AR3a improved Xover capacitors

[soundminded]

I've seen a lot of loudspeaker measurements but don't remember any of them looking similar enough that the speakers wouldn't be easily distinguishable. Speakers of different designs, and sometimes even mass produced speakers of the same design, just don't measure the same. Even if you could find two different systems that measure similarly on axis, their off axis curves would likely be highly different.

Of course, loudspeaker marketers, unsavory characters that they are, might be embelishing the curves to make them look more ideal. Are we talking reviewers curves or advertising curves? Do you have some examples? Please show us this textbook perfect speaker.

David

Since I asked the question, I will not attempt to answer it, only to rephrase it and elaborate on it so that it is clearer.

Going back as far as AR3, whether the published curves were the manufacturer's own or appeared in hobbyist magazines that had their own in house testing such as Audo Magazine, farmed their tests out to independent laboratories like Hirsch-Houck Labs (Stereo Review Magazine), CBS Laboratories (High Fidelity Magazine), magazine of today (Stereophile Magazine) we see many FR curves which appear nearly flat on axis with minor deviations often so random and insignificant as to be of seemingly no importance to the end user. In extensive listening comparisons between KLH Model 17 and AR3, while AR3 had clearly stronger deeper bass capability, KLH Model 17 was consistently brighter, clearer on all program material with tone controls and speaker driver controls set at their flat position. This even though AR3 had superior high frequency dispersion and more extended output at high frequencies than KLH Model 17. The same turned out to be true for KLH Model 6 and later with AR3a. The surprise in the LVR demo with AR3 was that it was even brighter than the live guitar. This evidently at least in part was the result of a treble boost on the PAS3x preamplifier altering the speaker's FR.

Many other speakers had similar published curves. Speakers today routinely exhibit poor HF dispersion, evidently intentionally to cater to market preferences for pinpoint imaging yet having nearly flat on axis response. These often demonstrate shrillness that was rarely if ever found in speakers of earlier eras of the 1960s. These speakers often use tweeters of similar design and very similar measured performance based on publications both by the manufacturers themselves and by independent labs.

I've read reports that depending on room acoustics, speaker placement, and listener location low bass response can vary 40 db or more for the same speaker in the same room at some frequencies. My own experience suggests that this is probably true.

So, in an industry whose most expensive products are justified by being based to a large degree on the claimed ability to hear differences not only in loudspeakers but amplifiers that test very similarly, speaker wires, interconnect wires, vacuum tubes of the same type, capacitors, resistors, why wouldn't different speakers that measure very similarly sound different?

You being a speaker design "professional" along with Ken Kantor, I reitterate my queston. What do you see in the performance of published test data that allows you to evaluate the relative merits of one speaker versus another that we amateurs don't see? Obviously we are missing something that is critical since there often appears to be little or no correlaton between performace specifications or data and subjectively perceived performance in use.

[kkantor]

....

You being a speaker design "professional" along with Ken Kantor, I reitterate my queston. What do you see in the performance of published test data that allows you to evaluate the relative merits of one speaker versus another that we amateurs don't see? Obviously we are missing something that is critical since there often appears to be little or no correlaton between performace specifications or data and subjectively perceived performance in use.

I think your question is both rhetorical and a bit of a straw man. But, I will fall for the bait, in lieu of going into the office too soon on Sunday.

1- I don't remember either David or me proposing to evaluate the, "relative merits," of any speaker based on "published test data." At most, it has been discussed that perhaps some clues about design decisions might be found by looking at all the available information. That seems reasonable.

2- I would classify individual measurements as design tools, not design goals. There are exceptions, but they are for specialized applications.

3- Pro's rely on dozens-to-hundreds of measurements to help guide each product. (Actually, thousands, but not that many are kept.) What is published by a manufacturer or a reviewer is, at best, illustrative of key issues.

4- I can't remember ever publishing response graphs for any of my system designs, (as opposed to drivers). At most, I would list things like operating limits and response window, and I would state the measurement conditions. Am I forgetting something?

5- This is old news. Julian Hirsch, by far the most influential and widely read audio product reviewer in history, raised many eyebrows when he explicitly decided to stop publishing loudspeaker response curves and editorialized that they were misleading.

6- Personally, I like glancing at the ones in Stereophile, but don't really judge a product based on them.

-k

PS- It is very difficult to build two loudspeakers, even of the same model, constructed on the same day, that cannot be readily distinguished by simple measurements. So, the idea that any two different products could even come close to matching one another is unrealistic.

[speaker dave]

We see many FR curves which appear nearly flat on axis with minor deviations often so random and insignificant as to be of seemingly no importance to the end user.

Why wouldn't different speakers that measure very similarly sound different?

Sorry, but I'm not buying your premise.

I'm not sure who's measurements you are looking at that show so many speakers approaching perfection. I have several of the High Fidelity anthologies with multiple curves from the CBS chamber and none of them look alike on axis. The off axis curve further differentiate one system from another. Ditto the curves from Audio Magazine, especially those tested by Don Keele. My impression with the Julian Hirsch curves was that they raised enough questions and manufacturers criticism, with their non flat response, that he gave up on printing them.

I've attached a few curves picked at random from the Stereophile web site. No danger of not being able to differentiate one system from the next here. Some of the systems you'll see there are just not well designed, but if you pick through the better ones (designed by professionals, no doubt) even they would have the odd dB or so trend in this octave or that, that would clearly be enough to devine one from the next in a careful listening test. Also clearly an order of magnitude more response error than any competently designed amplifier. Such are the challenges of loudspeaker design.

If you think that the KLH 6 and 17 were brighter and clearer than their AR counterparts I'd suggest that was evident in their curves, certainly in the curves I remember. Although the AR3a had drivers that were flat in their individual passbands, when measured in 2pi with no cabinet edges or grilles, the Allison paper shows what the full system looks like anechoically measured.

If you want correlation between measured performance and subjective ranking I'd recommend the excellent work of Sean Olive as quoted in Floyd Toole's recent book (both good professionals). He compared ranking in blind tests with a number of measurable factors and found best correlation with the following factors and weightings: on-axis narrow band smoothness--31.5% of the total, Low bass extension--30.5% of the total, Narrow band deviation in the in-room curve--20.5%, and smoothness of the in-room response--17.5%.

[soundminded]

Sorry, but I'm not buying your premise.

I'm not sure who's measurements you are looking at that show so many speakers approaching perfection. I have several of the High Fidelity anthologies with multiple curves from the CBS chamber and none of them look alike on axis. The off axis curve further differentiate one system from another. Ditto the curves from Audio Magazine, especially those tested by Don Keele. My impression with the Julian Hirsch curves was that they raised enough questions and manufacturers criticism, with their non flat response, that he gave up on printing them.

I've attached a few curves picked at random from the Stereophile web site. No danger of not being able to differentiate one system from the next here. Some of the systems you'll see there are just not well designed, but if you pick through the better ones (designed by professionals, no doubt) even they would have the odd dB or so trend in this octave or that, that would clearly be enough to devine one from the next in a careful listening test. Also clearly an order of magnitude more response error than any competently designed amplifier. Such are the challenges of loudspeaker design.

If you think that the KLH 6 and 17 were brighter and clearer than their AR counterparts I'd suggest that was evident in their curves, certainly in the curves I remember. Although the AR3a had drivers that were flat in their individual passbands, when measured in 2pi with no cabinet edges or grilles, the Allison paper shows what the full system looks like anechoically measured.

If you want correlation between measured performance and subjective ranking I'd recommend the excellent work of Sean Olive as quoted in Floyd Toole's recent book (both good professionals). He compared ranking in blind tests with a number of measurable factors and found best correlation with the following factors and weightings: on-axis narrow band smoothness--31.5% of the total, Low bass extension--30.5% of the total, Narrow band deviation in the in-room curve--20.5%, and smoothness of the in-room response--17.5%.

Dave;

As you have taken up the gauntlet of explaining Ken's statements, I'll quote them again;

"On the other hand, it is hard for a professional to ignore some of the "facts" that get stated."

"Yet, I do not believe it is good thing for advertizers to promote whatever they wish, with no respect for testing."

These two statements taken together send me two messages. First that there is something of critical importance to be learned from the measurements (testing) which are among the relevant facts about particular speakers. Second, that only professionals have the knowledge to unlock what those "facts" mean. If this is not the message, I'd like Ken to explain what it does mean.

Rather that try to hunt down the countless FR curves which appeared in the three audio hobbyist magazies during the 1960s through 1980s since they are no longer readily accessible insofar as I can tell, I'll try something else but first before I let that go for now, if you can find a copy of the review of Infinity ServoStatic Ia, I think their first product, published around 1970 or 1971, you will see a response curve that is almost as flat as a solid state amplifier's and a statement from the reviewer that they had never measured a speaker with a response that flat. Of course for a number of reasons the speaker became extinct not all that long afterwards and was superceded by that company by hundreds if not over a thousand models some costing far more.

So here are some curves of speakers we can easily reference. This first one is in a current issue of Stereophile Magazine and sells for $16,000 a pair. What in your judgement distinguishes this speaker from its "facts" that would suggest it should be priced competitively in this class?

http://stereophile.com/floorloudspeakers/v...ker/index5.html

Here's a model that sells for $68,000 a pair;

http://stereophile.com/floorloudspeakers/w...ker/index6.html

Is that one your middle graph? I certainly looks like it. What makes that one worth over four times as much as the previous one?

Here's one that's a mere $22,000 a pair. Except for the 4db bump at around 100 hz, it looks almost ruler flat above its low bass falloff. Certainly better than the $68,000 pair cited above and probably the best of the three to this amateur's eyes.

http://stereophile.com/floorloudspeakers/6...vel/index5.html

I've referenced two articles in which Gordon Holt expresses his views, one published in 1992 and the other in 2007 the first of which is extremely pessimistic, the second even moreso. These are linked on the thread about the expensive interconnects on e-bay in the Kitchen. They say in effect that the high end audio industry has given up on the goal of creating audio equipment that is intended to sound like acoustical instruments. Ken said in that thread that this could be done whether as they are heard in the listener's room or in a large venue but never said how it could be done. My own opinion for what it's worth is that they've given up because the problem defeated them but refuse to accept or admit it. I say they can't do it and won't change my mind until they demonstrate otherwise but that is a different discussion.

So what do you see in these three loudspeaker measurements that you can shed light on to help us understand what differences we would expect in their actual performance in use. Also since these are engineered products, perhaps you could shed some light on the engineering goals of each one, how the design was tailored to achieve those goals, and whether or not the goals were met and to what degree or where they fell short.

[speaker dave]

Rather that try to hunt down the countless FR curves which appeared in the three audio hobbyist magazies during the 1960s through 1980s since they are no longer readily accessible insofar as I can tell, I'll try something else...

I guess you've given up on your premise that loudspeakers frequently measured identically and yet sound quite different....

Nice change of topic. Is it time for the rant that all audio designers are inept?

David

[soundminded]

I guess you've given up on your premise that loudspeakers frequently measured identically and yet sound quite different....

Nice change of topic. Is it time for the rant that all audio designers are inept?

David

As one example look at the curve for Revel Ultima Salon II and compare it to the spliced curves for the drivers for AR3a and to measurements published in the three hobbyist magazines for AR3a and you will see how remarkably close they are. Never heard the Revel speaker but the design compared to AR3a is entirely different and I'd expect them to sound entirely different.

We are still waitng for Stereophile Magazine to publish its article about AR and its review of AR3a it promised last year. I wasn't joking when I said that Stereophile lives on publishing reviews of the greatest speaker in the world of the month, also the greatest amplifier in the world of the month, and the greatest preamplifier in the world of the month...etc.

You just resent that this whole industry is dying. I knew you wouldn't come up with an explanation about the three speakers I cited. IMO as a real engineer, most hi fi speakers are not engineered at all, they are tinkered together and with few exceptions always have been. When someone gets something he likes that he thinks will sell, that's the design and let the chips fall where they may. After the rhapsodic hype for the ads, the design is reverse justified in the ad and may actually get some technical explanation. The real effort has little rhyme or reason to it other than trial and error. Where I come from, that is not engineering.

Many of the most expensive speakers are made from off the shelf parts including the drivers. Where there are differences, they were not due to inadequacy of the performance of the driver manufacturer's standard product but the result of wanting to preclude reverse engineering by hobbyists at a fraction of the cost and to monopolize the market for replacement parts. A hobbyist could build many of them probabaly for no more than 10% or 20% of the retail price if fit and finish of the cabinet exterior, what I call the window dressing that has no effect on performance is not a concern.

[Carlspeak]

I totally agree.

It's also a case of "test until failure." Rather than defining the experiment a priori, then publishing a clear result, one can simply continue to move the goal posts, change the methods, pick the data, then exclaim, "Got It!" OK, if you got it, you can do again, right?

Anyway, I gotta really try and not go here. It's a belief system, I understand that. People do not like to be told that their primary senses are not 100% reliable. It's much more ego-syntonic to postulate weaknesses in the experiment or flaws in the whole rational method. I would be totally into accepting that idea if there was evidence.

http://www.kenkantor.com/publications/audi...ish_part_02.pdf

-k

This video is a bit long but the first 10 minutes are on topic here.

[JKent]

just a thought--maybe it's time to move this thread to the kitchen

[kkantor]

This video is a bit long but the first 10 minutes are on topic here.

Thanks for this link.

I missed this panel, but have done a few in the past. Intros aren't given, but for those who don't know, Jim Johnston is one of the towering figures behinds the scenes of modern audio. His work in psychoacoustics and sound perception underpins a great deal of what we hear, from surround sound to MP3 to cell phones.

-k

[kkantor]

I just watched it, and thought it was on the mark, although it is too bad that much of it was truncated. That topics like those need to be debated at all shames the hobby. I used to debate JJ on usenet years ago. We really went at it at times, although at bottom we mostly agree. I certainly agree with what he said on the video.

Howard Ferstler

Don't be ashamed of audio. These issues exist in many hobbies and fields. Any way you want to measure it, radical subjectivist tenets have lost almost all commercial impact. Compare this to, say, the field of cold remedies, colon or liver,"cleansers and detoxifiers," or male performance enhancements, gasoline additives, "low fat" foods.

Yes, I also used to talk with JJ on the boards, and even visited him at his lab at ATT during a project we did. However, it wasn't until several years later, when I was hired to work on:

http://en.wikipedia.org/wiki/Alcatel-Lucent_v._Microsoft,

that I became aware of just how ubiquitous and important his contributions are.

-k

[soundminded]

I just watched it, and thought it was on the mark, although it is too bad that much of it was truncated. That topics like those need to be debated at all shames the hobby. I used to debate JJ on usenet years ago. We really went at it at times, although at bottom we mostly agree. I certainly agree with what he said on the video.

Howard Ferstler

"That topics like those need to be debated at all shames the hobby."

That you and others waste your time debating this kind of junk with tyros only shames you. Whether or not one should be ashamed depends on whether those things that preoccupy them are far beneath their level of knowledge and intelligence. There is the other end of the spectrum to discuss but perhaps the real shame of it is that there are few to discuss it with.

For example, I've recently made a change to the signal processing loop of my simulator. I often find I've done the right things without knowing why until later. First the explanation. In Sam's "Handbook for Sound Engineers" 1988, the text I use as my reference, Brook and Uzzle write in 7.7.2 about the distinction between three categories of echoes, that up to 50 ms the early reflections associated with the direct field, the reflections arriving between 50 and 400 ms called the running reverberation, and the reflections after 400 ms called the terminal reverberation. R is defined as the ratio of the time intergral of the power squared of the running reverberation to the early reflections. "Running reverberance is perceived as liveness, adds fullness of tone, degrades definition. Terminal reverberance (and often thrilling) sound that persists after an abrupt stop in the music. Later they also talk about the angle of arrival of running reverberation, the wider the better.

It works! Recently I modified the signal processing loop in my simulator. Injecting a signal with additional 80 ms having an electrical RT of 0.3 seconds made a remarkable difference. (at 40 ms, much less so.) This added signal typically will appear from about 140 to 220 ms after the direct field arrival depending on the settings of the primary processor delay, typical is about 160 to 165. This effectively increases the density of all delays subsequent to that time. Decreased overall gain settings for the reverberant field still allowed for significant increase in the level of reverberation and and yet virtually eliminated the possibility of the Haas effect at any usable relative level. It also allowed for even further increases in overall RT. The angle of attack in my listening room of first running reverberation varies depending on where you sit from about 135 degrees to around 75 degrees horizontally and about 45 degrees vertically while secondary running reverberation occupies the balance of 360 degrees horizontally and is always at least 45 degrees vertically but can be nearly 90 degrees in some locatons. The angular density is very low by design, enormous diffusion the result of speaker design and placement. Any one reverberant speaker is unidentifiable until you are within about 2 to 3 feet of it and almost directly in line with its axis which is pointed straight up.

In 7.8.1 they state that "another product of Beranek's study of concert halls was the solution to the problem that had plagued acousticians since Sabine's time. The calculated measured reverberation times were often substantially greater than the measured times." Apparently this is due to the audience's absorption. Personally I have found that my own preferences seem to coincide with longer RTs at least electrically than most measured data. I also prefer listening in empty halls during rehearsals than in filled halls during concerts when I can. Unfortunately I don't have a method to measure the acoustical RT of the performace of the simulator and correlate it with the electrical settings of the processor. This calls into question IMO the standard test methodology which I'm going to try to investigate to see if it is somehow flawed.

Now if you think after this I am going to worry about capacitors, wires, and such, you're crazy.

[Zilch]

My brother collects vintage watches. They don't keep time any better than a $39 quartz Timex, but they sure are prettier, and watching the sweep hand move around the dial without jumping can be kind of hypnotic.

Well, there's your argument: time is analog; seconds do not occur as steps.

The quality of time as measured and displayed by vintage timepieces is obviously superior.

[Vinyl vs. CD, anyone.... ]

[kkantor]

I did the bulk of my "serious" B&W work with Ilford FP4, a nice medium-speed film. I did some 35 mm copywork with Panatomic-X, but generally when I did copying I used medium-speed 120 or 220 film and one of my Pentax 6x7 units and a huge 135 mm macro lense for that format. I never fooled much with pushing Tri-X, and if I needed it I would just use a flash. I liked powerful flash units and nearly always bounced the light off of a white ceiling (hopefully available) to get a good room fill. However, if I had to shoot with available light I would use my 35 mm gear and had 28 mm F2, 35 mm F1.8, 58 mm F1.2, and 85 mm F1.7 lenses on hand to handle that. Yeah, they were heavy.

The internet is not a hobby. It is a disease.

Howard

Howard, and All,

So, I have a question that relates something in your post to audio. It's a sincere question, and not an editorial.

The difference between F2 and F1.4 is one stop, and between F1.4 and F1.2 is only a half stop. A very rough analogy in amps would be 50W to 100W to 150W. The first jump is significant, in terms of listening results. The second is less so.

Yet, with a 50mm-ish lens, the price of an F1.2 is much higher, it's heavier, more delicate, has more vignetting and spherical aberration, etc. Still, I lusted after them, (though couldn't afford one). Is this just a case of GAS, like wanting a problematic 150W amp, rather than a conservative 100W amp, or did a F1.2 really open up significant photographic possibilities?

-k

[soundminded]

Howard, and All,

So, I have a question that relates something in your post to audio. It's a sincere question, and not an editorial.

The difference between F2 and F1.4 is one stop, and between F1.4 and F1.2 is only a half stop. A very rough analogy in amps would be 50W to 100W to 150W. The first jump is significant, in terms of listening results. The second is less so.

Yet, with a 50mm-ish lens, the price of an F1.2 is much higher, it's heavier, more delicate, has more vignetting and spherical aberration, etc. Still, I lusted after them, (though couldn't afford one). Is this just a case of GAS, like wanting a problematic 150W amp, rather than a conservative 100W amp, or did a F1.2 really open up significant photographic possibilities?

-k

The value of an F1.2 lens is in low light conditions where available light will be the only source of illumination. Experience with developing film quickly teaches the meaning of the phenomenon of reciprocity failure, the fact that beyond a certain point, halving the amount of light requires more than twice as much exposure time for the film to achieve a given degree of latent image intensity. This type of lens is especially useful for photography of nocturnal wildlife. In this type of photography some red light may be introduced because many nocturnal animals who might scatter in whiter light don't seem to be bothered by it. Even so, the lower the light level the better.

[Steve F]

Ah, the P-38 Lightening. Prettiest plane in history.

Howard Ferstler

No argument about the "Fork-tail Devil's" great appearance, although some would say it was more menacing than beautiful.

For absolute beauty, it's tough to beat the Spitfire XXII (with the bubble canopy), or the sheer Italian style of the Macchi-Castoldi MC 202 and 205 series.

Steve F.