AR woofer crossover points in perspective

[Aadams]

This chart shows some AR woofer crossover points in the context of the musical range.  I think it helps explain why the AR4x still gets rave reviews and why the AR12 is an under-appreciated performer compared to the AR5, 2ax and AR14.

Note also, the crossover points of the 12” woofers.  After 1979 until the end of the dome mids the crossover in a 3 way was 700hz except in the AR303 which was 650hz. 

The top 12 inch and 10 inch systems always had lower woofer crossovers and the same is true today for full range 8 octave speakers.  Even for satellites and powered woofers you are encouraged, almost required, to cross over low.   

I could be wrong but I think I see a pattern.

 

Adams

[ligs]

Another consideration is the distortion figures  of the woofer. The heavy 12" AR woofer, while it has very low distortions at low frequencies, has relatively high distortions at higher frequencies. If the goal is to have the cleanest sound then cutting off the woofer at a relatively low frequency is probably the way to go. The very competent NHT 1259 woofer used in NHT model 3.3 is used only to 100 hz !

[tysontom]

"The heavy 12" AR woofer, while it has very low distortions at low frequencies, has relatively high distortions at higher frequencies."

Ligs, the graph you show above is from CBS Labs test of the AR-3a (that picture, by the way, with no attribution, looks like the one I supplied some time back; it even has my file number on it).  Your comment about the "heavy 12-inch AR woofer having relatively high distortions at higher frequencies" is vague, incorrect and appears to be based on assumption.  "Relatively high" compared to what?  You have to understand the measurement conditions before you can make such a statement.

First of all, the input-power level for this particular measurement extended to very high; you would never see such sustain-wattage figures for normal listening.  In addition, CBS Labs tested into full space in an anechoic chamber and did not test using the normal 2Pi steradians solid angle for their measurements, thus the distortion is at least twice as high.  It would be akin to listening to your speakers hung from a rope in the center of a room; bass response suffers without boundary reinforcement, and the woofer has to work harder without a boundary for reinforcement.  Nevertheless, the comments made by the lab on this distortion was that it was the among the lowest ever measured for a loudspeaker; so even with the high-power measurements, the distortion level was considered very low when compared with other loudspeakers, which holds up well even to today.

--Tom Tyson

 

[ligs]

“Your comment about the "heavy 12-inch AR woofer having relatively high distortions at higher frequencies" is vague, incorrect and appears to be based on assumption.  "Relatively high" compared to what? “

 

Tom. I am comparing the distortions at 300 hz vs 80 hz from the same speaker, 4.2% vs about 1% at the same 100 db spl. The latter is exceptionally low and is probably still unchallenged even today.  I have the 1976 edition of High Fidelity’s Test Reports. The harmonic distortion figures of AR-LST are similar to yours. Interestingly the 8” AR-6 woofer is no slouch either.

[Aadams]

Thanks, both of you.  This post was pointing out that AR apparently saw a way to improvement in their full range TOTL by lowering the 12" woofer crossovers.  It never got lower than 525hz until they incorporated cone mids.   For me, the matrix shows, at normal listening levels, the distortion of the woofer at 80hz is equal to the rated distortion of some amplifiers, which is good for a speaker.   To bad they didn't test all the way to the 700hz crossover point of the 12" three way verticals.

IN MY OPINION  the AR 12" speakers with the lowest woofer crossovers sound better, in the music packed human voice octaves, than those with higher crossovers.

Adams

[tysontom]

5 hours ago, ligs said:

“Your comment about the "heavy 12-inch AR woofer having relatively high distortions at higher frequencies" is vague, incorrect and appears to be based on assumption.  "Relatively high" compared to what? “

 

Tom. I am comparing the distortions at 300 hz vs 80 hz from the same speaker, 4.2% vs about 1% at the same 100 db spl. The latter is exceptionally low and is probably still unchallenged even today.  I have the 1976 edition of High Fidelity’s Test Reports. The harmonic distortion figures of AR-LST are similar to yours. Interestingly the 8” AR-6 woofer is no slouch either.

Ligs and Aadams,

Thanks for the comments and thoughts about the distortion levels and crossover points.  The important thing to note is the very low distortion for all of the AR woofers tested, and this was typical throughout years from all testing laboratories and magazine reviews, etc. 

From the very beginning in 1954, AR designed its speaker components by using accepted scientific testing methods—repeatable by anyone following these test methods—to have the lowest-possible harmonic distortion and the flattest frequency response that AR could design.  The results from these measurements were widely published for all the world to see, something fairly unique in the high-fidelity loudspeaker industry during the day.  In fact, for many years after the first AR-1, Acoustic Research was the only hi-fi manufacturer I know of to publish detailed distortion, frequency-response and acoustic-power response tests for all of their speaker systems.  Reluctantly, a few other speaker manufacturers did follow suit after AR's early success in the market place and with critical reviews, but the main reason that most manufacturers did not publish such information: lack of comprehensive testing equipment, anechoic chambers and a real fear of the outcome of those measurements and possible public backlash.

The point I was making to Ligs was that there were often variances in the method of testing, and this could cause slight differences in results.  AR maintained that by using the accepted standard testing methods of the day, for example RETMA STandard SE-103 and 61 IRE.RP1 Recommended Practices for Loudspeaker Measurements, anyone could accurately replicate the same results that AR got with their testing measurements.  By the way, McIntosh Laboratories' Roger Russell did extensive lab measurements for their speaker products (though they did not publish all of the results), and McIntosh later built and maintained one of the largest anechoic chambers in the industry for this purpose.

As for the crossover points on the three-way AR speakers, the lowest crossover was always chosen based on the lower frequency-response linearity of the midrange speaker.  The issue was dispersion of the large woofer cone but not so much the distortion of the woofer at its higher range, as it actually gets lower and lower the higher it goes up to its operating limit.  For example, the AR-3 has a 1kHz crossover from the 12-inch woofer to the 2-inch dome midrange speaker, and this was the lowest frequency that the midrange could operate.  AR would liked to have a lower crossover, but the 2-inch dome could not do it; with the AR-3a and the new 1½-inch midrange with its lower resonance, the crossover was changed to 575 Hz., thus improving the off-axis response at those frequencies.  Therefore, the crossover was based mostly on the operating range of the respective drivers, and there were very few compromises made to dispersion.  AR could easily have used a low-resonance cone midrange in the AR-3 to go much lower, but the dispersion and linearity of the 2-inch AR-3 dome was considered much more important to the overall accuracy and sound quality of the speaker, and AR was unwilling to give up that accuracy in the design.

—Tom Tyson 

[Aadams]

2 hours ago, tysontom said:

As for the crossover points on the three-way AR speakers, the lowest crossover was always chosen based on the lower frequency-response linearity of the midrange speaker.

This brings attention back the point of greatest distortion at normal listening levels, the area around 300hz, which is only two semi-tones away from 262hz (Middle C) and the area around which most popular music vocals are performed.  

Here is an arcane bit of trivia.  If you listen at the end of “Paradise by the Dashboard Lights” which is sung well below 500hz, you can clearly hear in one channel Mr. Loaf singing “It was long ago and far away ……………… while in the opposite channel Ellen Foley is singing a different lyric in counterpoint.  The Ellen Foley lyric is not completely intelligible on a 3a but can be heard on a 9 or another speaker where the voice is carried in single driver.  The arcane part is she can be heard clearly in the 3a up close in the mid-range driver alone as harmonics and overtones or in the woofer alone as fundamentals but back off and let the two sounds merge and it gets garbled.

With a crossover above 300hz voices are split from their harmonics and overtones between the mid and the woofer.  A crossover at 200hz puts almost all voice in a single driver as with a crossover at 1000hz.  I don’t know if driver arrangement, distortion or woofer crossover point is the answer, but voices are clearer when the they are carried in a single driver IN MY EXPERIENCE.

Hence the 4 way and maybe the inevitable de-emphasis of dispersion and power response.

Adams

[tysontom]

Aadams:

I think you are missing the point somewhat, and what you are saying is sort of getting in the weeds a bit.  This is also listening up too close to these speakers.  Up in the direct field, you will unavoidably encounter interference and diffraction effects, especially from the AR-3a, but from the AR9 as well.  Back in the listening area you would not be able to easily make the distinction you mention.

More importantly, you would have to set up the AR-3a and the AR9, side-by-side, and carefully balance the output levels and then switch back and forth to make that determination (probably best done with Double-Blind Tests).  I understand what you are saying, but you can't go on assumptions or perceptions.  In the reverberant field, you are not likely to be able to clearly make those distinctions anyway, and if you were at the original venue of the recorded performance, and you were listening to the integrated output of the entire recording, you would also not be able to clearly make those distinctions. 

There is little doubt that the AR9 has somewhat better integration of sound through each individual speaker, and the 8-inch lower-midrange driver on the AR9 is better suited for mid-bass in the 200-700 Hz range than the 12-inch AR-3a woofer up to 575 Hz, but I am confident there would not be huge, noticeable differences.  Also, the AR 12-inch woofer is very linear and flat within its pass band.  It is not "muffled" or lost in the frequencies it is reproducing; it is not missing notes or failing to reproduce sounds in its operating range.  There is no ringing or overhang or other damping issues.  I think you are getting caught up in some interference effects by listening too close to the speaker.

On the other hand, if you are looking more for more "imaging" and more detailed, up-close, intimate sound of the performers or the instruments, then neither of these speakers is going to be the best solution for your listening, although the AR9 images better than the AR-3a.   

—Tom Tyson

[Aadams]

Keeping in mind the thread topic of woofer crossover points:

For me imaging is a fortuitous by-product.  I normally listen to these speakers at 12-15ft and just park my brain listening but sometimes one is drawn to the lyrics of a song.   When there is a solo voice I begin to focus on intelligible lyrics within the balanced overall sound, which is how I came to discover this threshold difference between the 9  and the 3a. 

The speakers are sitting side by side, inches apart, on different amps, same source, same equalizer and attenuators at zero.  The referenced passage is not absolutely clear on any speaker of mine at normal listening distant but is less so on the 3a.  To hear it clearly on the 3a your ear needs to be almost in the driver, maybe 3" away at low volume.  I know the woofer is doing its part but it does not reproduce frequencies much above the crossover point.  If a performer sings a 440hz A, then the first natural harmonic is 880hz. With a 575hz crossover design,  the fundamental and some overtones will be placed in the woofer and all prominent harmonics and remaining overtones are in the mid and that is what I hear. This is not true with the 9 or the 3 but it is for all other 3 way 12" ARs. 

 It is not a problem I was just pointing out some arcana that can be demonstrated at my house with my limited resources and what  I have observed when instruments are split by crossovers.  This same effect can be observed by crossing a powered woofer in the low mid range and spatially  separating it from the upper drivers.  A vocal recording will seem to bounce from speaker to speaker

And, I am not criticizing the 3a because I have never heard anything superior for presentation of orchestral recordings but, human voice can reveal its weakness perhaps because we are most sensitive to those frequencies and timbres in which we naturally communicate. Perhaps keenness for speech is why I have not read of Live vs Recorded tests being done with vocalists.

Aadams

Edited April 23, 2018 by Aadams

[tysontom]

Aadams,

Out of curiosity, what "equalizer" are you using with your setup and why?  Do you have a picture of your AR-3a/AR9 listening-room setup?

The AR-3a woofer is not going to contribute much to the overall sound of a vocalist because, as you noted, most of the overtones are going to be well into the midrange frequencies.  The woofer will reproduce the fundamental energy very accurately, of course, but the midrange is where most of vocal energy (the overtones) are reproduced and what you would hear.  

The AR-3a's dome-midrange driver reproduces these higher frequencies with nearly perfect perceived "clarity," insofar as this driver has nearly perfect transient response and is linear and flat with low distortion at normal listening levels.  However, it also has much better dispersion in the 600 Hz range than the AR9's 8-inch LMR cone driver, so there will be more interference effects and reflections from the floor and walls, grill molding and so forth, than the AR9, and this could reduce "intelligibility" somewhat.  This is typically never heard back in the reverberant field. 

The AR9's 8-inch lower-midrange driver operates from 200 Hz to 1200 Hz, so it covers a big range for vocalists, and this driver is more directional and will "image" better than a small dome driver.  In addition to all this, the average output level from the AR-3a's midrange driver is down 2-4 dB from the 12-inch woofer's level, whereas the output of the AR9's 8-inch lower-midrange driver is even with respect to its woofers.  For the AR-3a, this slight difference in spectral balance likely gives the sensation of slightly less output in that critical voice frequency range. 

I realize that you are not trying to criticize the AR-3a, but this discussion is largely academic, and it has little to do with the perceived accuracy the AR-3a reproducing music.  You stated, "...Perhaps keenness for speech is why I have not read of Live vs Recorded tests being done with vocalists."  There have been several LvR demonstrations done with vocalists, but AR used a string quartet, classical guitarist, drum solo and a 1910 Nickelodeon for its well-known public demonstrations.  The reason the quartet was chosen was the difficulty in reproducing the ensemble tone and acoustic-power balance, which proved to be difficult to accurately reproduce and as a result, easier for listeners to detect differences in the live and playback sessions.  Even with this, few listeners could reliably detect the switch-overs.  Voice reproduction is difficult, but probably not as much so as a string quartet.

—Tom Tyson

 

 

[Aadams]

1 hour ago, tysontom said:

Voice reproduction is difficult, but probably not as much so as a string quartet.

The point in question is not important to me and I did say it was arcane.  

I have absolutely zero dispute with anything you have said and have complete confidence it is accurate.  

Have you listened to the piece of music in question? The final 20 seconds is all this is about?

Regarding mid-range fidelity: I can verify you are correct. The mid range driver can be heard to reproduce the lyrics with unquestionable clarity if your ear is very close to the driver.  Same for the woofer, as I said before.  And you appear to agree with me the apparent improved vocal clarity of the AR9 is because the voice range is contained almost entirely in the 8” driver.

My point was merely that the AR3a ultimately is not as good at presenting vocals in my living room at 12-15ft as my AR9s and by extension any of the other 4 way ARs which are very similar.  Whether the cause is diffraction or splitting the instrument in the crossover and creating difference in timing or phase, related to driver location, I do not know but somebody at AR decided the difference was sufficiently great that in 1979 all top of the line systems became 4ways for at least a decade and the 12” 3 way became second tier.

The two speakers in question are on a 10 band software equalizer seen below

The room is 3000cft and rectangular, moderately live.  The full wall opposite the speakers in the long dimension is a bookcase full of irregular shapes.  Speaker attenuators or pots are set to max output.

I think you are going to say the equalizer settings put the 3a midrange at a disadvantage due to its slightly reduced output in relation to the woofer.  You very well could be correct, but I am not changing anything because this is not a genuine problem.

The only thing you have said that I take issue with is people are better at hearing differences in strings than voice.  The only reason I caught this is because I understand spoken English. If this had been German, Italian or Stradivari or Guarneri It would have been heard as fadeout music and completely unnoticed. 

If you hear a familiar voice you can put a face with it.  When you a hear a middle C can you, or most people, distinguish between a viola and violin?  Much less which luthier did the build. 

Cheers

Adams

[Aadams]

On 4/25/2018 at 8:12 PM, Aadams said:

I think you are going to say the equalizer settings put the 3a midrange at a disadvantage due to its slightly reduced output in relation to the woofer.  You very well could be correct, but I am not changing anything because this is not a genuine problem

Tysontom

In retrospect, the above comment sounds perhaps terse and dismissive.  It was not meant that way.  That equalizer influences the sound of all my systems and it is simply not worth the effort to live with the ripple it will create for me to attempt resolving this tiny bit of difference.

I am fortunate that someone with your deep knowledge of AR, and speakers in general, is around and willing to talk about this. There are many that hole up and don't bother.  Apology offered.

The topic is still AR woofers and crossovers.

Re the 3a. I spent a lot of time listening to a wide variety vocals yesterday.  Without a reference, it is difficult to find fault in how the 3a presents voice but in relation to more recent vintage speakers that have smaller drivers covering the voice range, my 3a seems to add a slight warmth or heaviness in the tenor and low alto voices.   The difference is most noticeable with female voice. 

The arcane 20 seconds of which I spoke earlier is a low alto female voice.

I am still surprised at the superior presentation by the 3a of string ensembles and live recordings.  I no longer shop for new speakers but I would be surprised to hear anything new that can top an AR3a in those areas.  I don’t think this is true with voice, now that I am beginning be more attentive to musical lyrics.  Most people today, who listen to music, listen to music with lyrics and that is why I suspect AR changed the sound of their systems in such a telling way, beginning with the AR9.

Edit: I think the difference could probably be equalized out,with a 1/3 octave equalizer, to be practically imperceptible but back then that wasn't a cost effective idea. 

Adams

[tysontom]

14 hours ago, Aadams said:

Tysontom

In retrospect, the above comment sounds perhaps terse and dismissive.  It was not meant that way.  That equalizer influences the sound of all my systems and it is simply not worth the effort to live with the ripple it will create for me to attempt resolving this tiny bit of difference.

I am fortunate that someone with your deep knowledge of AR, and speakers in general, is around and willing to talk about this. There are many that hole up and don't bother.  Apology offered.

The topic is still AR woofers and crossovers.

Re the 3a. I spent a lot of time listening to a wide variety vocals yesterday.  Without a reference, it is difficult to find fault in how the 3a presents voice but in relation to more recent vintage speakers that have smaller drivers covering the voice range, my 3a seems to add a slight warmth or heaviness in the tenor and low alto voices.   The difference is most noticeable with female voice. 

The arcane 20 seconds of which I spoke earlier is a low alto female voice.

I am still surprised at the superior presentation by the 3a of string ensembles and live recordings.  I no longer shop for new speakers but I would be surprised to hear anything new that can top an AR3a in those areas.  I don’t think this is true with voice, now that I am beginning be more attentive to musical lyrics.  Most people today, who listen to music, listen to music with lyrics and that is why I suspect AR changed the sound of their systems in such a telling way, beginning with the AR9.

Edit: I think the difference could probably be equalized out,with a 1/3 octave equalizer, to be practically imperceptible but back then that wasn't a cost effective idea. 

Adams

Adams,

I didn't take your comments to be terse and dismissive.  No apology needed!  I thought your comments were measured and reasonable.  Also, your equalizer settings are not a problem from what I can see.

The big thing that is happening here is the effect of wide dispersion of the AR-3a at those midrange frequencies and how it affects the perceived clarity of voice or instruments.  It also occurs to me that if you are mounting your AR-3a on a stand or table (above the floor but back against the front wall), your AR-3a will be susceptible to the "Allison Effect," a boundary-caused dip of several dB as the reflection from the wall behind the speaker (and the floor) interact with the direct output of the woofer and cancel (null) certain frequencies to some degree, usually around 300 Hz or so.  This could impact intelligibility at those frequencies, of course, and the way to fix this problem is to mount the AR-3a flush with the wall or in a bookcase flush with books around the speaker so that the woofer sees a true 180 solid angle.  It's not always easy to do it that way!

With the AR9, the idea was to get the woofers to operate close to the floor-wall boundary to get maximum reinforcement and minimum boundary dip, such as the wall and floor, and to limit the pass band of the woofers for smoothest response.  Therefore, with side-mounted 12-inch woofers operating up to 200 Hz, there had to be a lower midrange driver (requiring a 4-way configuration) to handle the mid-bass frequencies.  This also allowed a higher crossover into the 1½-inch midrange dome to improve its power-handling capability, and so forth in the design of the AR9.  It was definitely a step forward in design.

The AR-3a has been criticized in the past for its somewhat "heavy" sound, and some of that problem was due to a crossover issue during the changeover from the Alnico woofer to the newer ferrite woofer in the 1969-1970 time frame.  Also, the relative balance of the midrange output to that of the woofer seems to give a slight sense of heaviness, but it is very minor.  The reduced output of the tweeter also added to this, but the overall acoustic-power output of the AR-3a is very smooth and linear, though downward-sloping through the midrange and treble.  Ironically, the acoustic-power response of the AR-3a in the highest frequencies is greater with less drop-off than most speakers (for example the original Large Advent) that are "brighter" and more forward-sounding than the AR-3a on axis! 

—Tom Tyson

[Aadams]

Thanks. 

There is a thread here asking 3a owners where they set their mid and tweeter controls.  I think the vast majority said tweeter max and mid at the mid point. I was the same. This tells me the  relative lower sensitivity of the mid is not a problem so much as the inability to contour the specific output in the voice band which leads to equalizers or complexities that many don't want to deal with.

9 hours ago, tysontom said:

The reduced output of the tweeter also added to this, but the overall acoustic-power output of the AR-3a is very smooth and linear, though downward-sloping through the midrange and treble.  Ironically, the acoustic-power response of the AR-3a in the highest frequencies is greater with less drop-off than most speakers (for example the original Large Advent) that are "brighter" and more forward-sounding than the AR-3a on axis! 

Its also ironic that, at the time in their lives when they worry most about about tweeter roll off, most geezers can't hear above 10khz, frequently less.

 

The topic is still AR woofers and crossovers but borders on mods and tweaks. BTW all of my 12" 3 way ARs are 12" off the floor and about 1 ft from rear walls. 

 

9 hours ago, tysontom said:

With the AR9, the idea was to get the woofers to operate close to the floor-wall boundary to get maximum reinforcement and minimum boundary dip, such as the wall and floor, and to limit the pass band of the woofers for smoothest response. 

Regarding your statement above   Does this imply that if I were to place AR58s or 11s ( for instance) back to back, in AR9 fashion, on the floor and limit their output to 200hz, I could achieve AR 9 level bass to 32hz that could also be easily managed?  Assume the tweeters and mids have been cannibalized.

Adams

[tysontom]

2 hours ago, Aadams said:

Thanks. 

There is a thread here asking 3a owners where they set their mid and tweeter controls.  I think the vast majority said tweeter max and mid at the mid point. I was the same. This tells me the  relative lower sensitivity of the mid is not a problem so much as the inability to contour the specific output in the voice band which leads to equalizers or complexities that many don't want to deal with.

Its also ironic that, at the time in their lives when they worry most about about tweeter roll off, most geezers can't hear above 10khz, frequently less.

 

The topic is still AR woofers and crossovers but borders on mods and tweaks. BTW all of my 12" 3 way ARs are 12" off the floor and about 1 ft from rear walls. 

 

Regarding your statement above   Does this imply that if I were to place AR58s or 11s ( for instance) back to back, in AR9 fashion, on the floor and limit their output to 200hz, I could achieve AR 9 level bass to 32hz that could also be easily managed?  Assume the tweeters and mids have been cannibalized.

Adams

Adams,

Regarding the settings of the level controls on the back of the AR-3a (and other AR speakers of this vintage), most people do leave the midrange close to the "dot" or center position with the tweeter at or near full tilt.  Even with the midrange output down somewhat from the woofer level, with the control set to max, the overall output sometimes makes the speaker seem too bright, probably due to the very wide dispersion at all frequencies.  But the controls were designed originally to tailor the speaker's balance to match the "liveness" of the listening room.

Placing two AR11s, AR58s or even AR-3as back-to-back on the floor would not emulate the AR9's bass extension, simply because the system resonance (fc) of the 11s, 58s, etc is higher at around 43 Hz and the AR9 has a lower system resonance of around 32 Hz, thus has a more extended bass output.  Placing two AR-11s back to back will increase bass output in the region of bass resonance (making the two sound a bit bass-heavy), but the resonance frequency remains the same (due to the AR-11 box size, "Q" and other factors), and bass response falls off at the rate of 12 dB/octave below resonance.  A bass equalizer would have to be used to flatten and extend the output below resonance.  Even with that, however, the output would not be quite as smooth as the AR9 in the 20-200 Hz range due to the AR9's purpose-built crossover for the two 12-inch woofers.  Overall, the harmonic distortion and so forth would be equal, however.  

--Tom Tyson 

[Aadams]

On 4/25/2018 at 6:51 PM, tysontom said:

The AR9's 8-inch lower-midrange driver operates from 200 Hz to 1200 Hz, so it covers a big range for vocalists, and this driver is more directional and will "image" better than a small dome driver.  In addition to all this, the average output level from the AR-3a's midrange driver is down 2-4 dB from the 12-inch woofer's level, whereas the output of the AR9's 8-inch lower-midrange driver is even with respect to its woofers.  For the AR-3a, this slight difference in spectral balance likely gives the sensation of slightly less output in that critical voice frequency range. 

The modified chart below is what I see based on the clarification above and the tendency of owners to keep the 3a mid output at around the attenuator mid point.  Maybe the 10Pi addressed this.  I know the LST did. 

[RoyC]

5 hours ago, Aadams said:

The modified chart below is what I see based on the clarification above and the tendency of owners to keep the 3a mid output at around the attenuator mid point.  Maybe the 10Pi addressed this.  I know the LST did. 

Interesting thread, Aadams...

I have nothing to add of a technical nature, but subjectively there is no doubt in my mind that the reproduction of vocals is not one of the 3a's better attributes, regardless of iteration. This is based on having heard hundreds of 3a's along the way.

Below is a Consumer Reports review of the 3a. My guess is their opinion was based on subjective impressions of vocal heavy popular music. I don't agree with the conclusion regarding the 2ax, but there is something to the 3a's midrange response issue.

Roy

[Aadams]

8 hours ago, RoyC said:

My guess is their opinion was based on subjective impressions of vocal heavy popular music.

Agree .  Voice is the only sound most untrained ears can blindly discern with good discrimination IMO.  

IMO, the tweeter issue is unrelated to the woofer output at 500hz and was just an easy target. Also, probably the the only way to make a 2ax outperform a 3a is by focusing on a specific type of human voice and ignoring all of the other superior aspects of the 3a that are happening simultaneously.  Ultimately, with the emergence of high power amps and fascination with loudness,  even a perfected 3a would not be competitive.  Today, if I felt it were important enough, I would use a  $300 31 band equalizer to eliminate the issue it but in the 1970s an Altec Acoustavoicette was  around $700.   And, who sells a pair of speakers one moment and in the next asks "what about that equalizer?"

Adams

[Steve F]

What if some contrarian wise-*ss was to offer the opinion that it was all the tweeter? The AR-11 used the exact same midrange and woofer at the later 3a's, yet the 11 was never accused of sounding "thick and heavy." Indeed, the 11 is considered by most to be an exceptionally neutral, well-balanced speaker.

That contrarian wise-*ss would say that perceived "heaviness" is primarily a function of spectral balance--the overall balance of lows-to-mids-to-highs, the general shape and slope of the FR curve. The 3a's slope is downward; the 11's is far more flat.

Funny thing--when the treble control on your system is advanced to around 2:00 (say about 5dB more) when playing a 3a, its so-called "heaviness" miraculously disappears. Coincidentally, the 11 has about 5 dB more treble output from its tweeter than the 3a had from its. More treble output and the lower-midrange heaviness goes away. 

Spectral balance.

(BTW, that's what the LST had over the 3a, too--and the LST used identical drivers and x-o topology as the 3a. It's a matter of spectral balance. The LST had none of the lower-mid heaviness that the 3a was accused of.)

Steve F.

[Aadams]

21 hours ago, Steve F said:

BTW, that's what the LST had over the 3a, too--and the LST used identical drivers and x-o topology as the 3a. It's a matter of spectral balance. The LST had none of the lower-mid heaviness that the 3a was accused of.)

3 posts prior.  LST is cited as not having this problem but the LST also had a completely different approach to achieving "spectral balance".    I have never heard an 11 or a 10pi  and not seen a graph for either and don't know if that same notch at 575hz is present.  I also stated in the previous post that the problem could be solved with a 1/3 octave equalizer.   IMO neither the world's most powerful tweeter nor a simple treble control adjustment around 5khz will restore the missing harmonics and overtones, between 600 and 1600hz, so you can clearly hear what Ellen Foley is singing, which is the origin of this thread.

Good to see you back.

Adams

Addendum: I found this in archives:  A letter from Allison to Hoffman, 1970.  Items D and E give insight into the thinking of the creator about the objective of the 3a design-----------Clearly, not drum demos in club sized rooms.   There are no response curves for the ADD designs to be found.

 

[Aadams]

Thread creep alert

This thread IS NOT about how good or bad the AR12” woofer is at producing voice or mid-range frequencies, though apparently it is quite good in comparison to other similar woofers.

This thread IS about how evidence shows, with each new generation of speakers, AR sought to increasingly diminish the role of the 12” woofer in the critical range between 80hz and 1000hz when used in their premier products.  

Some of the reasons for crossover choices are known but one thing verifiable today, beginning in 1979 the “AR Sound” that is today still seen as a measurement standard, is the sound of the 4 ways, in which the woofer was used for its best purpose and the critical mid- range was left to smaller drivers.

AR3 - 1000hz

AR3a - 575hz

AR10Pi, 11 - 525hz

AR98 - 200hz

AR9, ls, lsi, - 200hz

 

Adams

 

[Steve F]

To avoid Thread Creep, I would agree with Aadams’ basic premise, that all things being equal, it’s generally advantageous to have a heavy, low-resonance woofer restricted to operating as low as possible, below 200-300Hz, max. “All things being equal.” We’ll let that rather vague qualification stand alone, in all its unspecified glory. (BTW, later 3a's--from around the foam/ceramic days, as opposed to cloth/Alnico days--had a crossover of 525Hz, not 575Hz. I think there was a woofer choke change then too.)

As to the letter from Allison, I have seen this letter many times and it always struck me as strangely off-target.

The 3a’s—or any AR speaker’s—goal was to reproduce the electrical input signal as accurately as possible. The goal of the 3a was not to replicate the spectral balance of any arbitrary symphonic hall or night club. Imparting a sense of the performance venue—if it is to be done at all—is the job of the recording engineer, not the loudspeaker. AR did not and should not have had any requirement for its speakers to impose the tonal characteristics of any particular performance space upon the playback process in a domestic living room. That entire concept is fallacious on its face.

The response of the 3a may well have been uniform in a living room from 250 to 2500Hz, but its response in the succeeding octaves fell off precipitously.

I reject “C” on its face as being flat-out untrue. AR’s own curves—like yours above—clearly show the mid level as being 2-3 dB below the woofer and the tweeter level being some 5 dB below the mid. So, the tweeter is about 7 dB below the woofer. Yes, yes, I know that these are “individual driver curves,” not an integrated system curve, but AR had the honesty to put the individual driver curves onto this graph in a real-life level relationship to the other drivers. This is a very accurate graph of how the 3a sounds. Why the heck do people think it’s a little “reticent” or “thick”? Because it sounds exactly as AR very honestly and accurately represents it to sound on this graph. They also say—quite honestly—on the 3a Technical Sheet that the amplifier’s treble control must be advanced to 2:00 o’clock for flat response. That’s about 5dB, exactly what I said the 3a needs for a treble boost and exactly how much more on-axis output the 10π/11’s tweeter has vs. the 3a’s.

The 3a was a bit dull and its woofer level was a bit too high in relation to the other drivers. However, it had a whole bunch or other attributes that were really great, so if you could tolerate a down-sloping response tilt, it was a great speaker: unsurpassed bass from a 25” enclosure, super-wide dispersion, very low overall distortion and very smooth (albeit down-sloping) response. Far more good than bad. Far more. I’ll take it over almost anything in its size/price class and it has those classic looks too. 

But it’s not perfect.

Steve F.

[RoyC]

3 hours ago, Steve F said:

The 3a was a bit dull and its woofer level was a bit too high in relation to the other drivers. However, it had a whole bunch or other attributes that were really great, so if you could tolerate a down-sloping response tilt, it was a great speaker: unsurpassed bass from a 25” enclosure, super-wide dispersion, very low overall distortion and very smooth (albeit down-sloping) response. Far more good than bad. Far more. I’ll take it over almost anything in its size/price class and it has those classic looks too. 

 

But it’s not perfect.

Steve F.

I agree, Steve. In fact I actually prefer the AR-3a over the AR-11 and 10pi, as I personally find them to be a bit harsh in the mid frequencies...not an acceptable trade-off for me.

I would only add, however, that I have experimented a great deal with tweeters in the AR-3a. More powerful tweeters (crossed over at 5000Hz) may be somewhat helpful but also increase sibilance. In my experience they do little to improve the lower frequency issue under discussion. On the other hand, as you have suggested in the past, some treble controls can be beneficial if their influence reaches low enough into the mid frequencies. Unfortunately better controls found in higher quality preamps often affect only the highest frequencies, and are less beneficial.

Discussing speaker nuances and characteristics is what we enjoy doing here, but in the end the popularity of the 3a spoke/speaks for itself.

Roy

[genek]

With regard to the comments about "vocal heavy popular music," the jazz and classical music that AR's classic era ads leaned toward both had plenty of vocals. And if I'm choosing a vocal to use as a listening test source, I'll take Sarah Vaughn or Maria Callas over David Lee  Roth or Axl Rose any day.

[Aadams]

On 5/1/2018 at 2:40 PM, Steve F said:

The 3a was a bit dull and its woofer level was a bit too high in relation to the other drivers. However, it had a whole bunch or other attributes that were really great, so if you could tolerate a down-sloping response tilt, it was a great speaker: unsurpassed bass from a 25” enclosure, super-wide dispersion, very low overall distortion and very smooth (albeit down-sloping) response. Far more good than bad. Far more. I’ll take it over almost anything in its size/price class and it has those classic looks too. 

The thread is still creeping but that is ok because it is near the end.  SteveF, my eyes were damp after reading your closing remarks.  Speaking of the 3a as you did, reminded me of the great warrior Achilles who, with his classic look, could do extraordinary things but was terrible at picking the right shoes.

I spent this afternoon listening only to my 3as.  In my experience, there is still no speaker better at presenting a large-scale string recording.  My 9s have more bass and the sound chiseled but they don’t ever provide a “you are there” experience as the 3s sometimes do.  Maybe that is why Roy Allison was so interested in the sound of Chicago Symphony Hall. 

On 5/1/2018 at 7:15 PM, genek said:

I'll take Sarah Vaughn or Maria Callas over David Lee Roth or Axl Rose any day.

That takes care of the 70s and 80s. 

Adams