Jump to content

Roy Allison interview in 1992 issue of The Audio Critic, David Ranada was the interviewer


Zilch

Recommended Posts

The Toole/Olive work demonstrates listener preference rating of the speakers under evaluation to be independent of the room, indicating that we "hear through" it, whereas common wisdom instead attributes prime, indeed, dominant, significance to the room. It's the "shopping mall" thing: perceived spectral balance (and timbre) do not change with the space; we know who's talking to us -- mom is easily identifiable from anywhere in the grocery store, the parking lot, or inside the car berating us for wandering off. Outdoors or in, a clarinet is a clarinet, and I doubt anyone familiar with the differences between good and bad ones requires a particular space in which to make that determination.

In "small" rooms, typical of home listening spaces, studies define the variables more precisely, and indeed find that cues for localization, spectral balance, and spaciousness are functions of delay. Linkwitz says 6ms, Geddes says 10ms, Kantor and Toole, 20 ms, and Allison, 30 ms. It's not brick walls, rather, different psychoacoustic continua for the three characteristics. Geddes cuts to the bottom line and keys on the conclusion common among all of these: very early reflections (VER) are inherently detrimental -- avoid them, design speakers and deployments which do not generate them, and gain the benefits of taking control AWAY from the room in the region of frequencies above the modal transition zone. That accomplished, the listener may adjust the balance between imaging and spaciousness virtually independent of the listening space, according to taste, without affecting the spectral balance.... :D

"The Toole/Olive work demonstrates listener preference rating of the speakers under evaluation to be independent of the room, indicating that we "hear through" it, whereas common wisdom instead attributes prime, indeed, dominant, significance to the room."

Zilch, the B&K technical paper Ken cited, the middle one which tested 5 loudspeaker systems in 3 rooms came to exactly the opposite conclusion, that the preference was strongly influenced by the room. So was the steady state FR measured with pink noise. There was strong correlation. (This paper was presented shortly before introduction of inexpensive graphic equalizers to consumers which changed everything, I'm sure much to the chagrin of many high end equipment manufacturers who worked hard to steer the cult culture away from them.) However, the context of the test is extremely limited and did not take into account many other variables among the speakers. Its conclusions therefore are not valid on the face of it owing to omission of those variables and the small sample size used.

But the entire study like Toole's brings up an entirely different issue. What is the purpose of these sysems, to please and entertain subject to the capricious tastes of the market or to accurately recreate? To those who study music itself as opposed to being preoccupied with recordings of music those are usually one and the same but to others they are not. There is a big difference here. As an analogy I used to like Fujichrome film in preference to Kodak Ectachrome. Fujichrome not only had better greens but color was more saturated. But it was not as accurate. I also like jukeboxes. The fascination of the mechanical selection of 45 RPM records, the light show, even better than watching a turntable spin or vacuum tubes glow. But it's not high fidelity and worthless for any serious listening. I feel much the same about audiophile equipment I hear. It's always a poor second best to the real thing. More like looking at a painting of a beach and imagining the experience than actually being there. To me that is what "hearing through the equipment and room" really means.

Link to comment
Share on other sites

  • Replies 83
  • Created
  • Last Reply
Yes, I probably do. Again an omni mic. has no proximity effect. You can bring it up to near contact and see no effect at LF. I've done it many times.

They offer no evidence other than Moller prefers a system that he personally designed, which followed the curve. As always we are confusing steady state measurements made with non-directional microphones with the perceived response from human hearing and processing in a live room. As you move to a larger and larger room the steady state measured response of a system tends to have more and more extreme rolloff at HF yet it may still sound flat. Numerous papers, that I have referenced in the past, all point to the notion that the room has little effect above 200Hz in the perceived frequency response. Measurements don't automatically equal perception.

Toole writes about the variable of the room vs. the variable of the speaker. He finds that when speakers are compared within a room the room effect is stationary and evaluations are ranked based on the speaker performance. He did find that the room could become a variable, but only when a system was binaurally recorded in a number of rooms and the rooms could be A-B ed with instantaneous switching. In other words the ear got used to the room it was in.

So you are saying if I understand you correctly that you think Atkinson's measurements of AR303 are correct and that decades after Villchur, Kloss, and Allison, Ken Kantor took an improved version of their woofer and put it in an enclosure that gave a 5 db peak at 60 hz due to underdamping which if compensated for by a gentle 6db per ocatve filter would yield an F3 substantially higher than the original design, has a sudden step discontinuity in its midrange, poorer high frequency dispersion, still has a high end rolloff, and this is what Atkinson (and you?) would call evolution and progress.

"The wide baffle does make its presence known, however: despite the use of a small-diameter tweeter dome, the top two octaves fall off more rapidly with off-axis angle than would be the case with a minimonitor. In a typical room, this will contribute to the speaker's rather mellow tonal balance."

AR3a had a wide baffle and had the widest dispersion of any single tweeter speaker I've ever heard of, maybe wider than any other speaker ever made except for AR2ax and AR5 that used the same tweeter. Wide uniform dispersion and mellow tonal balance due to hf rolloff are not mutually exclusive. In fact the rolloff is easily compensated for by equalization, non uniform dispersion (beaming where dispersion is a strong function of frequency) is a whole 'nother kettle of fish.

"Note that fig.3 reveals that, in the midrange, the 303 also has rather more limited dispersion than a typical two-way design. This is probably due both to the crossover performance and to the use of a 12" LF driver."

How can the use of a 12" woofer limit the dispersion of a midrange driver (unless he is talking about the upper bass)? I swear some people can write anything and get away with it.

"Despite its superficial resemblance to the vintage AR-3a, however, the AR 303 is undoubtedly a modern speaker, with no subjective or objective failings that could be laid at the feet of its heritage—other than its dulled tonal balance."

So in all of these ways, the design deviates from the original intent of its inventors, costs 3 times as much and is "an improvement."

I am not asking you to defend Atkinson's review, just wondering if you agree with it. I don't suppose Ken would comment on whether he thinks the measurement or evaluation is accurate or fair. Or on the other hand whether the test methodology was flawed or the speakers a defective sample.

BTW, on the other subject you commented on, it is my experience that the tonal balance of the first arrival strongly influences the perceived FR but the relative rate of hf falloff as a function of time in very large rooms serves to alter perceptions of tonality and space. This is why the steady state FR measurement of concert halls for rationalizing loudspeaker FR is flawed not reflecting the way people actually hear. Every speaker with a rolled off high end sounds dull to my ears (like AR3/AR3a) unless some other factor such as the program material or system equalization compensates for it. Another way to look at is is that the tonality perceived in a concert hall and the reverberation are intrinsically bound up with each other, different subjective aspects of the same physical phenomena. In order to accurately reproduce one, you must consequently reproduce the other.

Link to comment
Share on other sites

AR was not just good, it was great, and Villchur's handling of the company was great, too, with a brilliant and honest ad campaign and generous employee benefits and assurances that those benefits would continue when he sold the company to Teledyne in 1967. However, there were weak areas related to room/speaker interactions (Allison even noted some of them when he was at AR, as evidenced by his 1970 soundfield paper, and later papers on the existence of mid-bass suckout artifacts), and the later Allison lineup managed to smooth (literally) things out a bit, mainly in the mid-bass frequency range. From the get go, Allison also used liquid cooling in his drivers (company built, by the way, and not adapted from assorted OEM versions), and his outboard ESW bass equalizer allowed his speakers to get down lower in frequency (flat to 20 Hz with all but the smallest model) than any competing Altec, JBL, or Klipsch model, period, or even any AR model short of the AR-9.

So, yes, while AR was a champion-grade company, Allison did move things forward. Indeed, when his company began (with several members of the previous AR staff also joining) many individuals considered Allison Acoustics as and example of the "classic" AR company under a different name.

-_- zzzzzz

I suggest that you and Geddes set yourselves up in a good room, obtain several different pairs of good speakers (including any Geddes models you care to use), obtain some good, well-engineered recordings that work to simulate a live-space environment, and do some careful, level-matched A/BA comparisons.

I suggest that YOU get a pair of Geddes loudspeakers, deploy them as recommended in your perfect listening room, and review them.... :D

Link to comment
Share on other sites

So you are saying if I understand you correctly that you think Atkinson's measurements of AR303 are correct and that decades after Villchur, Kloss, and Allison, Ken Kantor took an improved version of their woofer and put it in an enclosure that gave a 5 db peak at 60 hz due to underdamping which if compensated for by a gentle 6db per ocatve filter would yield an F3 substantially higher than the original design, has a sudden step discontinuity in its midrange, poorer high frequency dispersion, still has a high end rolloff, and this is what Atkinson (and you?) would call evolution and progress.

The impedance curve shows that the system resonance is a little below 40Hz. The bump near 60 looks to me like some interaction between the woofer response and the dividing network. This sometimes happens when a high inductor value is in series with the woofer.

"The wide baffle does make its presence known, however: despite the use of a small-diameter tweeter dome, the top two octaves fall off more rapidly with off-axis angle than would be the case with a minimonitor. In a typical room, this will contribute to the speaker's rather mellow tonal balance."

AR3a had a wide baffle and had the widest dispersion of any single tweeter speaker I've ever heard of, maybe wider than any other speaker ever made except for AR2ax and AR5 that used the same tweeter. Wide uniform dispersion and mellow tonal balance due to hf rolloff are not mutually exclusive. In fact the rolloff is easily compensated for by equalization, non uniform dispersion (beaming where dispersion is a strong function of frequency) is a whole 'nother kettle of fish.

"Note that fig.3 reveals that, in the midrange, the 303 also has rather more limited dispersion than a typical two-way design. This is probably due both to the crossover performance and to the use of a 12" LF driver."

How can the use of a 12" woofer limit the dispersion of a midrange driver (unless he is talking about the upper bass)? I swear some people can write anything and get away with it.

I agree that the dispersion vs. baffle width comment doesn't make sense. The cabinet width will change the 4pi to 2pi transition frequency. Outside of those frequencies (below the frequency of transition for a bigger speaker and above the range for a smaller speaker) the driver diameter would be the primary factor. A wide dispersion device can approach 180 dispersion as limited by the baffle, if above the 2pi transistion. I also agree that dispersion and frequency balance are two different items.

His lateral difference curves show a little bit of fall off around 700 Hz which could be both woofer directivity and crossover effect (even with vertical drivers you get a little bit of differential phase shift as you move well off axis).

"Despite its superficial resemblance to the vintage AR-3a, however, the AR 303 is undoubtedly a modern speaker, with no subjective or objective failings that could be laid at the feet of its heritage—other than its dulled tonal balance."

So in all of these ways, the design deviates from the original intent of its inventors, costs 3 times as much and is "an improvement."

I am not asking you to defend Atkinson's review, just wondering if you agree with it. I don't suppose Ken would comment on whether he thinks the measurement or evaluation is accurate or fair. Or on the other hand whether the test methodology was flawed or the speakers a defective sample.

I've attached the frequency response of the 303 and the nearest AR3a response curve from the Allison paper. The AR3a curve is the anechoic response, including the edge reflection and diffraction problems. You questioned earlier if they might add even 1dB of response variation? Since we know the individual drivers are quite flat under ideal conditions all these other response "wiggles" are reflections or crossover effect. To be fair the 303 curve has a nearfield or 2pi LF (below 300) and some spatial averaging, although we can see from the the off axis curve plots that the lateral dispersion is very well behaved, so the averaging wouldn't have done much. At the same time the vertical dispersion looks good except for the inevitable x'over dips. It looks to me that Ken has made a nod towards the frequency response of the AR3a in that he has slightly rolled off the top end balance. Overall the response is very smooth and driver integration is commendable. As to 3 times the price, look up the inflation figures for the respective years.

BTW, on the other subject you commented on, it is my experience that the tonal balance of the first arrival strongly influences the perceived FR but the relative rate of hf falloff as a function of time in very large rooms serves to alter perceptions of tonality and space. This is why the steady state FR measurement of concert halls for rationalizing loudspeaker FR is flawed not reflecting the way people actually hear. Every speaker with a rolled off high end sounds dull to my ears (like AR3/AR3a) unless some other factor such as the program material or system equalization compensates for it. Another way to look at is is that the tonality perceived in a concert hall and the reverberation are intrinsically bound up with each other, different subjective aspects of the same physical phenomena. In order to accurately reproduce one, you must consequently reproduce the other.

Glad we agree about the importance of the direct and early sound. I also agree that modifying the brightness of the reverberent field (increasing HF RT) is likely audible in concert halls. As an asside, I am in the midst of research on cinema EQ and there is much evidence that the "X" curve traditionally used is an artifice that allows the direct field to be flat, when the steady state curve is greatly rolled off. That the response should measure rolled off, at distance, does not mean the the direct field should be intentionally rolled off.

David

post-102584-1276039200.jpg

post-102584-1276039211.jpg

Link to comment
Share on other sites

Attached is a copy of an early AR-3a data booklet that has some curves and commentary material that might interest both of you.

Howard Ferstler

Note that figure 3 and 4 from your brochure are both measured with units infinite baffle mounted: "grille cloth and decorative molding removed". The response curve I attached above shows the actual response of the system, as sold.

David

Link to comment
Share on other sites

Of course, one reason for this is my rather good main listening room, but given the kind of speaker prices we are talking about (the Cantatas and Allison IC-20s were priced similarly at over five grand per pair) most potential customers would be in a financial position to have that kind of listening room.

And us peasants? :D

Note that figure 3 and 4 from your brochure are both measured with units infinite baffle mounted: "grille cloth and decorative molding removed". The response curve I attached above shows the actual response of the system, as sold.

And recently verified via measurements of vintage AR3a's at ZilchLab. -_-

Attached below are the normalized horizontal polar response plots for my "SpitWad" project updating The Smaller Advent speakers with waveguide high frequency over on PE's Tech Talk forum, first at the same resolution provided in the AR3a brochure above using the closest frequencies available in CLIO, and then, with higher resolution.

Also posted in that thread is a graphic illustration of the adverse influence of that vintage product's 1" edge overhang upon the frequency response. The saga begins here:

http://techtalk.parts-express.com/showthre...899#post1647899

As may be seen in the polars, I have come very close to replicating the dispersion characteristics of AR3a through +/- 45° as published by AR in their brochure using an inexpensive modern waveguide. Once the project is complete, I will post the full directivity through +/- 90° for comparison to actual AR3a performance measured under the same conditions....

post-102716-1276052428.jpg

post-102716-1276053518.jpg

Link to comment
Share on other sites

So you are saying if I understand you correctly that you think Atkinson's measurements of AR303 are correct and that decades after Villchur, Kloss, and Allison, Ken Kantor took an improved version of their woofer and put it in an enclosure that gave a 5 db peak at 60 hz due to underdamping which if compensated for by a gentle 6db per ocatve filter would yield an F3 substantially higher than the original design, has a sudden step discontinuity in its midrange, poorer high frequency dispersion, still has a high end rolloff, and this is what Atkinson (and you?) would call evolution and progress.

"The wide baffle does make its presence known, however: despite the use of a small-diameter tweeter dome, the top two octaves fall off more rapidly with off-axis angle than would be the case with a minimonitor. In a typical room, this will contribute to the speaker's rather mellow tonal balance."

AR3a had a wide baffle and had the widest dispersion of any single tweeter speaker I've ever heard of, maybe wider than any other speaker ever made except for AR2ax and AR5 that used the same tweeter. Wide uniform dispersion and mellow tonal balance due to hf rolloff are not mutually exclusive. In fact the rolloff is easily compensated for by equalization, non uniform dispersion (beaming where dispersion is a strong function of frequency) is a whole 'nother kettle of fish.

"Note that fig.3 reveals that, in the midrange, the 303 also has rather more limited dispersion than a typical two-way design. This is probably due both to the crossover performance and to the use of a 12" LF driver."

How can the use of a 12" woofer limit the dispersion of a midrange driver (unless he is talking about the upper bass)? I swear some people can write anything and get away with it.

"Despite its superficial resemblance to the vintage AR-3a, however, the AR 303 is undoubtedly a modern speaker, with no subjective or objective failings that could be laid at the feet of its heritage—other than its dulled tonal balance."

So in all of these ways, the design deviates from the original intent of its inventors, costs 3 times as much and is "an improvement."

I am not asking you to defend Atkinson's review, just wondering if you agree with it. I don't suppose Ken would comment on whether he thinks the measurement or evaluation is accurate or fair. Or on the other hand whether the test methodology was flawed or the speakers a defective sample.

BTW, on the other subject you commented on, it is my experience that the tonal balance of the first arrival strongly influences the perceived FR but the relative rate of hf falloff as a function of time in very large rooms serves to alter perceptions of tonality and space. This is why the steady state FR measurement of concert halls for rationalizing loudspeaker FR is flawed not reflecting the way people actually hear. Every speaker with a rolled off high end sounds dull to my ears (like AR3/AR3a) unless some other factor such as the program material or system equalization compensates for it. Another way to look at is is that the tonality perceived in a concert hall and the reverberation are intrinsically bound up with each other, different subjective aspects of the same physical phenomena. In order to accurately reproduce one, you must consequently reproduce the other.

I was pretty happy with the Stereophile review of the 303, (despite wishing for a bit less ambivalence, either way). I certainly have no issues whatsoever with their measurement data.

The 303 was a carefully designed system, using full-custom drivers. During its development, hundreds of measurement curves were taken, optimized and traded-off. If the speaker game was as easy as making any one or two curves, (single or averaged), look better, this would be extremely easy to do with modern electronics.

But this is not the game. In fact, the reality is that speaker designers must consider and balance a huge amount of technical data in an attempt to optimize a product. Similarly, reviewers who attempt to introduce objective criteria into their loudspeaker reviews are faced with the problem of selecting presentations that are in some way useful to the reader and, over the long haul, compatible with their own listening evaluations.

-k

Link to comment
Share on other sites

Note that figure 3 and 4 from your brochure are both measured with units infinite baffle mounted: "grille cloth and decorative molding removed". The response curve I attached above shows the actual response of the system, as sold.

David

David,

You've opened up a 'can of worms' here that Tom and I went round and round on a while ago.

In this same paper, Allison shows a 2-pi curve of the 3a--with decorative molding and all the warts--and the curve shows the expected lift in the bottom end that one would expect from a 2-pi environment. As a matter of fact, the entire reason for Allison showing the 4-pi curve you reproduced was to show the effect of 4-pi on the woofer's response.

But, the thing Tom and I went around on was the fact that the 2-pi measurement showed considerably more upper-mid/HF energy that this 4-pi curve--even though they were identical speakers, both measured with the grille molding, through the x-o. But at those high frequencies, the 3a's baffle is large enough to be "2-pi" in the upper mid/HF region. So why the discrepancy? The 2-pi and 4-pi curves should have been identical above about, say, 8000, where the wavelength is less than 2 inches. The 3a's baffle is "2-pi" in either case above 8k, so the HF curves should be the same, no? But they're not.

I'll try to dig out the 2-pi curve (I have the paper somewhere), but if you have it, look and you'll see what I'm talking about.

Thoughts?

Steve F.

Link to comment
Share on other sites

I have it from a good source that any "vintage" AR-3a will have tweeters that are simply not up to spec. The internal foam damping material will have crumbled and the surround binders will have gotten brittle, too. As a result, no comparison of the speaker with any modern designs will be valid, and any current measurements by you or anybody else will be invalid, too, at least over the treble range. The midrange driver may also have deteriorated over time, so measurements and impressions over that range will be suspect, too.

I don't know why you're so anxious for every vintage AR speaker to be a crumbling wreck. My own "listening impressions" of the AR-2ax's that I've owned since I bought them new in 1975 is that their sound has not noticeably changed in that time, and my "impression" when I fired up the 3a's I acquired a couple of years ago was that they sounded exactly as I remembered 3a's sounding 40 years ago and expected them to sound (i.e., better than anything new I've heard in any demo room within a reasonable drive of where I live now).

What I "suspect" is that many people who acquire ARs for the first time as vintage speakers fall back on the deterioration theory simply because they are unable to reconcile what their ears expect a "good" loudspeaker to sound like (the ones that are being touted as "high end" by today's audiophiles) with their reaction to the signature sound of vintage ARs. They are what they are, either people like them as they are or they don't, and if they don't, then WTF are they doing wasting their lives hanging out on a site like CSP?

Link to comment
Share on other sites

I have done a lot of speaker reviewing and my take is that it is nearly impossible to draw conclusions about speaker sound when moving from room within minutes, let alone after forty years.

It's not difficult at all if you're not obsessed with measurements or proving anything "superior" to or "more accurate" than anything else. You connect the speakers to your sound system, arrange them in your listening space and play music. You either like what you hear or you don't.

Zilch's measurements of the 3a's in his possession indicate that the performance of their drivers is still consistent with AR's original published curves for them. Someone who believes that the signature AR sound is not "good" or "right" will certainly use this to bolster an argument that it never was, but for those of us who like the sound we're getting from our speakers, it should be encouraging news.

Link to comment
Share on other sites

Attached is a photo of a dissected 3a tweeter I took a few years ago. I have dissected over a dozen dead 2ax and 3a tweeters over the past few years, and every one has had dried out foam under the domes. In some cases, however, the foam was still in the original shape of the dome until touched, at which point it crumbled to dust. I have seen a number of deteriorated foam suspensions as well, usually where there has been a break in the original butyl rubber coating.

Having performed many restorations of ARs with this type of tweeter, the most obvious observation I can share is inconsistent high range behavior. Some sound good, others not so much. Gene's are apparently still acceptable, and Zilch's measurements suggest his pair of 3as are performing similarly to originally published measurements. A big problem, therefore, for folks just getting into these old timers, is the unknown condition of the tweeter. BTW, I have not seen similar variations in the midrange drivers. They have been remarkably stable and consistent over the years. (There is no foam in the construction of the mid.)

When we were putting together the AR restoration document, we sent Ken K. a number of tweeters to test in an effort to develop an appropriate crossover modification for the AB Tech replacement tweeter. At that time, Ken compared a number of original, used 3a tweeter specimens to Tom T's NOS tweeters, and there was a measurable difference between them. Two of mine died during testing.

Roy

post-101150-1276127079.jpg post-101150-1276127047.jpg

Link to comment
Share on other sites

Zilch's measurements of the 3a's in his possession indicate that the performance of their drivers is still consistent with AR's original published curves for them. Someone who believes that the signature AR sound is not "good" or "right" will certainly use this to bolster an argument that it never was, but for those of us who like the sound we're getting from our speakers, it should be encouraging news.

One is "better" than the other, much as Dave found with AR4x, but both are surprisingly in conformity with the original measurements in the Allison/Berkovitz paper. The data is posted on AudioKarma, where we went through all of this, including constant directivity as a distinguishing feature of AR3a, over a year ago. :rolleyes:

It is nice to have classic speakers on hand, and I would love to have a pair of AR-3 or 3a systems on hand to mess with. Even AR-2a or 2ax models would be nice. However, if I were going to listen to them seriously I would be on the hunt for some replacement tweeters at the very least (and maybe a replacement midrange for the 2ax), and I would be inclined to upgrade their crossover networks to second-order-slope status to maybe protect the drivers better and reduce interference effects that, while probably not audible in most rooms, would still get on my nerves.

Audible only in the very best of rooms and only when heard by the most golden of ears, no doubt. :P

The Allison tweeter employed in each uses long-life cotton batting under the dome for resonance control and a non-deteriorating foam damping ring at the perimeter, and the dome midrange uses similar foam in the oversized surround, with no batting at all under the dome.

But, of COURSE! :D

Link to comment
Share on other sites

What I "suspect" is that many people who acquire ARs for the first time as vintage speakers fall back on the deterioration theory simply because they are unable to reconcile what their ears expect a "good" loudspeaker to sound like (the ones that are being touted as "high end" by today's audiophiles) with their reaction to the signature sound of vintage ARs. They are what they are, either people like them as they are or they don't, and if they don't, then WTF are they doing wasting their lives hanging out on a site like CSP?

:rolleyes::P

Link to comment
Share on other sites

Attached is a photo of a dissected 3a tweeter I took a few years ago. I have dissected over a dozen dead 2ax and 3a tweeters over the past few years, and every one has had dried out foam under the domes. In some cases, however, the foam was still in the original shape of the dome until touched, at which point it crumbled to dust. I have seen a number of deteriorated foam suspensions as well, usually where there has been a break in the original butyl rubber coating.

Roy

Interesting photos. Old foam of all types can present a problem. (My piano hasn't had the felt changed for 50 years and its all as good as new.) I would think that if the under foam hadn't crumbled it would still fulfill its purpose. Plugs behind tweeters are much like stuffing in a bass cabinet: they should damp any resonances and possibly lower resonance slightly. The reflection type resonances are quite high in frequency and I've seen a number of tweeters that omit plugs under the dome and don't seem any worse for it. The 3 suspension spots are another matter. If they stiffen up then resonance can climb considerably.

I measured an AR3 a few months back and at first glance I thought the mids were dead. They had stiffened up to the point where the resonance was way up and they were missing half their range.

Don't I recall that some people were able to replace the suspension dots?

David

Link to comment
Share on other sites

Having performed many restorations of ARs with this type of tweeter, the most obvious observation I can share is inconsistent high range behavior. Some sound good, others not so much. Gene's are apparently still acceptable, and Zilch's measurements suggest his pair of 3as are performing similarly to originally published measurements. A big problem, therefore, for folks just getting into these old timers, is the unknown condition of the tweeter.

It would be interesting to know whether the condition of old tweeters is totally age-dependent, or if treatment enters into it. My 2ax's have never been subjected to temperature or humidity extremes, and even at my youngest and dumbest I never tried to use them as PA speakers or do stupid woofer tricks with them. OTOH, I know nothing about the history of my 3a's other than that they were the property of a "retired engineer" who had some very odd ideas about how to deal with corroded pots, and came with one dead tweeter that had to be replaced.

Link to comment
Share on other sites

Attached is a polar curve Roy Allison ran on an IC-20 from 63 Hz on up to 16 kHz. Here, we are talking about seriously wide dispersion. The graph stops at 180 degrees (90/90), but the response is actually very wide out to a full 270 degrees, at least up to 10 kHz. Consequently, a huge amount of energy is bouncing off of the front wall with very little in the way of delay.

I know you hate that kind of thing, but I love it.

Howard Ferstler

Wow, constant directivity! Maybe it is a good speaker!

I'm a little surprised there isn't more HF lobing, although I assume it is a full octave of noise per curve which would give some averaging.

Still, fairly uniform.

David

Link to comment
Share on other sites

David,

You've opened up a 'can of worms' here that Tom and I went round and round on a while ago.

In this same paper, Allison shows a 2-pi curve of the 3a--with decorative molding and all the warts--and the curve shows the expected lift in the bottom end that one would expect from a 2-pi environment. As a matter of fact, the entire reason for Allison showing the 4-pi curve you reproduced was to show the effect of 4-pi on the woofer's response.

But, the thing Tom and I went around on was the fact that the 2-pi measurement showed considerably more upper-mid/HF energy that this 4-pi curve--even though they were identical speakers, both measured with the grille molding, through the x-o. But at those high frequencies, the 3a's baffle is large enough to be "2-pi" in the upper mid/HF region. So why the discrepancy? The 2-pi and 4-pi curves should have been identical above about, say, 8000, where the wavelength is less than 2 inches. The 3a's baffle is "2-pi" in either case above 8k, so the HF curves should be the same, no? But they're not.

I'll try to dig out the 2-pi curve (I have the paper somewhere), but if you have it, look and you'll see what I'm talking about.

Thoughts?

Steve F.

I'm not seeing any particular discrepency. I'm looking at figure 5 and figure 9 of the AES paper. It appears that there is a 10dB shift in chart level. Then the differences in the 4 pi curve relative to the 2 pi are: 4 to 5 dB less at 100, but a couple dB more at 500 (very typical of 2 pi to 4 pi change), upper mid response is much the same except 4 pi is a little rougher. The tweeter response is similar except, again, a little rougher in 4 pi (I read this as a greater dip at 8k and 15k rather than a treble level shift). Note that figure 5 has the picture frame lip added but the system is flush, from the picture frame level, with the wall. This means that the internal reflections are present but the external edge reflection is not present, hence the generally smoother curve.

The previous curve #4 is similar to the ones in the brochure that Howard submitted yesterday. Drivers and network as stock but mounted on a flush infinite baffle, so they are generally smoother as explained in the text.

This is a very instructional look at how the raw sections of a multiway system add up to the final 4pi response.

Note that these curves are with the mid and tweeter at max, rather than at recommended positions. You can picture what the mid prominence at 400 - 500Hz in 4 pi would be with the pots at nominal. I seriously wonder if this is part of the reason for the inductor increase that came later.

Of course, none of this is audible unless you are one of those direct field hoodlums that listens to non-serious music.

David

Link to comment
Share on other sites

I would think that if the under foam hadn't crumbled it would still fulfill its purpose.

David

Agreed...I suspect it is foam that has crumbled into the voice coil gap that may be causing some variability (along with any changes in suspension compliance).

Roy

Link to comment
Share on other sites

Don't I recall that some people were able to replace the suspension dots?

Last year I sent the dead tweeter that came off one of my 3a's to someone who wanted to try to graft it onto a 2ax tweeter with a damaged dome. He said he was able to make it work, but I never heard the result and can't judge how well.

Link to comment
Share on other sites

Attached is a polar curve Roy Allison ran on an IC-20 from 63 Hz on up to 16 kHz. Here, we are talking about seriously wide dispersion. The graph stops at 180 degrees (90/90), but the response is actually very wide out to a full 270 degrees, at least up to 10 kHz. Consequently, a huge amount of energy is bouncing off of the front wall with very little in the way of delay.

There's an obvious broadband (up to 10 dB) suckout on axis, but since the curves are neither labeled nor in color (which I know you hate), I can only guess it's in the VHF (where there's no music :rolleyes:).

No matter, it's obvious that a pair of divergent-mounted SpitWad waveguides could do as well, IF we desired to spray the room with response-anomaly-inducing energy.

From this perspective, we might all be thankful that Villchur's classic ARs failed to meet this ill-conceived and long since repudiated wide-dispersion design/performance objective.... :P

Link to comment
Share on other sites

I'm not seeing any particular discrepency. I'm looking at figure 5 and figure 9 of the AES paper. It appears that there is a 10dB shift in chart level. Then the differences in the 4 pi curve relative to the 2 pi are: 4 to 5 dB less at 100, but a couple dB more at 500 (very typical of 2 pi to 4 pi change), upper mid response is much the same except 4 pi is a little rougher. The tweeter response is similar except, again, a little rougher in 4 pi (I read this as a greater dip at 8k and 15k rather than a treble level shift). Note that figure 5 has the picture frame lip added but the system is flush, from the picture frame level, with the wall. This means that the internal reflections are present but the external edge reflection is not present, hence the generally smoother curve.

The previous curve #4 is similar to the ones in the brochure that Howard submitted yesterday. Drivers and network as stock but mounted on a flush infinite baffle, so they are generally smoother as explained in the text.

This is a very instructional look at how the raw sections of a multiway system add up to the final 4pi response.

Note that these curves are with the mid and tweeter at max, rather than at recommended positions. You can picture what the mid prominence at 400 - 500Hz in 4 pi would be with the pots at nominal. I seriously wonder if this is part of the reason for the inductor increase that came later.

Of course, none of this is audible unless you are one of those direct field hoodlums that listens to non-serious music.

David

Yes, those are the two figures I'm referring to, 5 and 9. I superimposed the two on top of each other, then I "normalized" them between 400-1000 Hz (the midrange "perceived loudness" frequency region), and darkened the 4 pi fig 9 curve for clarity.

To my eyes, it looks as though the 4pi curve's treble region is a solid 4-5 dB lower than the 2 Pi curve--even though the 3a's baffle is providing a "2 pi" environment in both cases.

Both you and Tom say there's "no particular discrepancy." Maybe I'm just thick or stubborn or both. But a 5 dB lower treble region when the midrange regions are normalized to the same level, to me, is a big deal--and a mystery to me, if both curves are "2 pi" treble environments.

Steve F.post-100522-1276178394.jpg

Link to comment
Share on other sites

Yes, those are the two figures I'm referring to, 5 and 9. I superimposed the two on top of each other, then I "normalized" them between 400-1000 Hz (the midrange "perceived loudness" frequency region), and darkened the 4 pi fig 9 curve for clarity.

To my eyes, it looks as though the 4pi curve's treble region is a solid 4-5 dB lower than the 2 Pi curve--even though the 3a's baffle is providing a "2 pi" environment in both cases.

Both you and Tom say there's "no particular discrepancy." Maybe I'm just thick or stubborn or both. But a 5 dB lower treble region when the midrange regions are normalized to the same level, to me, is a big deal--and a mystery to me, if both curves are "2 pi" treble environments.

Steve F.post-100522-1276178394.jpg

I wouldn't normalize the way you did. Try it again with the 700-800 region equal, which also gets the 2k to 4k region equal. Then the 400 Hz region will be a net gain and the tweeter will only be lower by a little bit.

Note that this net gain in the 2 pi move to 4 pi transition is typical and is shown in the original Olson diffraction study curves. A woofer perfectly flat in 2 pi will droop downwards at low frequencies to a max of about 4 to5 dB. At the top of the transition range there is usually a bump in response. I've attached a photo from the web of what I mean and the same trend is pretty clear on the AR curves.

Why 4 to 5 dB you ask? Always two effects. First, the radiation resistance in 2 pi is double that of 4pi and the woofer puts out 3dB more power. Secondly, all the radiation that was going around behind the cabinet is constrained to the front. Even at low frequencies the woofer and cabinet have some directivity so the back energy does not equal the front energy, so you get less than another 3db, perhaps 1.5dB, for a total of 4.5

David

ps. I will agree that you are thick or stubborn or both!

post-102584-1276180886.jpg

Link to comment
Share on other sites

I wouldn't normalize the way you did. Try it again with the 700-800 region equal, which also gets the 2k to 4k region equal. Then the 400 Hz region will be a net gain and the tweeter will only be lower by a little bit.

Note that this net gain in the 2 pi move to 4 pi transition is typical and is shown in the original Olson diffraction study curves. A woofer perfectly flat in 2 pi will droop downwards at low frequencies to a max of about 4 to5 dB. At the top of the transition range there is usually a bump in response. I've attached a photo from the web of what I mean and the same trend is pretty clear on the AR curves.

Why 4 to 5 dB you ask? Always two effects. First, the radiation resistance in 2 pi is double that of 4pi and the woofer puts out 3dB more power. Secondly, all the radiation that was going around behind the cabinet is constrained to the front. Even at low frequencies the woofer and cabinet have some directivity so the back energy does not equal the front energy, so you get less than another 3db, perhaps 1.5dB, for a total of 4.5

David

ps. I will agree that you are thick or stubborn or both!

Thanks, Dave.

From figures 5 and 9, it would appear that a 3a mounted flush in a bookshelf (5), surrounded by books roughly flush with the front of the speaker's cab (the classic 'bookshelf mounting') will sound a bit brighter in the treble and fuller in the mid bass than a 3a mounted out into the room on an 'x' stand (similar to figure 9).

Regarding the charges of 'thick' and 'stubborn,' I plead guilty to both--probably more 's' than 't,' but a bit of both, to be sure!

Steve F.

Link to comment
Share on other sites

Thanks, Dave.

From figures 5 and 9, it would appear that a 3a mounted flush in a bookshelf (5), surrounded by books roughly flush with the front of the speaker's cab (the classic 'bookshelf mounting') will sound a bit brighter in the treble and fuller in the mid bass than a 3a mounted out into the room on an 'x' stand (similar to figure 9).

Regarding the charges of 'thick' and 'stubborn,' I plead guilty to both--probably more 's' than 't,' but a bit of both, to be sure!

Steve F.

Yes, assuming the bookshelves are fairly full then you would be getting back to the 2 pi (#9) curve, at least at LF. I doubt that you would have a smooth enough surface around the picture frame to get the HF curve back, at least entirely.

If you re-normalize as I suggest, it looks like there is still a 2-3dB discrepency in general tweeter level. Since reflection effects are largely to blame, it will shift around with small observer position changes. (Doesn't mean it doesn't matter!)

David

Link to comment
Share on other sites

Yes, assuming the bookshelves are fairly full then you would be getting back to the 2 pi (#9) curve, at least at LF. I doubt that you would have a smooth enough surface around the picture frame to get the HF curve back, at least entirely.

If you re-normalize as I suggest, it looks like there is still a 2-3dB discrepancy in general tweeter level. Since reflection effects are largely to blame, it will shift around with small observer position changes. (Doesn't mean it doesn't matter!)

David

2 pi #5 curve, right? Not #9. Just want to be clear......

Steve F.

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.


×
×
  • Create New...