Inconsistencies in AR's 12" woofer curves

[Steve F]

This, of course, is the essence of the ages-old argument--First-arrival vs. total energy.

Kloss, Petite (Kotsatos), Toole, et al think first arrival on-axis determines the speaker's character and voice in the listener's ear/mind, and the room relflections simply modify things from there. So for them, the 1m on-axis FR curve is not just a test/developmental tool, but a real-life indication as to how a speaker actually sounds.

Allison, Chamness, Moran, CU et al feel the total energy in the reverberant (far) field is what counts and that far-field energy swamps first arrival for listeners seated in the far field (as in a normal living room or den).

Of course, a good speaker can do both--have great 1m on-axis response and smooth, well-behaved far-field energy response.

The AR Verticals (the 9, 90, 91, and 92) are perfect examples of that, and that's why I feel they sound so good, even to this day.

Are you seated in the near, critical, or far field? Easy test: Hold up a sofa pillow about 6" in front of your face. If you're in the near field, you'll hear a great increase or decrease in high frequency energy when you raise or lower the pillow.

In the far field, the spectral balance is almost unchanged with the pillow raised or lowered.

This test conslusively and simply proves that in the far-field it's total energy that determines the speaker's tonal character, and it drives the First Arrival folks nuts. But there you have it.

Steve F.

[tysontom]

"Kloss, Petite (Kotsatos), Toole, et al think first arrival on-axis determines the speaker's character and voice in the listener's ear/mind, and the room relflections simply modify things from there. So for them, the 1m on-axis FR curve is not just a test/developmental tool, but a real-life indication as to how a speaker actually sounds."

"Allison, Chamness, Moran, CU et al feel the total energy in the reverberant (far) field is what counts and that far-field energy swamps first arrival for listeners seated in the far field (as in a normal living room or den)."

This is so true. The argument has raged for years with the near-field crowd winning—not by any quantitative proof that it's better—but by reason that it is the easiest, most straight-forward way to determine the output quality of a speaker itself. As Steve points out with the "pillow test," the first-arrival sound is not the true tonal character of the speaker—and is only a fraction of what one hears—so therein lies the conundrum: total energy into the room or first-arrival energy from the speaker measured anechoically or gaited? The answer probably lies within tests of both. Ken Kantor used to say that no one measurement was the sensible way to judge speaker quality, and that you had to use many tests and make a judgment based collectively on the results of them.

I still believe that the argument that hurts the 1-meter steady-state frequency-response measurement the most is microphone position. If the microphone is moved only a few inches one way or the other, the frequency-response measurement will change due to different interference effects; for standardization, and for this reason, RETMA and ASA long ago established the 1-meter, on-axis measurement technique to at least get a standardized method of measurement. Acoustic-power measurements, on the other hand, show close correlation when measurements are taken in different parts of an area back within the reverberant field. So... do you listen up close to the speaker or back in the listening room?

When Julian Hirsch first tested the "all-conquering" Acoustic Research AR-3 in 1960, he found that this speaker was, by far, the finest loudspeaker he had ever tested in his laboratory setup. "The sounds produced by this speaker are probably more true to the original program than those of any other commercially manufactured speaker system we have heard." Hirsch went on to say, "The listening tests showed beyond doubt that this was a superior speaker system—a match for any composited system we could compare it to." "When very good program material was used, the realism of the sound was striking." However, when Hirsch measured the AR-3 at a 1-meter distance, he noted lots of interference effects and dips and peaks throughout the response. Granted, he was not using a gaited system back then, but he placed the speaker on its back and measured facing up into the sky outdoors, giving a semi-anechoic environment. He told me later that when he moved the microphone into different positions, the response changed completely, with the dips and peaks moving up or down, or disappearing entirely, in the response. This demonstrates the difficulty in making quality judgment by simply using the 1-meter system frequency response measurement.

"Of course, a good speaker can do both--have great 1m on-axis response and smooth, well-behaved far-field energy response. The AR Verticals (the 9, 90, 91, and 92) are perfect examples of that, and that's why I feel they sound so good, even to this day."

I agree with Steve on this statement to a degree: speakers (such as the AR Tower speakers) can be designed to perform well with both types of tests, but it is the method of measurement for near-field testing should probably be anechoic rta measurement rather than 1-meter frequency-response measurement. I believe that AR's measurements of the AR9 and other Tower speakers were done in this fashion. AR even used the outdoor method by raising the AR9 on a tower above the ground for measurement, something that had been done since prior to WWII by Jensen, Western Electric, University Sound and others.

—Tom Tyson

[speaker dave]

Are you seated in the near, critical, or far field? Easy test: Hold up a sofa pillow about 6" in front of your face. If you're in the near field, you'll hear a great increase or decrease in high frequency energy when you raise or lower the pillow.

In the far field, the spectral balance is almost unchanged with the pillow raised or lowered.

This test conslusively and simply proves that in the far-field it's total energy that determines the speaker's tonal character, and it drives the First Arrival folks nuts. But there you have it.

Steve F.

Really? You can hold a pillow in front of your face and not hear a difference in a distant speaker's sound? Far field or not, I have never been unaware of the direct field of the speaker. Are you really conducting this test or simply assuming an outcome based on your beliefs?

A couple of my observations: If you go to the cinema and they play background music prior to the show, you can close your eyes and point to the sound sources hidden behind the screen (no need to close your eyes I guess). This is well beyond the critical distance and the direct field is still detected and given an importance beyond its simple level would suggest. As a second experiment, from when we were debating cabinet diffraction, I had my wife take a ruler on edge near a tweeter and slowly vary its distance to a tweeter (thus imparting a comb filter to pink noise). I could hear it at any distance in a lively listening room. In fact when I put the speaker in one corner and moved to the diagonally opposite corner (as far as I could get away in that room), I could still hear the effect.

Study after study is showing the greater importance of the direct field rather than the reverberant field.

David S.

[Steve F]

I try to pick my words very carefull and intentionally:

"In the far field, the spectral balance is almost unchanged with the pillow raised or lowered."

Not "completely" unchanged, as if I could "not hear a difference in a distant speaker's sound."

But, without question, the difference in this test (which I have done countless times with many different speakers in many different rooms) is strikingly more dramatic in the near field than in the far field. Strikingly.

However, as I said, there is no good reason why a speaker can't have both an unfettered, non-comb-filtered, non-diffracted, vertically-aligned, stable image near-field response and a smooth well-behaved far-field response. No reason at all.

Does the AR-91 (which is excellent in both near-field and far field) still sound better than the AR-3a (even after you correct the 3a's treble defficiency with about a 5 dB treble boost above 8k or so), even thoough both use essentially identical drivers (the same woofer, the same dome midrange, and very similar 3/4" tweeters)?

Yeah, it does. I compared the 91 and the AR-11 (which use identical drivers, including the 3/4" ferrofluid-cooled soft-dome tweeter), which I owned at the same time, and the 91 was better.

Was it better because of its superior, more well-behaved near-field behavior? After all, the 91 has the diffraction-reducing Acoustic Blanket and its drivers were neatly aligned vertically, whereas the 11 had no Blanket and its mid-tweeter were side-by-side (horrors!) when the speaker was on a stand.

Probably.

But the 91 was only this much better. (fingers a millimeter or two apart).

So, ok, all things being equal, get the direct sound right. No reason not to.

But the best modern speakers I've heard (like Mike Chamness' VR-M Series at BA) were designed to consiously and intentionally do both near- and far-field right.

Like I said, no reason not to.

But far too many oh-so-hip nouveau designers these days reject total energy altogether, as not even being part of the "how it sounds" equation.

To my ears--which have voiced and launched millions of top-selling and highly-reviewed products, I say, "Ignore far-field at your peril."

As someone on this site says often, "YMMV."

To each his own. If there were one and only one right way, that's the way everyone would do it.

Steve F.

[speaker dave]

If you can optimize both then you have covered all your bases, but, of course, sidestepped the issue of relative importance.

That is the more interesting question: if one can be optimized to be ideal (direct field or reverberant field) and the other must be compromised, then which one should get more of the designer's attention?

I always point people to Floyd Toole's original study of the 25 systems that listeners rank ordered and that he subsequently measured. From best to worst there is a clear correlation between the axial frequency response of the systems and their order of preference. Looking at the power response of the same systems I would defy you to find any rhyme or reason connecting their rank ordering and their power response. Now, neither Floyd, nor Lipshitz and Vanderkooy found that power response was inconsequential. Peaks in power response are audible if high enough, and there seems to be a range of downward power response contours that are safest, but there is no particular power response curve that is mandatory, in contrast to axial frequency response.

This is not the fashion of the hip, but more the understanding that 50 years of research have revealed since Vilchur's day.

David S.

[Steve F]

I'm not sure why the choice has to be made to optimize one and compromise the other. Good engineers get good results with both and the speakers sound correspondingly terrific.

As to your implication that energy is also an important design consideration to nouveau designers, I'd say no, it itsn't. The vast majority of speaker engineers measure 1m, on-axis gated FR--maybe, maybe a few off-axis curves also--and when they get the curves to look good, they proclaim, "Done! I've designed a great speaker!" Then there are a few listening tests, some tweaking of the tweeter resistor or the woofer choke, some more 1m gated FR curves to check that it's still ok, and then it's really done.

Energy measurements? I'd offer two things:

1. Not only don't today's engineers measure far-field energy, they don't even have a reliable, repeatable method for measuring far-field energy.

2. There are very few--if any--speaker companies that have a reverberant chamber to measure energy. Mike Chamness had a very specific way that he used to measure energy in his lab. I loved to be there during those measurements. I also remember Gerry S. at BA using his trusty old Ivie RTA to check far-field energy, just to make sure things were behaving ok.

But today's boys measure 1m gated FR, and a few off-axis curves. That's it. They don't measure energy. They don't even think of energy. I'd say they don't even know the term.

If the design is good (if they're using the drivers within their non-directional passbands, the drivers' outputs are summing properly through the x-o, the lobing is correct and controlled, the diffraction effects are minimized, etc.), then the far-field energy is probably ok.

But it's ok only by accident, not because they've intentionally taken it into account during the design process, or assigned it any specific weight.

Steve F.

[speaker dave]

Hi Steve,

Well, its not as simple as you say. First off there is always a compromise between axial response and spherically averaged response. The best axial performance will come from an LR approach. This always comes with a mild power hole, so that is a compromise. Subjective tests show (I believe) that power response holes are benign but it doesn't sound as if you would accept that. If you want your networks to add in quadrature then you can have flat power through the crossover but then you will have a rising lobe in the up or down direction. I don't know of any way to get dead flat response in both measurements, and to some degree a compromise is required.

As to "1m on axis and you're done". I don't think that characterizes any of the better designers I know. Most are aware that 1m is a bit close for a good approximation to the far field, and will be looking at 2m response as well (which is generally adequate for the development of the far field response). Most look at response at a number of angles, possibly emphasizing near angles around a likely listening window, but nonetheless getting a good look at off axis performance.

You keep repeating the term "energy" but I don't think that is being correctly used. Energy is seldom measured in acoustics although intensity (Watts / meter squared) is sometimes measured. You probably mean power response, another misnomer, used to refer to the power or energy weighted (i.e. pressure squared) spherical average frequency response. Frequency response (pressure) is sampled at points around a sphere and each curve is squared and weighted by the area it represents. The end result is not power (at least not without the right weighting factor), so "power response" is not strictly correct. Reverb rooms can be used for this, as long as they are calibrated and the corrections are applied. At the same time gated response in close is just as good for calculating the power curve, and the errors of 1m measurement are actually inconsequential if a spherical average is needed. Most of these "nouveau" designers should be able to calculate the spherical average if they wanted to .

This is frequently measured in pro products as it is needed for directivity index, but it is not used much in consumer because equalization of power response is known to be an unreliable approach. This crops up frequently these days as DSP is making room EQ so easy. Everyone seems to be practicing a steady state approach which gives a result that is midway between correcting the axial response and the power response. Right off the bat they find that they need some sort of room curve (flat power response is always too bright) and those that have much experience at it are confounded that the room curve that works changes from room to room and speaker to speaker. Again, this is because concentrating on power response sends you down an problematic alley.

Did I mention that SMPTE is working on totally revising the X Curve equalization method because they find that it doesn't make any 2 theaters sound the same? Again, looking at the room curve brings in the power response and equalizing power response is not going to get you to consistent sound.

I know everyone here loves the early AR products, but I seriously doubt that EV would want everyone to worship them to the point of denying that not everything was known and understood in the classic days, and that real progress hadn't been made since.

Really.

Regards,

David

[Steve F]

We're getting very close to the point of an endless loop, where no good new info is being presented and the readership is getting bored.

I've been in so many situations with so many very highly regarded and accomplished engineers using the 'ol meter stick to measure out from the tweeter axis of the prototype speaker under test that I've lost count.The 1m on axis FR measurement dominates. Others used as well, like 2m and off-axis? Thank G*d for that. But 1m is where 99.99999% of designers start and they keep at it until that result is good. Then they may move on to other tests/measurements. Or not.

I've never seen an engineer--except for Chamness--do a full-fledged power response (call it whatever you please) measurement, graph it, analyze it, study it. Or even just calculate one. Ever. Yet for some strange coincidence, by some sheer stroke of uncorrelated luck, Chamness' speakers at BA were flat-out fabulous.

He did all the usual 1m gated, off axis, etc. measurements too, but in addition to power.

Hey, your XA's were some of my favorites also, so there's more than one way to skin a cat, d'accordo?

Steve F.