vissa steg med vissa högtalare som ger ett konstlat djup i ljudbilden. Är det så enkelt?

[DanNorman]

Lek med en eq! Det något många skulle lära sig mycket av. Bland annat hur kort ljudminnet är
6db hit och dit.. man vänjer sig snabbt.. sen kan man tro att man hör skillnad på ..... och ........, speciellt utan att blindtesta

[skrutten]

Detta är alltså den raka motsatsen till pip (och guru qm10)?
Ingvar hade problem att hitta en tillräckligt högspridande diskant till pip, medan här väljer man att stoppa den i en waveguide. Intressant.

Kanske att wave guiden ifråga sprider ljudet annorlunda än förväntat?
Bara en tanke... Hursomhelst en intressant tråd detta.

/G

[DanNorman]

Ja, visst. Jag har ingen aning. Bara en tanke.

[Jacro]

Hi James, welcome to the forum, what do You think is the secret about tubes and that they almost always sounding more pleasant and true to the ears? And do we really need all that watts, low powered amps seems to gain in transparent sound, may be construction simplicity or cost effectiveness? Why do I think carbon sounds more real than metal as a conductor? Any clue?

Objektivisten;

First, thank you for the friendly welcome. Everyone has been very friendly.

These are big subjects, and often difficult to explore effectively in a short post, but I will start the discussion.

Lately, I tend to avoid answering these types of questions in that I find that many people have already made up their minds and aren’t interested in hearing an answer that contradicts their beliefs.

I have been involved in extensive research in this area, both independently in the early 1980’s and again in the 1990’s when I ran the R&D department at Carver Corporation.

In this post I’ll start with some well-verified, general principles, and if there is further interest I can go into deeper specifics.

One thing that is important to understood is that to find truths about such matters is a complex process and not at all fun if done properly.
Hard work.

It takes a lot more effort and rigor than most audiophiles are willing to endure. For many, audio is a joyful endeavor, and it should be.

It is fun to do casual listening tests and derive and proclaim judgments about many aspects of audio, in terms of what is good, what is bad, all from a casual environment without establishing careful controls and repeating hundreds of tests, over and over again to be sure that one is not fooling themselves.

It is not easy or fun, and I find that there are not many that are willing to put forth the great effort it takes and be open to challenging ones most cherished beliefs.

To me listening to music through an audio system is one of the most fun hobbies one can have. Doing the research to find out which things really sound different, or better, and what really makes a difference, is a completely different kind of, serious, rigorous, activity.

Many seem to confuse the two.

I’m sorry to start by preaching, but I think it is important to establish how careful one has to be to really answer these types of questions with a degree of certainty.

So, I’ll tell you what, after great efforts, I have found to be true.

I’m not saying you should believe me, and I suggest you don’t just believe what anyone says, but test carefully and thoroughly and always find out for yourself. But, be sure to be thorough and rigorous in your testing. Question your most fundamental beliefs as you work through your tests.

Because the subject is so involved, I’ll start with your last question for this first post, and work back to the first question in subsequent posts.

These cannot be complete answers in a single post, but hopefully they will provide a starting point for thought and discussion…and entertainment.

Your Question:

1) WHY DO I THINK CARBON SOUNDS MORE REAL THAN METAL AS A CONDUCTOR?

First, I’m not sure why you think that, as I don’t know how you have tested your hypothesis to arrive at your conclusion.

So maybe the question is not, "Why do you think carbon sounds better than metal?", but maybe your real question is, "Is it true that carbon sounds more real than metal?"

If my answer about whether it is true or not, doesn't satisfy, then we can explore the psychology and process of "Why do YOU think it is true?"

I have found that one must not make judgments about any component out of the context of the system that it is used in. It has almost no meaning to discuss component types outside of the circuit that they operate in, because the quality of their performance is dominated by the relational impedance interface within the system.

You might say, I have found this system (amp, pre-amp, loudspeaker, etc.) that when I use resister “A”, it sounds better to me that when I use resister “B”. You may even say, that I have found this pattern in a number of systems.

I would tend to ask first, what is it about these systems that they react in an unfavorable manner when used with resister “B”? Is the circuit interface compatible with resistors that are of type “B”? Is resistor type “B” being used in the manner that it is best suited?

While it might seem more obvious to question the component “resistor type B”, it turns out, that unless one has a component that is fundamentally defective, or is functioning outside a standard deviation of some important parameter, it is more likely that what you are experiencing as a sonic difference, is the system mismatch, not the inferiority of the component in isolation.

Sometimes one will find that in a certain component category, such as a capacitor, the one that is the “audiophile favorite” is not necessarily the best to use in a particular circuit.

At a recent show, people kept asking me what kind of capacitors I used in the crossover of ( the clue ) loudspeaker. I would have to say, that I cannot give a simple answer to the question.

I have a number of networks designed that all sound the same. Each network has different components from different vendors.

The reason I had to design different networks is that I have certain components that I cannot second source, so if they become unavailable, I have to substitute another component from another vendor.

But, in some cases, another component will not work perfectly, just substituted in, even if it has the same specified values. The transfer function will have changed a little. So, when I have to change one component, such as a capacitor from a one brand name to one of another, I may have to change the components that interface the input and/or the output of that capacitor to maintain the same dynamic transfer function, and sound quality. One cap wasn’t better than the other, but the system that it was operating within, was better matched to achieve my ideal input to output function.

So, to be able to use alternative components in certain places in the crossover network, I designed different networks to best work with different component set combinations while maintaining the same dynamic transfer function.

When I first substituted the replacement part, and found the system to perform worse, I could have blamed the quality of the part and decided not to use it, and call it “a bad part”. But, it wasn’t “a bad part”, it was a part that the system was not matched to. Once properly re-matched with new associated components, one could not hear any difference.

It can easily happen that substituting the most expensive component will actually disturb the transfer function of the total system. Sometimes it will sound different, when put in the circuit, and listeners will assume the difference means “better” because they think it is a better, more expensive component, that has a good reputation, but it doesn’t always work that way.

Good audio is predominately a systems design approach, not an individual component selection approach. The components have to be appropriate within a standard deviation, but beyond that, it is all based on the impedance interfacing of the system.

Some designers operate from a systems approach, and some try to use all the most expensive components and hope for the best result.

Whether one has an unlimited budget or is designing for lowest cost, the systems approach will provide a better result and will tend to be less wasteful. But, it takes longer and requires more careful assessment.

Cheers,

- James

[Kraniet]

Im a little bit more interested in the waveguide. I see you use a very small WG, so small that some people (like geedle) dont even call it a waveguide.
However I think reasonably sized WGs are more interesting and also would have more relevance in home-audio (as its more "design-driven").

Ive been looking at JBL* and what they call "elliptical oblate spheroidal waveguide" and by the looks of it it seem to be only 5x8cm or something like that. Thats very small when it comes to waveguides but they (JBL) seem happy with the directivity control, or matching, they get.
Looking at the site you linked to, small waveguides seem rare.

Do you fell that the WG have to be big to be effective? If one assumes a 2kHz crossover, is a size like JBL WG enough or is it just for show?

The DXT-lens is a diffraction lens and maybe not a true WG, but it is very effective att directivity control down to 2Khz. But being a diffraction-device there should be alot of HOM being created. Seeing the "rave" that this lens have got (like in the speakers from Acoustic Energy) it would seem that HOMs arent that bad?

In what way are you using the WG in your speaker and isnt it very short and of a "simple" profile to be effective as a WG? (no disrespect intended of course)


Heres the link to JBLs Linear Spatial Reference Studio Monitor System

*http://www.jblpro.com/catalog/general/Product.aspx?PId=26&MId=5

[Laila]

. . . . . .

Good audio is predominately a systems design approach, not an individual component selection approach. The components have to be appropriate within a standard deviation, but beyond that, it is all based on the impedance interfacing of the system.

Some designers operate from a systems approach, and some try to use all the most expensive components and hope for the best result.

Whether one has an unlimited budget or is designing for lowest cost, the systems approach will provide a better result and will tend to be less wasteful. But, it takes longer and requires more careful assessment.

Cheers,

- James

James, welcome to the forum !

Like poetry in my ears(eyes) . . . . wounderful words.

[Jacro]

james,

thanks for your post, i need some time to think it throu. i have much to ask but one thing comes to mind; this might be against your policy but, what do you think about technologies like www.embracingsound.com

best regards

Celef,

I look forward to your next questions.

In terms of the Embracing Sound System, I think it is a very promising technology.

The basic principle of an crosstalk cancellation systems have has been attempted and explored since the mid 1960’s.

In 1963, Schroeder and Atal out-lined a method for generating a phantom source. These ideas were elaborated by Damaske in 1971 and were followed in the mid-1970’s by Iwahara in Japan, with the JVC’s Bi-Phonic processor. As the pure theory suggests, all were made for use with Binaural recordings. In 1979 Robert Carver created a modified version of the JVC unit and offered a processor recalibrated for stereo recordings called Sonic Holography.

All were attempts to fix some of the fundamental spatial and tonal flaws of 2-channel stereo. Theoretically, under ideal conditions, it can be a superior approach, but is much more critical to optimize than conventional stereo. While potentially better than stereo, if it isn’t perfectly calibrated it can sound much worse than stereo. Ultimately, it is best suited for reproducing a binaural based recording.

These types of systems were always problematic with widely spaced loudspeakers, because the cancellation signals were very difficult to match, as they had to include the frequency response effects of the sound rapping around the face, to the opposite ear.

Around the same time (about 1980-1) we built a crosstalk isolation wall/barrier with a loudspeaker on each side of the barrier, with the wall projecting from the loudspeakers to the face of the listener. The spatial effects were quite impressive, and formed a reference sound field for further development work but the structure was not practical for consumer use. (Don Keele wrote an AES paper on this technique in 1986) With this structure as a reference, we worked on the processor approach and found that if we placed the loudspeakers close together that the correction for face-related-transfer-functions were much easier to apply and retain neutral tonal balance, and the spaciousness was even more impressive.

Unfortunately, due to the new movement towards multi-channel Surround Sound becoming popular, our two-channel project was halted and 5-speaker surround sound was where development efforts were redirected.

In 1995 new work began by Kirkeby, Takeuchi and Hamada, (University of Southhampton and Tokyo Denki University) as filed in British Patent App. # 9,603,236.2 “Stereo Dipole”. This closely spaced cross talk canceller was found to have lower coloration and greater stability of image with head movements. The same group wrote a series of very interesting papers on the subject of Stereo Dipoles, provide thorough analysis and synthesis information that was much more advanced than the work we did in the 1980’s.

It appears that the Embracing Sound Systems are based on the work of Kirkeby, Takeuchi, and Hamada (I don’t think their patent was ever granted) and if properly implemented, should be very good.

It has always been a system with great potential, particularly with binaural recordings. While it can provide an impressive demo with standard stereo recordings, it doesn’t reproduce an ideal transfer from stereo recordings, at least not as an accurate representation of the source, but it is a lot of fun to experiment with.

Now if we could get all the recording companies to offer a binaural alternative to every stereo recording, life would be great, but I don’t think that is going to happen anytime soon.

So, in the meantime, those of us that design reproduction systems, must work blind to the recording process, never knowing what kind of microphone technique is going to be used. Ultimately, an impossible task, but we persevere, searching for the best ways to create a believable facsimile of the original sound from an imperfect model.

All the best,

- James

[paa]

James,
Maybe you have a comment to Professor Edgar Choueiris work also?

http://www.princeton.edu/3D3A/

[paa]

I have been working on new waveguide topologies that are based on an elliptical variation of the Oblate Spheroid.
- James

Seems like Procella has gone towards that direction too (and it's smaller than it looks in the picture):

[Lust]

Here we have some larger wave guides ...

http://www.gedlee.com/

Experience/comments someone?

[Jacro]

Im a little bit more interested in the waveguide. I see you use a very small WG, so small that some people (like geedle) dont even call it a waveguide.
However I think reasonably sized WGs are more interesting and also would have more relevance in home-audio (as its more "design-driven").

Ive been looking at JBL* and what they call "elliptical oblate spheroidal waveguide" and by the looks of it it seem to be only 5x8cm or something like that. Thats very small when it comes to waveguides but they (JBL) seem happy with the directivity control, or matching, they get.
Looking at the site you linked to, small waveguides seem rare.

Do you fell that the WG have to be big to be effective? If one assumes a 2kHz crossover, is a size like JBL WG enough or is it just for show?

The DXT-lens is a diffraction lens and maybe not a true WG, but it is very effective att directivity control down to 2Khz. But being a diffraction-device there should be alot of HOM being created. Seeing the "rave" that this lens have got (like in the speakers from Acoustic Energy) it would seem that HOMs arent that bad?

In what way are you using the WG in your speaker and isnt it very short and of a "simple" profile to be effective as a WG? (no disrespect intended of course)


Heres the link to JBLs Linear Spatial Reference Studio Monitor System

*http://www.jblpro.com/catalog/general/Product.aspx?PId=26&MId=5

Thou shalt not judge a man by the size of his waveguide
- Anonymous Swedish philosopher

A waveguide doesn’t always have to be big to be effective. Generally, the purpose of a waveguide is to increase the directivity of the tweeter to more closely match the directivity of the woofer for frequencies that are at, and near, the crossover. To achieve that purpose, the waveguide should be comparable in dimension to the active radiating portion of the woofer.

Also, generally, one wishes to maintain the smallest center-to-center distance between the woofer and the tweeter, to avoid interference nulls in the vertical response.

So, if one had a 160mm woofer, with a tweeter attached to a 300mm waveguide, the overly large size of the waveguide would be problematic in at least two different ways:

- The tweeter/waveguide would have significantly more directivity at the crossover frequency than the woofer, causing a mismatch and very poor off-axis frequency response

- The center-to-center spacing between the woofer and the tweeter would be quite large (at least 230mm) causing a very narrow vertical listening window between the vertical nulls in the polar response

An additional point is that is significant, is that it is difficult to determine the radiating size of a given woofer just by measuring the size of the diaphragm. Some woofers and midranges will have a smaller source size with increase in frequency, due to the outer portion of the cone breaking up and decoupling from the center portion of the cone. So the waveguide may need to be a bit smaller than the cone diameter to have matched directivities at a given frequency.

Big woofers = Big waveguides.

Small woofers or midranges = Small waveguides.

With the JBL link that you provided, the crossover to the tweeter/waveguide is stated as 2.2 kHz. The waveguide appears a little small but again, it may be size-matched to the characteristics of the midrange. JBL considers off axis frequency response, and directivity matching, to be one significant parameters, so one can assume they chose their waveguide size very carefully.

The DXT waveguide does have diffraction steps but they are fairly shallow and hopefully don’t develop significant amounts of high order modes, but I am not certain what amount of HOMs are created in the DXT waveguide.
I would expect that the very good directivity control of the DXT waveguide when combined with a high quality transducer will produce good results.
It would be interesting to add a foam insert to absorb the high order modes and find out if eliminating the HOMs significantly improved the sound quality.

After extensive testing, the form and size of the waveguide in ( the clue ) was chosen to achieve the desired effect in matching the interaction with the cone driver over the transition frequencies. If we can make the system perform better by making the waveguide larger, or of a more complex form, we would do so, but so far, we have found the current waveguide to be the best match.

(No disrespect taken )

- James

[Kraniet]

Thats reassuring to hear. Listening to Geddes one get the feeling that everything goes as long as its by his defenition..

You mentioned Revel and JBL earlier and their work on evaluating meaningful speaker attributes. I think some of it we could all agree to, like a reasonably flat on-axis and a even and slowly falling off-axis for higher frequencies. But their testing method seem limited to "voicing" and percieved sonic balance. The issues with the stereo flaw doesnt seem to enter the picture.

Why is that do you think? Is it a little to "subjective" of a thing for them to be able to "double blind" it? As you say they seem aware of their limitations.

But isnt it also a good way to go, establishing the "true objective" attributes of speakers like frequency/power response, low distortion etc? I mean looking at the industry at a whole there seem to be a lack of an "even ground" for how a speaker should reasonably work for it to have a chance of being a true reproducer och the source material. Lacking those attributes it seems useless to optimize for stereo imaging. Or?

[Jacro]

James,
Maybe you have a comment to Professor Edgar Choueiris work also?

http://www.princeton.edu/3D3A/

paa,

The work of Dr. Choueiri is very impressive, and encouraging, in its thoroughness.

He has taken the cross-talk cancellation process beyond Local simplicity and has included the Global issues, such as listener body related transfer functions and the room issues. His use of the power of digital signal processing to address the secondary effects of XTC is what makes his system more encouraging than the work that has been attempted by others.

There is still the issue that post processing of this type is only “correct” for binaural recordings, but as he suggests, it will also work reasonably well for any stereo recording that has attempted to preserve the original inter-aural time and level differences.

Choueiri is one of the few that has suggested, and required, directivity control in the loudspeakers used with his system. He addresses how the room effects distort the correction signals and he has developed a very sophisticated filter system to address the tonal coloration found in all the prior art systems of this type.

His custom filter calibration to listener location and attributes is also a step forward.

I have been following his work since he first made it public, and I have not yet measured or heard one of his systems. But the fact that he is not just talking about the difficulty of secondary problems, and instead is actually addressing them is very encouraging.

Theoretically, he appears to be on the right track in many ways.

Cheers,

- James

[Kraniet]

seems limited to one person listening though. he speaks of it being viable for TV but I cant see how it would work in a normal living room with 4-6 people getting the full experience.

[Jacro]

seems limited to one person listening though. he speaks of it being viable for TV but I cant see how it would work in a normal living room with 4-6 people getting the full experience.

Yes, your observation and concerns are correct.

If one moves their head, even though the Stereo Dipole is more robust than other XTC systems, it still tends to be limited to an ideal listening region of less than half a meter square...Not enough for more than one listener.

Currently, the best 2-channel approach for multiple listeners is to use a conventional stereo system, incorporating time/intensity trading, with the most intense axis of the loudspeakers crossing in front of the listener, as originally proposed by Ben Bauer in the 1960s and Mark Davis at DBX in 1983 (US 4,503,553).

Modern versions of this approach use waveguides for directivity control and cross fire the loudspeakers in front of the listeners.

As a general rule though, the potential of the accuracy of an audio signal at the listening seat is inversely proportional to the size of the listening region around the head.

So, the best sound is always in the center, but the best cross axis solutions can make the sound quality remain quite satisfying a few people sitting together.

Best regards,

- James

[Jacro]

I have been working on new waveguide topologies that are based on an elliptical variation of the Oblate Spheroid.
- James

Seems like Procella has gone towards that direction too (and it's smaller than it looks in the picture):

Yes, a number of groups are starting to work with this form of waveguide, as it is found to be quite effective in regards to consistent directivity control, low linear/diffractive distortions, and allows reduced center-to-center spacing compared to axi-symmetric waveguides.

Some, such as Geddes, point out that it doesn't match the directivity of the midrange/woofer in the vertical axis as well as an axi-symmetric waveguide and they believe that the center-to-center spacing advantage is not as important as directivity matching.

The counter argument is that while the local directivity of an axi-symmetric waveguide and woofer, of approximately the same size, DO match better at the crossover frequency, the Global, combined vertical directivity of the two symmetrical units causes a significant directivity change at the crossover frequency.

And the pro- axi-symmetric group claims they can control that effect in the crossover... and... and...

Proper system design can result in great performance from either approach, with one offering some advantage over the other depending on all the system variables.

- James

[Jacro]

Hi James, welcome to the forum, what do You think is the secret about tubes and that they almost always sounding more pleasant and true to the ears? And do we really need all that watts, low powered amps seems to gain in transparent sound, may be construction simplicity or cost effectiveness?

Hi Objektivisten,

I still owe you an answer on your Tube amp and Power requirement questions.

I’m still not sure how to address such vast topics adequately in a single post.

So, to start, I’ll provide you with some information that you may find interesting from earlier research that we did, and I will spread the discussion over a few posts.


Part One (A partial review of Large Signal/Overload Issues):

1) WE FOUND THAT PARTIAL CLIPPING OF AN AMPLIFIER IS NORMAL USAGE FOR MANY LISTENERS.

- We monitored the systems of 92 different audiophiles over a 2-year period by monitoring their amplifiers in their homes and collecting data on how often their amplifiers were clipped.

When using high quality program material, we found that over 70-percent of the test subjects used their systems in a manner that regularly caused the onset of overload, or mild clipping, even though most of them would not be considered to listen at inappropriately loud levels.

(We eliminated the data of the subjects that obviously drove their systems beyond their capability and into hard clipping.)

(The average rated wattage of their amplifiers was 45 watts per channel, with the lowest being 9 watts and the highest being 350 watts per channel. Average rated loudspeaker sensitivity was 88dB/1 watt.)

2) WE FOUND THAT SOME AMPLIFIERS HAVE EUPHONIC CLIPPING CHARACTERISTICS THAT CAN ACTUALLY ENHANCE SOUND QUALITY DURING MILD CLIPPING LEVELS.

- We found that there is a wide variety of overload parameters and associated audible artifacts that are derived from those overload parameters.

• Harmonic distortion: Depending on whether the amplifier generated predominately low order*, even harmonics*, or high order**, odd harmonics** determined whether the clipping sounded “enhanced*” or “irritating**”.

Many tube amplifiers generate low order, even harmonics during overload, due to their asymmetrical clipping characteristics.

Most solid-state amplifiers produce primarily odd-order harmonics due to their symmetrical clipping characteristics.

• Other characteristics that can provide either ‘euphonic’ or ‘irritating’ sound quality during overload;

Overload In-phase or out-of-phase cross talk:
(Out-of-phase cross-talk can cause enhanced spatial effects. Some tube amplifiers exhibit this characteristic during mild overload)

Overload recovery time:
(Can prolong either positive or negative overload effects, unless recovery is extended due to “sticking” wherein an amp will stay “stuck” to one power supply rail is itself a malady of the design)

Overload output impedance increase:
(Can cause under-damping of loudspeaker during overload. High feedback amplifiers can have a greater delta between linear and overloaded output impedance, which may cause a more disturbing sonic result.)


3) SOME HIGHER POWERED AMPLIFIERS ARE CRITICIZED FOR POORER SOUND QUALITY WHEN COMPARED TO LOW POWERED AMPLIFIERS:

- It turned out, that in nearly every case, the higher-powered amplifier was not inherently worse sounding, but because it had more unclipped power, it was found that it could more easily over-drive the loudspeakers into greater non-linearity, causing the loudspeaker (not the amplifier) to have worse sound.

But since the amplifier was the only element changed in the system when comparing the sound, the higher-powered amplifier was most often unfairly blamed for the decreased sound quality.

4) In later studies we have found that many class-D amplifiers have an even more severe problem in that they start to have significantly increased distortion effects as they approach clipping, but have not yet overloaded. Because of this, they may not be able to be used in a manner that fully utilizes their full power capabilities without sonic changes.


So, in conclusion:

A - Hard clipping is very irritating in any amplifier, but mild onset clipping can provide a wide variety of sonic effects, some of which can actually be experienced as “more enjoyable” than unclipped. Many tube amplifiers exhibit similar overload characteristics that may be experienced as “euphonic”.

This effect of mild overload, being sometimes experienced as favorable, is partially due to the fact that program peaks tend to be very brief and the ear’s ability to accurately analyze the quality of the peak energy is greatly reduced and more easily fooled as compared to the ear’s ability to evaluate long-term average levels.

B – Larger amplifiers can easily be blamed for poorer sound quality if they are used with loudspeakers that change sound quality when driven to the greater peak levels of the larger amplifier, even when the average levels are held to the same as comparable low powered amplifier.

Some loudspeakers have such low levels of linear capability that merely changing from a 10-watt to a 35-watt amplifier may cause this confusing effect of the larger amplifier being the cause of the sonic deterioration.

Again, this is a very limited discussion of the topic, but it may possibly explain experiences that some of you have had with differences between power amplifiers.

If there is further interest, I’ll try to follow up soon with a bit more information on the topic, including a few small signal issues with power amplifiers.

I originally interrupted your discussion here just to answer a few loudspeaker questions, and I don’t want to hi-jack your forum, so let me know if I am inappropriately dominating your discussions.

I would be curious to see what everyone else’s thoughts are on some of these subjects, and/or what everyone considers to be the largest remaining questions and areas of concern with their audio systems.

Best regards,

- James