Fullrange Drivers

[GettinDegreez]

While browsing a bunch of DIY sites the other day, I had an idea. Is it possible/a good idea to use full ranger drivers such as Lowther, Fostex, etc. in a multi-way speaker? Too my knowledge no one has ever done this. My thought behind it would be that a full range driver could be used to handle from about 100Hz up to say about 7-8kHz, and then a tweeter to handle the rest up to 20kHz, this would help in different ways. First you would have a speaker that would work well in that range eliminating the need for complex multi-driver crossover that has midbass, midrange, and tweeter, and the crossover could be designed using a first-order design. This supposedly makes the speaker time coherent because it doesn't introduce any phase changes(granted you also time align the drivers on the front baffle). Setting the crossover this high also eliminates the need to have the crossover in the range of 1kHz-5kHz which is apparently the most sensitive part of the human hearing range and when a crossover is put in there it can mess stuff up. You also would be able to use a smaller driver such as a 5 inch driver that would handle that whole range. It's my understanding that when the wavelength approaches the diameter of the driver the sound begins to start being beemed, so therefore the speaker would have better off axis response. Doing all this would also allow you to have a very high sensitivity speaker being that most full range drivers are very high sensitivity, normally 96db+. Couple that with a high sensitivity tweeter like a ribbon would allow you to have a speaker that could play very loud with little power. As for the bass end of things you could either use a dedicated subwoofer or make a bi-ampable speaker and install a woofer that uses an active crossover between the woofer and the full range driver.

Any comments on this? Anyone ever seen this before?

 

[Swerd]

Have a look at the Fostex BK-16 kit sold by Madisound.
http://www.madisound.com/bk16.html
I'm not sure what the crossover frequency is, but you could ask Madisound.

It's my understanding that when the wavelength approaches the diameter of the driver the sound begins to start being beemed, so therefore the speaker would have greater off axis response.

When a speaker starts beaming, it has less off-axis response, not greater. Was that a typo, or did you get it backwards? In the frequency response graph for that Fostex driver, as the frequency goes above 2000 Hz, the response at 30° and 60° off-axis falls off.

http://www.madisound.com/pdf/fostexdrivers/ff165k.pdf

There are two broad peaks between 3-4 kHz and 8-9 kHz. I would expect the crossover to smooth out the response in that area, if it is included in the design. Still, not bad at all for a 6.5" driver.

 

[WmAx]

Is it possible/a good idea to use full ranger drivers such as Lowther, Fostex, etc. in a multi-way speaker? Too my knowledge no one has ever done this. My thought behind it would be that a full range driver could be used to handle from about 100Hz up to say about 7-8kHz, and then a tweeter to handle the rest up to 20kHz, this would help in different ways.

Most full range drivers are nothing special, in so far as their response characteristics. But one exception would be some of the metal-coned Jordan drivers. Some of these, used as a mid-range, would be superb, if you wanted to use a 1st order crossover, because the response is extraordinarily smooth for an extended range. However, they[as most speakers] will have very poor off axis dispersion past 3kHz, unless you go for a very small diameter driver, which will prevent a low distortion 100Hz lower limit as being feasible, that you seem to imply is important. I would use a 1st order slope at around 2.5Khz on such a driver. However, finding a tweeter to compliment this, will be a serious challenge.

First you would have a speaker that would work well in that range eliminating the need for complex multi-driver crossover that has midbass, midrange, and tweeter, and the crossover could be designed using a first-order design. This supposedly makes the speaker time coherent because it doesn't introduce any phase changes(granted you also time align the drivers on the front baffle).

True, but the importance of minimizing phase deviation as specified above is not demonstrated to be important in highly credible perceptual research projects.

Setting the crossover this high also eliminates the need to have the crossover in the range of 1kHz-5kHz which is apparently the most sensitive part of the human hearing range and when a crossover is put in there it can mess stuff up.

The problem in this range has to do with matching the dispersion characteristics of the transitioning transducers. You need to have the smoothest transition possible, if you want the best characteristics. This is one thing that is not a concern with full ranges, since there is not crossover transition to another transducer. But if you use a full range driver in a multiway system, the same problems will now be present. With good choices of drivers that have the closest dispersion characteristics at crossover transition point possible, along with a steeper crossover, to minimize problems as frequency rises above the crossover point, you can prevent this from being a problem.

You also would be able to use a smaller driver such as a 5 inch driver that would handle that whole range. It's my understanding that when the wavelength approaches the diameter of the driver the sound begins to start being beemed, so therefore the speaker would have better off axis response

.

Actually, the driver starts to beam when a frequency is radiated from multiple points that are spaced as to exceed 1/2 the wave length of that frequency. It is not as accurate to refer to driver diameter, because drivers are not perfect pistons, and the radiation from edges to center may be substantially different for higher frequencies, where that driver is no longer a perfect piston. A prime, exaggerated example, is the Manger transducer, which was specifically engineered to take advantage of this effect for a specific purpose. As a result, the driver has far superior dispersion at frequencies that would not be possible if it was a conventionally designed driver. However, lacking specific polar response plots for a driver, you can use the driver diameter as a rough guide, and it will generally suffice.

Any comments on this? Anyone ever seen this before?

I do remember seeing some multi-ways using full range drivers, but I can not remember specifics.

-Chris

 

[skrivis]

Most full range drivers are nothing special, in so far as their response characteristics. But one exception would be some of the metal-coned Jordan drivers. Some of these, used as a mid-range, would be superb, if you wanted to use a 1st order crossover, because the response is extraordinarily smooth for an extended range. However, they[as most speakers] will have very poor off axis dispersion past 3kHz, unless you go for a very small diameter driver, which will prevent a low distortion 100Hz lower limit as being feasible, that you seem to imply is important. I would use a 1st order slope at around 2.5Khz on such a driver. However, finding a tweeter to compliment this, will be a serious challenge.

-Chris

I agree that the Jordan modules are the best of the bunch when it comes to full range drivers. Ted Jordan has forgotten more about audio than the rest of the designers put together.

The Jordans will evidently do rather well when stacked vertically. Take a look at the Jordan Linear Array here: http://www.ejjordan.co.uk/systems/jxr6_systems.html

 

[skrivis]

True, but the importance of minimizing phase deviation as specified above is not demonstrated to be important in highly credible perceptual research projects.

-Chris

The studies relating to phase/time coherency or whatever you'd like to call it were flawed because they didn't have a valid control.

They started with a system that had phase/time problems and then added on more from there. The result was that nobody could detect a difference betwween screwed-up and more screwed-up.

 

[WmAx]

The studies relating to phase/time coherency or whatever you'd like to call it were flawed because they didn't have a valid control.

They started with a system that had phase/time problems and then added on more from there. The result was that nobody could detect a difference between screwed-up and more screwed-up.

I reviewed the text body of several phase distortion perceptual research papers this morning in order to review the transducers used. Let's examine one of the most heavily weighted/credited tests that have been undertaken in this regard:

On The Audibility of Midrange Phase Distortion in Audio Systems
Lip****z, Stanely P.; Pockock, Mark; Vanderkooy, John
JAES, Vol. 30, No. 9, 1982 September

In this study, multiple high-quality headphones and a measured/verified loudspeaker pair were used. The loudspeaker pair at subject here was the Quad ESL, and it has superior phase measurements than even most current hi-fi loudspeakers.

In order for phase distortions to become audible on loudspeakers used in a normally reverberant room, special test signals must be used to cause a reasonably audible difference; on actual music program, the differences disappear, or become so subtle as be difficult to perceive. Headphones were used on the test because they simulate an anechoic environment, and allow such distortions to be greatly enhanced in audibility. Here, audibility was subtle at best on music[and only with very specific music] and easily audible on special electronic test signals.

There is a good summary on Audioholics that reviews the current state of phase audibility distortion research, and includes feedback from Floyd Toole, one of the most informed loudspeaker perceptual scientists of today:

http://www.audioholics.com/techtips/roomacoustics/HumanHearingPhaseDistortio.php

-Chris

 

[skrivis]

I reviewed the text body of several phase distortion perceptual research papers this morning in order to review the transducers used. Let's examine one of the most heavily weighted/credited tests that have been undertaken in this regard:

On The Audibility of Midrange Phase Distortion in Audio Systems
Lip****z, Stanely P.; Pockock, Mark; Vanderkooy, John
JAES, Vol. 30, No. 9, 1982 September

In this study, multiple high-quality headphones and a measured/verified loudspeaker pair were used. The loudspeaker pair at subject here was the Quad ESL, and it has superior phase measurements than even most current hi-fi loudspeakers.


-Chris

I stand corrected. I had forgotten that they used the Quad ESL for that one. It's been a long time since I read that article.

There is also useful information at:

http://www.ocf.berkeley.edu/~ashon/audio/phase/phaseaud2.htm

 

[WmAx]

I stand corrected. I had forgotten that they used the Quad ESL for that one. It's been a long time since I read that article.

There is also useful information at:

http://www.ocf.berkeley.edu/~ashon/audio/phase/phaseaud2.htm

The Berkley link basically mirrors existing conclusions based on valid perceptual research. The sample he used in DBT[castanets sample from PCABX source site], required him to do multiple rapid switching back and forth in order to detect the difference, and he admitted it was subtle; this sample is particularly sensitive to time distortion(s) due to it's rather unusual composition[close-microphoned castanets], but it is still subtle, at best, and with repeated rapid-switched methodology required to identify the difference. This is exactly the kind of result one would expect with that particular sample, based on existing research.

In order to achieve 'optimal' phase response, which is marginally audible[and only on very particular samples], a typical multi-way system must be heavily compromised in so far as it's power response characteristics, which are of verified important audible relevance. In order to achieve both good phase response, and then also the verified important parameters, cost of transducers/system may have to be raised considerably in order to accommodate. An example is the Manger transducers, which due to their very effective bending mode behaviour, are able to produce smooth, rather good power response even into the treble band, even if not optimal, is far superior than a solution such as a direct radiation ESL. Beveridge had a special acoustic-lens ESL that had excellent power response, but this system raises cost considerably, and reduces the already low efficiency of an ESL. Another solution is to use co-axial designs. Theile has had some success in this area. Another solution is to use a filler driver method within a narrow band between the midrange/tweeter. This raises cost of the system considerably. One more solution is DSP FIR filters that can have both rapid roll-off and phase linearity. However, this is only true for on-axis response. Off axis, the response will be severely distorted with such filtering; this particular issue awaits proper perceptual testing, as it may end up negating the marginal advantage it provides. Combing this with the lens system used by Beveridge would remove the off axis problems, but would further increase cost, and result in reduced sensitivity.

-Chris

 

[skrivis]

In order to achieve 'optimal' phase response, which is marginally audible[and only on very particular samples], a typical multi-way system must be heavily compromised in so far as it's power response characteristics, which are of verified important audible relevance. In order to achieve both good phase response, and then also the verified important parameters, cost of transducers/system may have to be raised considerably in order to accommodate.

A look at Fried and Meadowlark show what kind of drivers are required. In particular, the Hiquphon tweeter is very capable and will handle a 2.5K 1st-order crossover with no problems. A 6.5" driver below this on a TL gives a -3 dB point of 35 Hz in several speaker systems that I know of. Dynamics are superb and there is no apparent compression up to uncomfortable listening levels.

These aren't inexpensive drivers, but neither are they exorbitant in price.

the system considerably. One more solution is DSP FIR filters that can have both rapid roll-off and phase linearity. However, this is only true for on-axis response. Off axis, the response will be severely distorted with such filtering; this particular issue awaits proper perceptual testing, as it may end up negating the marginal

Aha, a fly in the ointment. I assume you are referring to the type of thing DEQX is doing? The other info I have seen on these corrected brickwall filters has been uniformly glowing. I have been hoping to get a listen to the NHT Xds, but NHT doesn't seem to want to sell any since they keep dropping the ball after they said they would arrange a local audition for me.

The response plots shown for the NHT XD in the Stereophile review (http://www.stereophile.com/standloudspeakers/1105nht/index4.html) doesn't look bad at all. I realize this doesn't say that all speakers using things like DEQX will have wonderful polar response, but it says that it's not completely impossible either.

 

[WmAx]

A look at Fried and Meadowlark show what kind of drivers are required. In particular, the Hiquphon tweeter is very capable and will handle a 2.5K 1st-order crossover with no problems. A 6.5" driver below this on a TL gives a -3 dB point of 35 Hz in several speaker systems that I know of. Dynamics are superb and there is no apparent compression up to uncomfortable listening levels.

These aren't inexpensive drivers, but neither are they exorbitant in price.

I should have clarified. The problem is not that you can not find drivers that can handle this range(indeed, you can, for a higher price than a driver that would otherwise handle the range with a sharper cut-off for a substantially lower price), but let's examine the situation:

Multiway open radiator system with non-coaxial drivers: (1) The substantial overlap of the mid-range into the treble band lends to very poor off axis characteristics as the vectored points interfere with each other from two individual spaced sound sources. (2) The response is using a Butterworth(or similar) topology that sums on axis to 0dB, but each driver is at -3dB at crossover point(each is 90 degrees rotated, opposite of the other, at this point). As you move off axis, the signals will begin to approach points where the summed output is correlated in phase, thus causing peaks off axis, instead of nulls, which are less objectionable.

Note: A 6.5" driver should preferably not be used as both the bass driver and simultaneously into the kHz range. At louder passages, where bass moves the diaphragm a substantial amount, relative to where the highest frequencies are also launched from this same plane, an inter-modulation of phase occurs with the higher frequencies, that is very objectionable. The exception to this is where only low volume will be used, or where a number of 6.5" units are used so that movement remains small[a line array is a prime example].

The response plots shown for the NHT XD in the Stereophile review (http://www.stereophile.com/standloudspeakers/1105nht/index4.html) doesn't look bad at all. I realize this doesn't say that all speakers using things like DEQX will have wonderful polar response, but it says that it's not completely impossible either.

Note the pre-ringing evident in the step response plots. The ringing is inverse pressure of each other for the high band and low band driver, thus cancelling. However, this is exactly the problem, as off axis, the drivers can no longer match each other, due to the differing polar responses inherent to each one, and the vectoring points of distance difference, since this is a situation of multiple invidual drivers launching the signal from different origin points. Thus, it is no longer cancelled under these conditions. This is what remains to be tested in perceptual study. I have no idea at this point, if the effect will be inaudible, or if it is audible, to what degree it will effect sound quality[if at all].

-Chris

 

[skrivis]

I should have clarified. The problem is not that you can not find drivers that can handle this range(indeed, you can, for a higher price than a driver that would otherwise handle the range with a sharper cut-off for a substantially lower price), but let's examine the situation:

Multiway open radiator system with non-coaxial drivers: (1) The substantial overlap of the mid-range into the treble band lends to very poor off axis characteristics as the vectored points interfere with each other from two individual spaced sound sources. (2) The response is using a Butterworth(or similar) topology that sums on axis to 0dB, but each driver is at -3dB at crossover point(each is 90 degrees rotated, opposite of the other, at this point). As you move off axis, the signals will begin to approach points where the summed output is correlated in phase, thus causing peaks off axis, instead of nulls, which are less objectionable.

Yes, these are the standard objections to 1st-order crossovers.

If the output is good or even superior on-axis, need we worry about off-axis problems?

I realize this gets into discussions of the benefit of even power response and that people like Linkwitz make a good case for requiring good off-axis behavior. If I remember, this was even at the heart of the type of speaker that O'Toole found people preferred.

But do we care?

I've asked a number of people whether even power response will equate to early reflections from the room that have the same FR as the on-axis sound. I never really got an answer. Nor did I get an answer when I speculated that we should work on avoiding early reflections and not worry so much about controlling the source of those reflections. (I take that back... Linkwitz implied I was a moron in one e-mail , but that's cool with me. hehe)

There are certainly a lot of people working on designs that will display lobing. Most ESL speakers qualify, as well as other panel and line array designs. Then there's Vandersteen and Thiel, Johnson, Schuemann. Formerly we had Bud Fried and Pat McGinty. These aren't stupid people...

Note the pre-ringing evident in the step response plots. The ringing is inverse pressure of each other for the high band and low band driver, thus cancelling. However, this is exactly the problem, as off axis, the drivers can no longer match each other, due to the differing polar responses inherent to each one, and the vectoring points of distance difference, since this is a situation of multiple invidual drivers launching the signal from different origin points. Thus, it is no longer cancelled under these conditions. This is what remains to be tested in perceptual study. I have no idea at this point, if the effect will be inaudible, or if it is audible, to what degree it will effect sound quality[if at all].

-Chris

Ditto.

Do we need to worry about off-axis performance so much?

As long as we get good performance at one point, that's still better than all the other speakers that _never_ achieve the same accuracy.


I'm not saying that there is only one way to design a speaker system. Objectively, there is room for argument.

Subjectively, I can say you don't know what you're missing. hehe

 

[WmAx]

Y
If the output is good or even superior on-axis, need we worry about off-axis problems?

Yes, you must worry, if you have a reverberant field(as opposed to an anechoic chamber) in your listening room.

I've asked a number of people whether even power response will equate to early reflections from the room that have the same FR as the on-axis sound. I never really got an answer. Nor did I get an answer when I speculated that we should work on avoiding early reflections and not worry so much about controlling the source of those reflections. (I take that back... Linkwitz implied I was a moron in one e-mail , but that's cool with me. hehe)

The more linear the power response, the closer the room reverberant field will resemble the on axis response. That is a given, since the definition of power response is the acoustical power radiated off axis.

As for importance, Toole has [1]shown conclusively, in statistically significant, rigorous scientific study, that all things being equal, the off axis response linearity determines perceived sound quality.

You can [alternatively] use high levels of acoustical treatments to remove the reverberant field from contributing much, therefor the off axis response is irrelevant, for the most part. The problem with this approach for stereo reproduction is that you end up removing the reverberant field that lends to creating an enveloping ambiance for the listener. Stereo reproduction is not a true reproduction format, and there are no standards in producing music for this format, either. The stereo format can not reproduce spatial realism on it's own. But with the assistance of contributing room reflections, it can be greatly enhanced, due to psycho-acoustic effects of the reflections(if the reflections comply with several specific conditions).

There are certainly a lot of people working on designs that will display lobing. Most ESL speakers qualify, as well as other panel and line array designs. Then there's Vandersteen and Thiel, Johnson, Schuemann. Formerly we had Bud Fried and Pat McGinty. These aren't stupid people...

Designs with poor off axis behaviour require substantially more room treatment to perform optimally, as compared to speakers with an even power response. As for these people being 'stupid', as you put it, I have no comment on that, specifically. But if someone ignores the current state of perceptual science as it relates to their work, they are not demonstrating anything that anyone can conclude as being especially wise.

As long as we get good performance at one point, that's still better than all the other speakers that _never_ achieve the same accuracy.

I wonder what accuracy, since there are NO standards. I [and I wager most people] do not want a studio monitor sound in my listening room. I want something that can be perceived as realistic in more than just tonality. But objectively, off axis power response is critical to sound quality in real [reverberant] rooms. It can not be ignored unless you remove the reverberant field by a large degree[which has other problematic effects upon sound quality, as discussed above].

-Chris

[1]Loudspeaker Measurements and Their Relationship to Listener Preferences: Part 2
Floyd E. Toole
JAES, May, 1986, Vol. 34, pages 227-235

 

[skrivis]

The more linear the power response, the closer the room reverberant field will resemble the on axis response. That is a given, since the definition of power response is the acoustical power radiated off axis.

I would perhaps restate that. The more linear the power response, the more closely the sound radiated to various points in the room will resemble the on-axis response.

The actual reverberant field presented to the ears of the listener may or may not resemble what the speaker radiates. (I have asked a number of people about this and have not received an answer so far.)

As for importance, Toole has [1]shown conclusively, in statistically significant, rigorous scientific study, that all things being equal, the off axis response linearity determines perceived sound quality.

It's been a long time since I read these. I'll have to dig Toole's articles up and give them another look.

One thing that always bothered me was that speakers such as Bose designs would evidently do quite well in Toole's tests. However, they do not sound at all accurate to me. The Ohm Walsh radiators would also appear to do well, and, in my opinion, do sound better than Bose, but that's not saying much.

You can [alternatively] use high levels of acoustical treatments to remove the reverberant field from contributing much, therefor the off axis response is irrelevant, for the most part. The problem with this approach for stereo reproduction is that you end up removing the reverberant field that lends to creating an enveloping ambiance for the listener. Stereo reproduction is not a true reproduction format, and there are no standards in producing music for this format, either. The stereo format can not reproduce spatial realism on it's own. But with the assistance of contributing room reflections, it can be greatly enhanced, due to psycho-acoustic effects of the reflections(if the reflections comply with several specific conditions).

I don't think I can agree with this at all. For many years I was lucky enough to live in a house with a very large, open yard, and no close neighbors. I regularly put my stereo outside during the summer and listened to various things. I listened to a number of different speakers and electronics, and of course a wide variety of music.

I heard a fair bit of spacial "realism." In fact, I would say that a normal listening room muddles things up due to reflections. Now, I'm sure that the "realism" I heard was not what would have been heard at the original performance, but the same could be said about speakers in a reverberant room.

The best case scenario for stereo reproduction would be as if there were a wall between you and the performance, with two holes in that wall. (Two spaced microphones with an acoustic performance, and then two loudspeakers placed at two vertices of an equilateral or isoceles triangle, with the listener at the third vertex.)

Multi-mono studio recordings can only really give us the amplitude cues for locating sounds, so they are probably flawed right there. (We do then get head transform and time cues, but they're governed by the speakers and not the original performance.

It is probably necessary to eliminate the crosstalk between the speakers, either mechanically or with transaural processing. Otherwise you hear two events at each ear instead of one.

Designs with poor off axis behaviour require substantially more room treatment to perform optimally, as compared to speakers with an even power response. As for

There is a definite difference between sound events emanating from the speaker and then that same event as reflected back to the listener by the room. There will be time differences, and, unless the room reflects all frequencies evenly, frequency response differences. (I think there will be frequency-variant attenuation due to passage through the air and distance too.)

I feel that more room treatment is a good idea in all cases.

Pat McGinty made a statement something like," When you get the time and phase response right, the frequency and power response fall right into line." The measurements of his speakers don't appear to be horrible.

But if someone ignores the current state of perceptual science as it relates to their work, they are not demonstrating anything that anyone can conclude as being especially wise.

I'm sorry. I'm just having a hard time (and have had a hard time) swallowing this because the speakers I have heard that claimed to follow Toole's precepts did not sound very good to me. Perhaps I have not heard good examples of these, or perhaps there were other factors that swayed my opinion. (And opinion it is.)

[1]Loudspeaker Measurements and Their Relationship to Listener Preferences: Part 2
Floyd E. Toole
JAES, May, 1986, Vol. 34, pages 227-235

I'll have to look and see if I have this or not.

 

[WmAx]

I would perhaps restate that. The more linear the power response, the more closely the sound radiated to various points in the room will resemble the on-axis response.

The actual reverberant field presented to the ears of the listener may or may not resemble what the speaker radiates. (I have asked a number of people about this and have not received an answer so far.)

Unless you have a mechanism in place to damp the off axis output into the room within a specific bandwidth, it will be reverberated in a more or less even fashion. There is nothing inherent of a room to cause averaged excitation of certain frequencies over another in any substantial degree, except into lower frequencies where modal resonances begin to be a real factor.

One thing that always bothered me was that speakers such as Bose designs would evidently do quite well in Toole's tests. However, they do not sound at all accurate to me. The Ohm Walsh radiators would also appear to do well, and, in my opinion, do sound better than Bose, but that's not saying much.

Absolutely not. Bose speakers have a very poor power response, overall. They use very directive tweeters/drivers, that are simply pointed in more than one direction simultaneously. The speakers fail to meet the requirement as shown in Toole's research of smooth response.

I don't think I can agree with this at all. For many years I was lucky enough to live in a house with a very large, open yard, and no close neighbors. I regularly put my stereo outside during the summer and listened to various things. I listened to a number of different speakers and electronics, and of course a wide variety of music.

You need not agree with it, but your agreement is not in line with perceptual research preferences of listeners. Tool speaks about this very issue in one of his documents, though I can not remember specifically which at the moment.

As an anecdote[I am not introducing this as any sort of evidence, just sharing as my perspective, and I am placing the entire anecdote in italics so that it is not confused with any other parts of the post]: I have a considerable amount of room acoustic treatments on hand, as well speakers that have wide linear off axis response[almost omnipolar, even into high treble range] available. I can set up the room optimally, to balance out some reflections at the proper distances(>5ms), and dampen other spurious reverb and slap echo in the room, but not dampen 1st reflections of the ceiling, side walls and rear wall. Now, I can take this set up and add large absorption panels at all of the first reflection points. The sound quality takes a rather significant step down in realism. It goes from what I perceive as virtually real sounding vs. just a clear, detailed presentation. Adding further acoustic absorption into the room, even when not at 1st reflection points, also eventually hurts the presentation. A certain balance of room ambiance is needed with standard stereo recordings. My perspective of judgment is using Linkwitz's suggestion of going to live unamplified events and paying very close attention to the actual sound characteristics.

I can appreciate a near-field or such other set-ups with no or very little room effects, but they are not what I consider to be enjoyable, as they do not sound realistic with commercial recordings.

Toole agrees that a dead acoustic environment is not desired.

I heard a fair bit of spacial "realism." In fact, I would say that a normal listening room muddles things up due to reflections. Now, I'm sure that the "realism" I heard was not what would have been heard at the original performance, but the same could be said about speakers in a reverberant room.

I have no comment, as I can not even know what your perspective considers. I'm sure it could sound tonally accurate, and I'm sure it could even have some narrow depth to it. But the lack of room ambiance contributions prevents even a crude pschoacoustics trick[like stereo can do using strategic room ambiance] of any sort of enveloping acoustic perception.

It is probably necessary to eliminate the crosstalk between the speakers, either mechanically or with transaural processing. Otherwise you hear two events at each ear instead of one.

Toole has done extensive research in this area. In one situation, he poses a binaural speaker arrangement in an anechoic chamber. This uses active cancellation systems to remove crosstalk. Of course, standard stereo recordings can not work on this system, it uses binaural recordings like those made specially for headphones using a dummy head.

There is a definite difference between sound events emanating from the speaker and then that same event as reflected back to the listener by the room. There will be time differences, and, unless the room reflects all frequencies evenly, frequency response differences. (I think there will be frequency-variant attenuation due to passage through the air and distance too.)

When you exceed 5ms time difference, the brain begins to perceive the delay combined with direct signal as ambiance. In a recording with ambiance cues in it, the delays inherent of this supercede the room time difference, if the recorded delays/reverb are of longer duration than the room's. The multiple reflective points act as phantom sound sources, beneficial in simulating a real environment such as a performance hall or jazz club. As for frequency variant attenuation, this is well known. As frequencies rise into the treble region, especially high treble, they are attenuated by air. This can be several dB comparing 1' to 10' distance at 15khz, for example.

Pat McGinty made a statement something like," When you get the time and phase response right, the frequency and power response fall right into line." The measurements of his speakers don't appear to be horrible.

That statement makes no sense. Perhaps you took it out of context. But it does not make sense from what I see. So, an ESL with good time and phase response(which means by definition good frequency response) will have good power response? Or maybe he means something else other than 'good' when he states 'fall right into line'.

I'm sorry. I'm just having a hard time (and have had a hard time) swallowing this because the speakers I have heard that claimed to follow Toole's precepts did not sound very good to me. Perhaps I have not heard good examples of these, or perhaps there were other factors that swayed my opinion. (And opinion it is.)

Nor did you listen to these speakers under control conditions, I wager, to actually determine why you did or did not like something. Also, who knows if you have a preference that is like the overwhelming majority of listeners. There are always statistical anomalies. Perhaps you are one, which seems very likely, if you really prefer an anechoic environment for listening to stereo.

-Chris

 

[skrivis]

Note: A 6.5" driver should preferably not be used as both the bass driver and simultaneously into the kHz range. At louder passages, where bass moves the diaphragm a substantial amount, relative to where the highest frequencies are also launched from this same plane, an inter-modulation of phase occurs with the higher frequencies, that is very objectionable. The exception to this is where only low volume will be used, or where a number of 6.5" units are used so that movement remains small[a line array is a prime example].
-Chris

We would certainly prefer not to use a 6.5" driver over such a wide range. In fact, a more expensive model of speaker would and does include a "subwoofer" crossed over at 100 Hz to relieve the 6.5" driver from having to handle the lowest frequencies, thus increasing power handling and reducing IM.

Is this IM objectionable? I won't go so far as to say it's beneficial, as someone at the new Fried Products Co. seems to feel, but how damaging is it?

A number of years ago I saw some measurements on a similar speaker that indicated that IM was not appreciably worse than in more conventional speakers. I don't have access to these graphs anymore though, so this will have to remain anecdotal.

I have noted that cone excursion, as observed visually, is not as extreme with a TL loading the driver as it is with vented or sealed designs. I wonder how much influence the enclosure design has?

I also own a pair of Pinnacle Classic Gold Towers, which I recently found out were reasonably highly regarded in a review on this site. The bass of the TL speakers I've been talking about seems to be more extended, tighter, and more "real" than that of the Pinnacles.

I also note that you have to play the Pinnacles at higher volume levels because they seem to compress the signal. Solo piano, for instance, just does not sound dynamic with the Pinnacles, so there's a tendency to turn up the volume to "wake up" the sound a bit.

I imagine that the Pinnacles would produce higher SPLs. However, there is a rather limited range of levels where the Pinnacles sound ok. Below this range it seems as if you need a loudness filter, above this range they become harsh and unlistenable. The TL speakers seem to have a much wider range at which they are accurate.

Assuming that the perceptual research is valid, I would agree that a very strong case exists for your viewpoint. I'm just not as convinced that it is valid as I am with preamps and amps. My opinion is subject to revision though, so it's good to discuss these things.

I wish that it were easier to setup rigorous DBT conditions, or at least easier to get people interested in such testing. I've talked to a few people locally about this and they just aren't interested. They're solidly in the subjective camp.

I guess I'm still straddling the fence a bit. It seems like there are some areas where engineering and science have a very good handle on things and the perceptual tests are in agreement. But with loudspeakers it seems that there is more wiggle room in the engineering; there is certainly more room in the area of loudspeaker/room/ear interaction. So does the perceptual research tell us everything we need to know?

I can envision a loudspeaker system composed of a multitude of organ pipes. Such a system could do very well at steady-state tests like FR, PR, IM, THD, etc. It could have a very well-matched on and off axis frequency response. But with all of those high-Q resonators you won't get accuracy when playing music.

If the perceptual research tells us that the most important factor is smooth power response, do we then design speakers to that standard and more or less ignore other things? At what point is power response smooth enough to give us a good speaker? What role do other factors play in this? (As I said in a previous post, I need to dig up Toole's article.)

 

[skrivis]

Unless you have a mechanism in place to damp the off axis output into the room within a specific bandwidth, it will be reverberated in a more or less even fashion. There is nothing inherent of a room to cause averaged excitation of certain frequencies over another in any substantial degree, except into lower frequencies where modal resonances begin to be a real factor.

I'm sure this is true in a room with bare walls and floor. Does it also apply to a more normal listening room, or even more important, one with some judicious room treatment?

Absolutely not. Bose speakers have a very poor power response, overall. They use very directive tweeters/drivers, that are simply pointed in more than one direction simultaneously. The speakers fail to meet the requirement as shown in Toole's research of smooth response.

I was just looking, and it's certainly hard to find measurements on Bose speakers.

points, also eventually hurts the presentation. A certain balance of room ambiance is needed with standard stereo recordings. My perspective of judgment is using Linkwitz's suggestion of going to live unamplified events and paying very close attention to the actual sound characteristics.

There's a difference between reproduced ambiance and produced ambiance. Ambiance is produced by the hall or room that the recording was done in. Ambiance is produced by the listening room, unless it is anechoic.

I don't think that we can _reproduce_ the ambiance of the original room in our listening room. The best we can do is produce some pleasant euphonic effect, or we can try to reduce the room's involvement as much as possible.

I can appreciate a near-field or such other set-ups with no or very little room effects, but they are not what I consider to be enjoyable, as they do not sound realistic with commercial recordings.

Toole agrees that a dead acoustic environment is not desired.

I guess I am unusual in this respect.

crude pschoacoustics trick[like stereo can do using strategic room ambiance] of any sort of enveloping acoustic perception.

I'm not sure I understand what you mean by "enveloping."

Toole has done extensive research in this area. In one situation, he poses a binaural speaker arrangement in an anechoic chamber. This uses active cancellation systems to remove crosstalk. Of course, standard stereo recordings can not work on this system, it uses binaural recordings like those made specially for headphones using a dummy head.

And the problem with those is that the head transform of the dummy head does not match that of the listener.

I have experimented with crosstalk cancellation using BruteFIR (not real-time, I was pre-processing the wav file ripped from a CD and then I burned it to a blank CD). I tried this with a normal speaker setup as well as something like the "stereo dipole." It seemed to me that I heard "measures of greatness" with that. I bought a Creative/Cambridge Soundworks PS 2000 speaker system for $20 that I have to take apart one of these days so I can use the processor with my regular speakers.

That statement makes no sense. Perhaps you took it out of context. But it does not make sense from what I see. So, an ESL with good time and phase response(which means by definition good frequency response) will have good power response? Or maybe he means something else other than 'good' when he states 'fall right into line'.

I will look and see if I can find the original reference.

Nor did you listen to these speakers under control conditions, I wager, to actually determine why you did or did not like something. Also, who knows if you have a preference that is like the overwhelming majority of listeners. There are always statistical anomalies. Perhaps you are one, which seems very likely, if you really prefer an anechoic environment for listening to stereo.

-Chris

And that's why I specifically said it was my opinion. If I had applied any sort of proper test conditions I woudl have said so.

I really do think that I prefer listening to only the original signal, since I would gladly listen only outdoors if that were possible. However, I do not like the perspective I get with headphones as much as that with speakers, so I don't think it's accurate timbre only that does it for me.

 

[WmAx]

I'm sure this is true in a room with bare walls and floor. Does it also apply to a more normal listening room, or even more important, one with some judicious room treatment?

Room treatments have a broad band absorption effect, with the non-linearity usually appearing in two circumstances: (1) a convoluted foam will have high co-efficient in a reverberant field when the convoluted surface(which is very low effective mass) is absorbing more energy due to increased surface area at high frequencies, than lower ones. (2)the absorption product reduces in coefficient as frequency lowers past it's effective bandwidth. Furniture, with very large foam pieces, such as a couch, has similar properties, unless it's covered with something like leather, in which case it becomes more reflective at high frequencies, and more absorbent at lower midrange frequencies. Many objects in rooms act as diffusers and reflectors as opposed to absorbers. Air produces more absorption at high treble frequencies. But these things are not a cause of some radical off axis response problem(s), as is the case with a speaker with poor off axis response. You can measure at various points in a room, and then average the points, and find that the power response is very well reflected in the room response in a normal situation[as opposed to some odd situation where a very usual device was used to create a substantial imbalance]. Probably the most important consideration is the speaker response relevant to the points where the 1st mirror reflections occur that are reflected back towards the listener.

I was just looking, and it's certainly hard to find measurements on Bose speakers.

There is no need, because by simple inspection of the drivers used, it is obvious that it is highly improbable that 2" or 3"[depending on model] paper cone tweeters produce a smooth and extended off axis response.

There's a difference between reproduced ambiance and produced ambiance. Ambiance is produced by the hall or room that the recording was done in. Ambiance is produced by the listening room, unless it is anechoic.

I am not intending for this to mean the room is creating more ambiance. It is not[except in the case of playing back a near-field mic-ed recording that has no reverb, in which case the room reverberation will become obvious]. What is happening is the phantom reflections are being picked up by the ears, and the brain is confused, causing a psycho-acoustic trick, fooling the brain into thinking it's an enveloping[surrounding] diffuse sound field. It only works properly with speakers that have an excellent and wide off axis linearity, and proper spacing from walls, in a symmetrical room placement.

I don't think that we can _reproduce_ the ambiance of the original room in our listening room. The best we can do is produce some pleasant euphonic effect, or we can try to reduce the room's involvement as much as possible.

Of course, you can not recreate the actual original sound-field with stereo standards[lack of] outside of special lab conditions. But you can create a realistic simulation that sounds like(or very close to) 'a' realistic sound-field.

I'm not sure I understand what you mean by "enveloping."

I mean, a surrounding immerse diffuse soundfield.


-Chris

 

[WmAx]

A number of years ago I saw some measurements on a similar speaker that indicated that IM was not appreciably worse than in more conventional speakers. I don't have access to these graphs anymore though, so this will have to remain anecdotal.

One would wonder the methodology used for this testing. What were F1 and F2 excitation frequencies? Amplitudes? You can mask results easily by using inappropriate excitation fundamentals and/or insufficient amplitude. There is an article/analysis on this subject at sound.westhost.com that you may find interesting, titled "Doppler Distortion in Loudspeakers".

I can envision a loudspeaker system composed of a multitude of organ pipes. Such a system could do very well at steady-state tests like FR, PR, IM, THD, etc. It could have a very well-matched on and off axis frequency response. But with all of those high-Q resonators you won't get accuracy when playing music.

Such an arrangement would produce some extreme phase response issues. At such extremes, I wonder if it would become easily audible.

If the perceptual research tells us that the most important factor is smooth power response, do we then design speakers to that standard and more or less ignore other things?

Ignore what other things? Are you talking about phase response again? If so, that is plainly obvious why it's not given much weight, when reviewing credible perceptual research on that subject.

At what point is power response smooth enough to give us a good speaker?

I do not know the level of threshold, or rule to this. Personally, I shoot for extreme values of off axis linearity to be sure.

-Chris

 

[skrivis]

axis response. You can measure at various points in a room, and then average the points, and find that the power response is very well reflected in the room response in a normal situation[as opposed to some odd

This is very good to know. Prior to this nobody else was willing to make such a statement to me (for whatever reason).

Probably the most important consideration is the speaker response relevant to the points where the 1st mirror reflections occur that are reflected back towards the listener.

I was involved in an interesting discussion about this here: http://www.audiocircle.com/circles/viewtopic.php?t=26784

I am not intending for this to mean the room is creating more ambiance. It is not[except in the case of playing back a near-field mic-ed recording that has no reverb, in which case the room reverberation will become obvious]. What is happening is the phantom reflections are being picked up by the ears, and the brain is confused, causing a psycho-acoustic trick, fooling the brain into thinking it's an enveloping[surrounding] diffuse sound field. It only works properly with speakers that have an excellent and wide off axis linearity, and proper spacing from walls, in a symmetrical room placement.

This still sounds to me like an artifact of the listening room. I heard this kind of thing from the Ohm Walsh speakers. It was pleasant at times, but you couldn't turn it off, nor did it vary depending upon the recording.

Of course, you can not recreate the actual original sound-field with stereo standards[lack of] outside of special lab conditions. But you can create a realistic simulation that sounds like(or very close to) 'a' realistic sound-field.

I've had described to me some recent work on producing "ambiance" by active means. In other words, adding extra speakers to the sides and back and using DSPs to extract and reproduce the ambient cues from the original recording. I suppose "surround sound" systems could do this too, but I don't think anyone is using it for anything but spectacular sound effects. The person I talked to reported excellent results, and others have reported the same after listening to his setup.

 

[skrivis]

One would wonder the methodology used for this testing. What were F1 and F2 excitation frequencies? Amplitudes? You can mask results easily by using inappropriate excitation fundamentals and/or insufficient amplitude. There is an article/analysis on this subject at sound.westhost.com that you may find interesting, titled "Doppler Distortion in Loudspeakers".

I don't recall. That's why I marked it anecdotal. I can't defend my statement.

I've already read most of Rod's articles including that one, but thank you for the reference. It's not quite the best article for explaining IM, but reducing excursion of the cone will reduce these types of distortion. My observation is that the 6.5" driver loaded by a TL exhibits less excursion at low frequencies than the same size driver in another type of enclosure. (This would be an interesting thing to try to measure.)

Such an arrangement would produce some extreme phase response issues. At such extremes, I wonder if it would become easily audible.

It might very well be audible. But it might also just measure wonderfully if all you're looking at is steady-state FR and PR.

A look at the Bose Waveradio and Acoustimass technologies is interesting too, and quite germane.



Also, it appears that Toole's collected papers aren't available at this point, and the 2 AES Loudspeaker anthologies I have don't contain the articles you've been referencing. I guess I need to cough up the money for #3 and #4. I also want the anthology with Heyser's collected works.