AR-3a Response Curves and Test Reports

[Carlspeak]

I loved those old Rodrigues cartoons in SR like the one shown in the Hirsch review of the AR303 in your attachment. I'm still waiting for the tech to come to my house and remove my tweeters.

I have a whole book of those classic cartoons.

[Shacky]

Following attachment is the Electronics World test of the AR-3a showing the frequency-response averaged curves done by Julian Hirsch:

--Tom Tyson

Tom,

Is that dip at 75 Hz similar to what was just posted on our other 3a curve thread - being a room effect?

Jim

[tysontom]

Tom,

Is that dip at 75 Hz similar to what was just posted on our other 3a curve thread - being a room effect?

Jim

Jim,

That dip is almost surely attributed to a boundary, and it's probably the "suck out" that was measured as a cancellation of energy due to the reflected energy off an adjoining boundary (wall), out of phase, with the woofer's frequency response at those frequencies. The cancellation is manifested in the form of a dip in the frequency response, and this relative "dip" in response is heard throughout the sound field -- direct or reverberant. It is no fault of the AR-3a's woofer at all! Every enclosure-type speaker ever made (except for those with the woofers close to a boundary) will suffer from this pertubation, and this is the reason Roy Allison made the Allison: One. The AR-9 was the first AR speaker to reflect this thinking. Allison didn't set out to discover the room-boundary cancellation ("Allison Effect") when he did the article with Berkovitz; he was trying to prove the existence of the predominently reverberant field in a typical listening room. During the investigation, Berkovitz and Allison noticed the re-occurence of this dip, and he did further research into this phenomenon. It had been understood previously to some degree, but never quantified in the manner in which Allison did later on after he left AR. As for that dip, it can be easily corrected by mounting a speaker such as the AR-3a flush with a boundary, such as in a bookcase, for flattest bass response. You can take the speaker and put it further out in the room such as many audiophiles tend to do, but then you are looking at more of a 360-degree radiation angle, and the bass suffers badly there as well. 2-Pi (against the wall) is proper, but flush with the wall or bookshelf is ideal!

--Tom Tyson

[Carlspeak]

Jim,

That dip is almost surely attributed to a boundary, and it's probably the "suck out" that was measured as a cancellation of energy due to the reflected energy off an adjoining boundary (wall), out of phase, with the woofer's frequency response at those frequencies. The cancellation is manifested in the form of a dip in the frequency response, and this relative "dip" in response is heard throughout the sound field -- direct or reverberant. It is no fault of the AR-3a's woofer at all! Every enclosure-type speaker ever made (except for those with the woofers close to a boundary) will suffer from this pertubation, and this is the reason Roy Allison made the Allison: One. The AR-9 was the first AR speaker to reflect this thinking. Allison didn't set out to discover the room-boundary cancellation ("Allison Effect") when he did the article with Berkovitz; he was trying to prove the existence of the predominently reverberant field in a typical listening room. During the investigation, Berkovitz and Allison noticed the re-occurence of this dip, and he did further research into this phenomenon. It had been understood previously to some degree, but never quantified in the manner in which Allison did later on after he left AR. As for that dip, it can be easily corrected by mounting a speaker such as the AR-3a flush with a boundary, such as in a bookcase, for flattest bass response. You can take the speaker and put it further out in the room such as many audiophiles tend to do, but then you are looking at more of a 360-degree radiation angle, and the bass suffers badly there as well. 2-Pi (against the wall) is proper, but flush with the wall or bookshelf is ideal!

--Tom Tyson

Aren't you talking basically about baffle step diffraction which occurs with centrally located woofers in speakers located out from the wall? There are BSC circuits commonly in use today to correct for this.

[tysontom]

Aren't you talking basically about baffle step diffraction which occurs with centrally located woofers in speakers located out from the wall? There are BSC circuits commonly in use today to correct for this.

Carl,

Well, that dip could very well be the result of baffle-step diffraction, but what it looked like to me was an abrupt dip in the response, more akin to a boundary issue. I think that baffle-step diffraction is a more of a gradual reduction (more like 6dB/octave) in output as a result of the transition from 2π to 4π radiation, whereby a transducer becomes increasingly omnidirectional in its radiation pattern below utlimate radiation impedance, down to resonance, when the loudspeaker is measured into 360 degree solid angle. The response rolls off gradually down to resonance. This is the reason that measuring a speaker's low-frequency output in a 4π anechoic chamber is wrong and gives the rolled-off appearance (a la Toole's AR-3 early measurement). Measuring the speaker in a 2π environment eliminates most of this issue.

--Tom Tyson

[tysontom]

I loved those old Rodrigues cartoons in SR like the one shown in the Hirsch review of the AR303 in your attachment. I'm still waiting for the tech to come to my house and remove my tweeters.

I have a whole book of those classic cartoons.

I used to have that book, "Harmonic Disturbsion," or something to that effect.

--Tom Tyson

[Zilch]

An interloper responds:

Thank you for providing this bonanza of AR3a data, Tom!

A great deal has been said about the AR-3a’s performance from an objective-measurement standpoint, and in my opinion, this discussion has for the most part been excellent! Despite some naysayers, the fact is that the AR-3a measures extremely well: in its day it was perhaps the flattest, smoothest loudspeaker money could buy; even today it is still a superb loudspeaker, capable of holding its own against many of the top loudspeakers currently available! It also shows that a speaker that measures well will also sound good!

I certainly agree with that last part.

There is really little question that what you hear in a normally damped listening room is a combination of direct and reflected sound, not primarily direct sound (the exception might be some high-energy, on-axis horn tweeters). This has been known for many years and reported in the literature (and vigorously disputed by some on this forum and elsewhere), but there were no published objective measurements of the home-listening environment that existed prior to the time the AR-3a was introduced in 1967, and for this specific reason, Roy Allison and Robert Berkovitz did the research in the Boston area and reported in their landmark AES paper, The Soundfield in Home Listening Rooms, posted on the CSP in its complete form thanks to the effort of Howard Ferstler. Allison and Berkovitz set about to answer the primary question, “…is the field at any practical listener location primarily direct rather than reverberant?” Their conclusion, verified by their extensive response-curve data, was simply “We are convinced that home music listeners perceive the spectral balance of the sum of the direct and reverberant fields, and that the very small time difference between them has no effect on this perception of balance.”

Agreed, all of that, but you've neglected what follows therafter:

"For a realistic assessment of loudspeaker spectral performance, it is clear that both types of response curves are needed: reverberant energy response and anechoic curves to investigate directivity." [Emphasis original.]

Therefore, anywhere in the listening environment what the listener hears is reverberant energy -- the sum of the speakers’ radiation at all angles. It is difficult to interpret this paper in any other way! The direct radiation is embedded in the total energy radiation, and it does not predominate. AR’s experience was that any normal listening room integrates the total output of a speaker system at all angles and that beyond three to four feet from the speaker a listener is no longer in the direct field but actually in the reverberant field.

Allison and Berkovitz posited this hypothesis, but what they proved is that the Beranek model fails in small rooms, instead, and the "UT, oh" moment is clearly revealed:

"There is one obvious contaminant left in the data: the effect of average room absorption with frequency"

Yup, that, in combination with the reflectivity and diffusivity of surfaces and objects within the space, is what kills the imaginary reverberant field above the transition frequency.

The AR-3a has such wide dispersion at all frequencies it is definitely affected by cabinet-molding diffraction, and this anomaly causes the direct frequency-response measurement to look jagged and rough, but this has no affect on the total acoustic output. The same is true of driver overlap at crossover and so forth, but these anomalies aren’t audible in the reverberant field, as shown by the total acoustic output of the speaker.

The measurements were taken with 1/3 octave resolution, masking the detail. Read up here:

http://www.churchsoundcheck.com/resolution2.html

Once 22 of the measurements are averaged together, it's all gone, and only the basic trend remains visible.

This up-close fr measurement of the AR-3a also caused the Consumer Reports reviewers to be critical of the AR-3a. In order to measure the total energy output prior to the development of computer-generated tests, AR developed its semi-reverberant test chamber, and an accurate representation of the total energy of the loudspeaker could be assessed without having to factor in room-response curves for every different room. “Semi-reverberant” in that there was a 2π baffle facing into a reverberant chamber.

Such was the mantra at the time, but you mischaracterize the chamber. Allison and Berkovitz described it:

"It is commonly recognized that interference and diffraction do not change the total energy radiated by a speaker system; they merely redirect it, bunching it at favored angles for particular frequencies. The truth of this can be verified easily by diffuse-field measurements in a very reverberant environment. Fig. 17 is the response curve of this same AR-3a system taken in AR's reverberant chamber. Input to the system in this case is pink noise. The microphone (located behind the cabinet, so as to prevent any direct radiation from reaching it) is flat for random-incidence energy. Its output is fed to a General Radio swept 1/3-octave filter with coupled chart recorder. Superimposed on the response curve is the chamber calibration for flat energy input."

A photo is shown in an additional link you provided -- no baffle, and not 2-Pi:

http://www.classicspeakerpages.net/IP.Boar...ost&id=4807

Not "semi-reverberant," but "very," instead.

AR measured each driver individually, on- and off-axis, in the anechoic chamber. This quantified the accuracy and smoothness of each individual driver without the influence of interaction of drivers at crossover, diffraction effects and other anomalies. The woofers were measured out doors flush in the ground facing upward, 2π-fashion, and these curves were spliced to give one an idea of what the overall frequency response of the speaker would look like without interference effects. Nonetheless, the acoustic-power curve is the most revealing data in that it shows what the speaker will be doing in a real room and what is being heard by the listener.

Ken Kantor informs us that the reverberant chamber was retired shortly thereafter, which may be viewed as confirmation that Allison and Berkovitz's landmark paper made it clear to AR that it was all but useless for the purpose. Indeed, as stated in the "Million measurements" paper:

"Here are some things we have learned about testing: our findings are that the most important measurements required to asses the accuracy of a speaker system are:

1) Frequency response on-axis,

2) Frequency response off-axis, and

3) Integrated power output,"

industry standard to this day....

[Zilch]

Regarding Hirsch and his measurements:

This is the best frequency-response curve we have ever measured using our present test set-up.

Even more significant to us was the flatness of the output between 550 and 15,000 Hz; ... the total variation was +/- 1.5 dB over this entire frequency range.

The AR 303 had significantly flatter response at the high frequencies, however, about 3 to 5 dB stronger than the AR-3a above 6 kHz.

On-axis at 1 meter, it's output variation +/- 1.5 dB from 300 Hz to 18 kHz, certainly among the best responses we have yet measured from a speaker.

The NHT data reveals the facts....

[Shacky]

Jim,

That dip is almost surely attributed to a boundary, and it's probably the "suck out" that was measured as a cancellation of energy due to the reflected energy off an adjoining boundary (wall), out of phase, with the woofer's frequency response at those frequencies. The cancellation is manifested in the form of a dip in the frequency response, and this relative "dip" in response is heard throughout the sound field -- direct or reverberant. It is no fault of the AR-3a's woofer at all! Every enclosure-type speaker ever made (except for those with the woofers close to a boundary) will suffer from this pertubation, and this is the reason Roy Allison made the Allison: One. The AR-9 was the first AR speaker to reflect this thinking. Allison didn't set out to discover the room-boundary cancellation ("Allison Effect") when he did the article with Berkovitz; he was trying to prove the existence of the predominently reverberant field in a typical listening room. During the investigation, Berkovitz and Allison noticed the re-occurence of this dip, and he did further research into this phenomenon. It had been understood previously to some degree, but never quantified in the manner in which Allison did later on after he left AR. As for that dip, it can be easily corrected by mounting a speaker such as the AR-3a flush with a boundary, such as in a bookcase, for flattest bass response. You can take the speaker and put it further out in the room such as many audiophiles tend to do, but then you are looking at more of a 360-degree radiation angle, and the bass suffers badly there as well. 2-Pi (against the wall) is proper, but flush with the wall or bookshelf is ideal!

--Tom Tyson

I didn't think it was the AR's Thanks for the education.'

BTW beautiful bookshelf

[Pete B]

An interloper responds:

Now Hear This, you interloper: Resistance is Futile, you must join the cult.

Clear your mind, stare off into space and repeat after me:

All that matters is the reverberant field, ... field ... field

All interlopers must assimilate!

[Carlspeak]

I used to have that book, "Harmonic Disturbsion," or something to that effect.

--Tom Tyson

The title is "Total Harmonic Distortion". It was published in 1988 shortly before the sale of Stereo Review's owner to a French publishing giant.

Surprisingly, there is a used copy for sale at Amazon.com for $250!

[tysontom]

An interloper responds:

Thank you for providing this bonanza of AR3a data, Tom!

I certainly agree with that last part.

Agreed, all of that, but you've neglected what follows therafter:

"For a realistic assessment of loudspeaker spectral performance, it is clear that both types of response curves are needed: reverberant energy response and anechoic curves to investigate directivity." [Emphasis original.]

Allison and Berkovitz posited this hypothesis, but what they proved is that the Beranek model fails in small rooms, instead, and the "UT, oh" moment is clearly revealed:

"There is one obvious contaminant left in the data: the effect of average room absorption with frequency"

Yup, that, in combination with the reflectivity and diffusivity of surfaces and objects within the space, is what kills the imaginary reverberant field above the transition frequency.

The measurements were taken with 1/3 octave resolution, masking the detail. Read up here:

http://www.churchsoundcheck.com/resolution2.html

Once 22 of the measurements are averaged together, it's all gone, and only the basic trend remains visible.

Such was the mantra at the time, but you mischaracterize the chamber. Allison and Berkovitz described it:

"It is commonly recognized that interference and diffraction do not change the total energy radiated by a speaker system; they merely redirect it, bunching it at favored angles for particular frequencies. The truth of this can be verified easily by diffuse-field measurements in a very reverberant environment. Fig. 17 is the response curve of this same AR-3a system taken in AR's reverberant chamber. Input to the system in this case is pink noise. The microphone (located behind the cabinet, so as to prevent any direct radiation from reaching it) is flat for random-incidence energy. Its output is fed to a General Radio swept 1/3-octave filter with coupled chart recorder. Superimposed on the response curve is the chamber calibration for flat energy input."

A photo is shown in an additional link you provided -- no baffle, and not 2-Pi:

http://www.classicspeakerpages.net/IP.Boar...ost&id=4807

Not "semi-reverberant," but "very," instead.

Ken Kantor informs us that the reverberant chamber was retired shortly thereafter, which may be viewed as confirmation that Allison and Berkovitz's landmark paper made it clear to AR that it was all but useless for the purpose. Indeed, as stated in the "Million measurements" paper:

"Here are some things we have learned about testing: our findings are that the most important measurements required to asses the accuracy of a speaker system are:

1) Frequency response on-axis,

2) Frequency response off-axis, and

3) Integrated power output,"

industry standard to this day....

Zilch,

Thanks for your comments. I think AR had more than one reverberant chamber; the first one was built well before the introduction of the AR-3a -- perhaps in the mid-1960s -- and it was fairly small. My reference to “semi-reverberant” meant that the chamber was not 100% reverberant, and that a portion was the 2π-baffle portion. I have a picture of the first chamber somewhere if I can locate it. I think Roy Allison also used such a chamber in the mid-70s at Allison Acoustics in Natick.

Perhaps there is more agreement than disagreement, overall.

--Tom Tyson

[Zilch]

Perhaps there is more agreement than disagreement, overall.

Perhaps so.

The core issue in the prior discussions here has been whether there exists in typical listening spaces a diffuse, isotropic reverberant field which normalizes the spectral balance and takes advantage of the wide dispersion AR strove to achieve in the individual drivers.

Allison and Berkovitz demonstrated that it's NOT there, that flat energy response in a highly reverberant chamber does NOT translate to flat in-room acoustic response, and it's easily seen that what we actually hear is the direct field combined with early reflections as revealed in the anechoic measurements. The reverberant field thesis has long since been repudiated by subsequent researchers, and AR themselves abandoned it as a design principle soon thereafter.

More fundamentally, if what we like about the way AR3a's sound is derivative of wide dispersion, I submit that it is far more productive to ascertain the actual basis of our preferences than to continue relying upon this fictitious one....

[Steve F]

Perhaps so.

and AR themselves abandoned it as a design principle soon thereafter.

For historical perspective, it should be noted that the ADD line (10 Pi, 11, 12, 15, 18, etc.) which followed the Classic line (3a, 5, 2ax, 6, 7, etc.) used the exact same or extremely similar driver layouts and had the exact same dispersion goals. The Verticals (9, 90, 91, 92), then the next series of "serious" ARs (the 9ls, the 98LS, and the 78LS) thru 1984-ish also used the 1 1/2" dome mid and 3/4" dome tweets--the latter ones using the 'dual dome' mid/tweeter assembly for even GREATER, more uniform vertical dispersion than earlier AR speakers.

After these speakers (with the extremely notable exception of Ken's Magic Speaker), AR became much more of a marketing-oriented company, and its bread-and-butter items were there to fill price points and driver configurations as dictated by competitive market conditions.

That doesn't mean that their speakers were bad or somehow disingenously-designed; far from it: the various BX, BXi, TSW, Connoisseur, Spirit, M Series and the like up through the early-90's were pretty good, to varying degrees. But they weren't guided by an overall corporate performance philosophy like the Classics or ADD's; they were guided by marketing needs.

AR didn't 'abandon' its core product philosophy of 1954-1984 because they somehow had an epiphany that their original philosophy was wrong; rather, their product focus became scattered because they let the market get away from them and lost their initiative. Their rise and fall would make a perfect graduate Marketing case study.

Whether or not their 'core product philosophy' was wrong or not is not what I'm speaking to here. I'm just pointing out that the fact that later AR's didn't adhere to the exact same performance philosophy as the 3a is not an automatic indicator that AR had an engineering change of heart.

By the late '80's, AR's 32% market share was LONG gone, corporate profitability was under extreme stress, and those realities influenced their product decisions--as they always do.

Steve F.

[Zilch]

AR didn't 'abandon' its core product philosophy of 1954-1984 because they somehow had an epiphany that their original philosophy was wrong; rather, their product focus became scattered because they let the market get away from them and lost their initiative. Their rise and fall would make a perfect graduate Marketing case study.

I'm not saying any of that, particularly NOT that they abandoned max dispersion, rather, the reverberant chamber and, implicitly, an imaginary reverberant field as a primary element in their design approach. It's appparent in AR 303, and in reissues of AR3a, even, that AR engineering learned somewhat more than many here surmise from the Allison and Berkovitz study.

Ken has discussed the objectives of AR 303 design elsewhere in this forum, and the result is documented in the NHT measurements and Hirsch's review. Do we like AR 303 as well as AR3a? Better?

[Steve F]

I'm not saying any of that, particularly NOT that they abandoned max dispersion, rather, the reverberant chamber and, implicitly, an imaginary reverberant field as a primary element in their design approach. It's appparent in AR 303, and in reissues of AR3a, even, that AR engineering learned somewhat more than many here surmise from the Allison and Berkovitz study.

But they DID abandon the 'max dispersion' approach in the late '80's. That's simply a fact--look in the Library on this site. The Connoisseurs and TSW's had 6 1/2" midrange units (the 810 and 910 having DOUBLE 6 1/2" units, quite the opposite of the 3a and 11's single 1 1/2" dome), so quite clearly, the 1984-1992 AR's were not max dispersion speakers. The 1992 Classic 18-22-26-28 speakers had dual 6 1/2" or dual 5 1/4" mids--again, not max dispersion.

What I'm pointing out is that this was done because AR had become more of a marketing-driven company by then (and a financially struggling one, to boot), and their products fit marketing-oriented price and configuration points to a far greater degree than 20 years earlier.

Ken's 303 family was in the 1993 and later time frame. I'm not addressing those. I'm addressing the original drift from wide dispersion (3a-5-10 Pi) to random design philosophy (58 BX, 810, M6, Classic 28) and pointing out that those 1984-1992 speakers were not driven by an overriding, consistent, wide-dispersion/engineering philosophy to the extent that the earlier ones (1954-1984) were.

That's all.

Steve F.

[Zilch]

The fact is that in any enclosed space short of VHB airplane hanger size the reverberant field will dominate over the direct field, provided the off-axis output of the speaker is decently strong. And in a smaller room, such as a typical home listening room with the boundaries closer than what we have in a concert hall, the reverberant field will be even more emphatic, due to the shorter delay times.

It's high-order early reflections, Howard, not an imaginary steady-state reverberant field, which in-room absorptivity precludes from even developing in typical "small" rooms, substantially as Toole describes. You have not demonstrated that a reverberant field even exists, in any significant measure, and neither did Allison & Berkovitz. Kill the early lateral reflections (they're easily located), and the soundfield collapses. You have more work to do, Howard, before publicly embarrassing yourself by declaring Toole "clueless...."

[genek]

Ken has discussed the objectives of AR 303 design elsewhere in this forum, and the result is documented in the NHT measurements and Hirsch's review. Do we like AR 303 as well as AR3a? Better?

In my case, the answer was no, I didn't (sorry Ken). My listening impression of the 303 was that it was more brightly voiced than the 3a or 10pi, but had a narrower "sweet spot" than either. I would describe it as sounding esentially like an AR-3 (not 3a) with its treble on steroids. After a weeklong trial comparing them side-by-side with my AR-2ax's, I returned the 303's.

I never had a chance to listen to the 303a, so I don't know what changes might have been made to that model from the 303.

[Zilch]

After a weeklong trial comparing them side-by-side with my AR-2ax's, I returned the 303's.

Clearly, it's not a fine line distinguishing their relative spectral balances.

How about spaciousness?

[genek]

Clearly, it's not a fine line distinguishing their relative spectral balances.

How about spaciousness?

I thought the 303 had less than the 2ax. That was part of what I was referring to when I mentioned "sweet spot." If I was positioned just right between the speakers they sounded grand, but if I got up and walked around the room the illusion collapsed, compared to my 2ax's and the 3a's I have now, which allow me to walk from one side of the room to the other, including locations outside the stereo pair, without any perceptable change in virtual source locations or that quality of sound I generally think of as "airiness." With the older speakers, moving around is like walking across the rear of a hall, while with the newer ones it was like leaving the room while the musicians played musical chairs.

FWIW, my living room has a hard floor, 14-ft vaulted ceilings, no carpeting or area rugs, mostly leather and wood furniture and an almost entirely glass wall on one side, and there's a definite echo in there that carries every sound all the way to the other end of the house. Every home I've owned for the past 30 years has had a similar configuration, but this one is more so than any of them.

[Zilch]

I thought the 303 had less than the 2ax.

If AR2ax may be said to exhibit spaciousness, it is not born of "max dispersion:"

http://www.classicspeakerpages.net/IP.Boar...ost&id=4808

[genek]

If AR2ax may be said to exhibit spaciousness, it is not born of "max dispersion:"

I'm not getting involved in that discussion (I don't have the measuring equipment). I know what the speakers I have sound like, and that I've not heard any others that come close since AR discontinued the ADD series (I missed my chance to hear the Allison efforts). I'll leave the theories on what's been missing since then to others, and be ready to listen to and report on anything anyone produces that claims to have put whatever it is back.

[tysontom]

If AR2ax may be said to exhibit spaciousness, it is not born of "max dispersion:"

http://www.classicspeakerpages.net/IP.Boar...ost&id=4808

Zilch,

Where did you get this term, "max dispersion?" Is this a reference to a pulsating sphere, or is it a snide reference to wide-dispersion tweeter design? Quite frankly, wide dispersion and "spaciousness" do go hand-in-hand, in my view, and the AR-2ax (while not as good as the dome midrange versions -- see attachment below) does have wide dispersion and it definitely has "spaciousness." Interestingly, the AR-2ax with its 3 1/2-inch midrange driver might "image" a bit better than an AR-3a or AR-5 due to its poorer dispersion in the upper midrange, but some might like that tonal balance. But since "imaging" is largely nonsense anyway, what difference does that part make? Having a more spacious stereo presentation is better, in my view, once back in the (and I hate to say it!) reverberant field. In other words, I think a speaker with wide dispersion calls less attention to itself than one that has a narrower dispersion through the midrange.

Comparison of the anechoic-chamber radiation pattern of the 3-1/2-inch AR-2ax midrange with the 1 1/2-inch AR-3a midrange (the AR-303's response pattern is very similar to this driver):

--Tom Tyson

[Zilch]

Where did you get this term, "max dispersion?" Is this a reference to a pulsating sphere, or is it a snide reference to wide-dispersion tweeter design?

It is a characterization of a significant design objective of the era, once the deficiency of cone drivers was recognized and acknowledged as such. Clearly, Allison was a subscriber. Different manufacturers pursued alternative means of achieving ever wider high-frequency dispersion capability.

Quite frankly, wide dispersion and "spaciousness" do go hand-in-hand, in my view, and the AR-2ax (while not as good as the dome midrange versions -- see attachment below) does have wide dispersion and it definitely has "spaciousness."

What I said. If AR2ax has spaciousness, it's not coming from wide dispersion. There's not much upon which Howard and I agree, but among those few items, that wide dispersion begins at 120° beamwidth is major (it took me over a month to coax him to commit), and AR2ax might just as well forgo having a tweeter at all at 60° off-axis.

Interestingly, the AR-2ax with its 3 1/2-inch midrange driver might "image" a bit better than an AR-3a or AR-5 due to its poorer dispersion in the upper midrange, but some might like that tonal balance. But since "imaging" is largely nonsense anyway, what difference does that part make?

If the sample of visitors to ZilchLab is any indication, it's apparent that few vintage loudspeaker enthusiasts have even a clue what imaging actually is.

Having a more spacious stereo presentation is better, in my view, once back in the (and I hate to say it!) reverberant field. In other words, I think a speaker with wide dispersion calls less attention to itself than one that has a narrower dispersion through the midrange.

Midrange is rarely the issue. We now know what cues are responsible for the perception of spaciousness, and it is possible to have both spaciousness and imaging concurrently. Employing gratuitous wide dispersion, they are in large degree mutually exclusive, as you observe. I believe Howard agrees on this point, as well, and advocates multi-channel as the solution, particularly, adding a center channel.

Comparison of the anechoic-chamber radiation pattern of the 3-1/2-inch AR-2ax midrange with the 1 1/2-inch AR-3a midrange (the AR-303's response pattern is very similar to this driver):

I have not seen the AR-303 polars, I don't believe. Did I miss them somewhere here?

[tysontom]

It is a characterization of a significant design objective of the era, once the deficiency of cone drivers was recognized and acknowledged as such. Clearly, Allison was a subscriber. Different manufacturers pursued alternative means of achieving ever wider high-frequency dispersion capability.

It really has nothing to do with the shape of the driver, only its diameter. A tiny cone has just as good dispersion as a tiny dome. The advantage of the dome is the fact that it is more rigid, less prone to break-up because it is rim- driven. Allison's tweeter was an inverted cone, but it had asymetrically radiating portions which contributed to its very wide dispersion. So Allison didn't subscribe to the "deficiency of cone drivers." So the characterization "max dispersion" came from you, eh? It certainly doesn't come from the literature.

What I said. If AR2ax has spaciousness, it's not coming from wide dispersion. There's not much upon which Howard and I agree, but among those few items, that wide dispersion begins at 120° beamwidth is major (it took me over a month to coax him to commit), and AR2ax might just as well forgo having a tweeter at all at 60° off-axis.

Look again at the anechoic on- and off-axis response of the 3/4-inch tweeter, same as used in the later AR-2ax. For entertainment, I've compared the performance of the waveguide tweeter, which is not generally rated up above 15kHz, to that of the AR 3/4-inch dome tweeter. Granted, these were not measured in the same venue under the same conditions, but the off-axis characteristics are clearly evident, and the dome tweeter is far superior off axis than the waveguide, which is really not rated off-axis beyond 15kHz, as I said. Note too, that the waveguide is charted at 10dB/division vs. 5dB/division for the AR tweeter, so the waveguid looks smoother in that presentation, and more linear in its off-axis decline. It's a less-rigorous method of measuring, for that matter.

If the sample of visitors to ZilchLab is any indication, it's apparent that few vintage loudspeaker enthusiasts have even a clue what imaging actually is.

Who might these visitors be? Does anyone have a clue as to what imaging is? Could you possibly render a definition of the term?

Midrange is rarely the issue. We now know what cues are responsible for the perception of spaciousness, and it is possible to have both spaciousness and imaging concurrently. Employing gratuitous wide dispersion, they are in large degree mutually exclusive, as you observe. I believe Howard agrees on this point, as well, and advocates multi-channel as the solution, particularly, adding a center channel.

With a directional midrange, "spaciousness" is not going to exist, particularly with a high crossover frequency to the tweeter in 3-way designs. Wide dispersion across the board is needed for spaciousness. A multi-channel surround system largely solves the problem, but for stereo-only, wide dispersion is a necessity for "spaciousness," in my view.

I have not seen the AR-303 polars, I don't believe. Did I miss them somewhere here?

No, I mischaracterized radpat. I do have the polar plots somewhere, however.