Matching up Advents for Stacking

[dingus]

I measure, I post; I hold no one but myself to my standard.

I also listen and enjoy, but know that has nothing to do with what the speaker IS. That's why you'll rarely find me saying anything subjective about how any speaker sounds -- I am no more reliable in this respect than anybody else. I like what I like, and you like what you like, none of which has any bearing on the facts.

FACT is, when studied under controlled conditions, central tendencies reveal themselves regarding listener preferences, and those are well correlated with objective measurements of loudspeaker performance. We each and all determine for ourselves whether that has any relevance to our individual experience....

fact is, that myself and others believe stacked Advents are a worthwhile endeavor because they sound really good. fact is, you contend we are wasting our efforts because according to the measurements, stacking "doesnt work". thats a standard i wont be held to.

my opinion of a speaker is based on how it sounds to me, but your opinion based on data of "what the speaker is" is no less subjective. the reality is that the facts which determine the merits of a speaker are determined by ones own priorities. the data is less important to me because it cant tell me what i like or dont like, or how something sounds. i'll rely on my ears to be the ultimate and final authority for this. if you place what you hear at a lower priority to the data, you are missing the point.

[Zilch]

i'll rely on my ears to be the ultimate and final authority for this. if you place what you hear at a lower priority to the data, you are missing the point.

No, you are missing the "well correlated" point. I listen, I hear. You're not talking about speakers, rather, about yourself.

It's a significant distinction....

http://www.zaphaudio.com/evaluation.html

[dingus]

that guy is as big of a knucklehead as you are.

A speaker can't be evaluated like a painting. The nature of a painting forces subjectivity while a speaker *can* be objectively tested.

paintings are evaluated objectively, from style, color, shading, lighting, individual brush strokes, texture, grain, materials used etc. these aspects can be quantified and measured just as well as frequency response and driver performance are. you've both missed the point, and that is, once all the testing and measuring is done and one listens to the sound of the speaker, then the data stops. the endeavor moves from a technical exercise to an auditory experience. we dont hear the data or the spec, we hear the sound, and it absolutely is about the individual because we all interpret what we hear differently and form our individual opinions from that. your problem is that you try to treat this as a single event when it is two separate and distinct processes. i'm not knocking anyone for enjoying the technical aspects of audio, but i do take exception when i am told that my subjective interpretations of what i enjoy do not matter.

back to stacked Advents, despite the measurements and data, they still manage to sound and work very well for many and the data cant change that. just because it looks good on paper doesnt mean its going to sound good in practice and vice-versa.

No, you are missing the "well correlated" point. I listen, I hear. You're not talking about speakers, rather, about yourself.

It's a significant distinction....

http://www.zaphaudio.com/evaluation.html

[Zilch]

your problem is that you try to treat this as a single event when it is two separate and distinct processes. i'm not knocking anyone for enjoying the technical aspects of audio, but i do take exception when i am told that my subjective interpretations of what i enjoy do not matter.

They matter, of course, but objectively speaking, only to you.

[OH, and the salesman, of course, albeit fleetingly.... ]

[dingus]

They matter, of course, but objectively speaking, only to you.

[OH, and the salesman, of course, albeit fleetingly.... ]

would you agree that it matters to anyone who enjoys the sound of stacked Advents as well?

[Zilch]

would you agree that it matters to anyone who enjoys the sound of stacked Advents as well?

Doubtless; there is no conflict here....

[speaker dave]

When stacking Advents are the top units inverted to keep the tweeters fairly close together? If this is not done one would think that the rather distant spacing between the tweeters would cause some odd interference effects. With the top units inverted, the tweeters would form a short, vertical "line" over much of their operating ranges, which would help to reduce vertical radiation in the upper midrange and treble. No telling how the widely spaced woofers would intereact with each other, however.

Howard Ferstler

This is my problem with stacked Advents. If the tweeters are mounted together (top speaker inverted) then, yes, there are likely to be interference nulls unless the listener very carefully sits at just the right height (on the bisecting plane between the two speakers).

A lot of people stack Advents and like the result. For some, that is the end of the story, which is fine.

For those that are technically curious, here an experiment. Feed pink noise or FM interstation hiss into one of your Advent pairs. Have a friend slowly slide the upper system forwards and backwards a few inches (nearer to your listening position and farther from it). From your listening position you will hear a swishing sound as the treble range goes in and out of phase. This is classic comb filtering. If your ear level isn't exactly one "Advent" height above the floor the only way to achieve flat response is with the upper system slightly farther back than the lower system, to bring the two tweeters into phase. This assumes your ear level is more than half way up the stack.

Once you have heard this effect, I don't know why you wouldn't want to correct for it.

In effect, by stacking a pair of speakers you have designed a symetrical array, a popular layout for high end systems. The difference is that most designers, when creating such arrays, find ways to reduce tweeter lobing. Unit to unit spacing (versus wavelength) is a problem: the tweeters are too far apart. Most such systems cross over to a single central tweeter to prevent the problem.

Your stacked Advents, like all such arrays, are more directional. If you allign them correctly the direct sound level at the listeners ears goes up 6dB while the power into the room has gone up 3dB (for all but lowest frequencies). So by definition the directivty index has gone up 3dB (Q has doubled). This will give a more focussed soundfield, but with less spaciousness. Some will like that, others won't.

If you don't allign them carefully then you probably have some serious suckouts in the treble range. You may like that too, but acknowledge it as an abberation rather than an improvement.

There's no magic here.

David

[Zilch]

The situation is worsened, David, by the fact that the speakers are not mirror-imaged, and thus, when stacked, the tweeters are not only displaced vertically, but horizontally, as well.

When I measured a stacked pair, there was only one line in space (rather than a plane) equidistant from all drivers, and once I located that, I started from there as baseline. It fell apart immediately with movement from that axis in any direction.

With a pair of stacks, the two lines become a point at their intersection when properly toed in. I suppose one might optimally aim one stack at one ear, and the other at the other to get it "right...."

[speaker dave]

When I measured a stacked pair, there was only one line in space (rather than a plane) equidistant from all drivers, and once I located that, I started from there as baseline. It fell apart immediately with movement from that axis in any direction.

Clearly that's your problem: you've got to stop measuring things!!

David

[kkantor]

I love the sound of stacked speakers, sometimes. It gives a them a sense of authority and richness, that might be part psychological, part radiation pattern, part changes in the tonality. Comfort food for the ears. Great for that heaping plate of spaghetti with lots of butter and garlic kind of mood.

I also know that stacked speakers are less "accurate" to the signal, according to some of the time-honored ways that accuracy has been defined in the field of audio. So, stacked speakers are not my go-to approach when my listening goal is to hear exactly what is on the record. But, both types of goal seem to me to be authentic, and we are lucky that we don't have to give up one for the other.

Now, there are legit arguments to be made about the benefits and weaknesses of both objectively based assessment (which certainly can be listening-centric itself), and purely subjective approaches. A lot of it, I think, comes down to the type of person doing the evaluation. Measurements or structured listening have very little value to most hobbyists. But, they are essential to designers and pro reviewers.

-k

[Zilch]

Clearly that's your problem: you've got to stop measuring things!!

YOU, sir, are in large part to blame for this "affliction...."

[dingus]

...A lot of it, I think, comes down to the type of person doing the evaluation. Measurements or structured listening have very little value to most hobbyists. But, they are essential to designers and pro reviewers.

thank you sir for summing up this point so succinctly.

[speaker dave]

Now, there are legit arguments to be made about the benefits and weaknesses of both objectively based assessment (which certainly can be listening-centric itself), and purely subjective approaches. A lot of it, I think, comes down to the type of person doing the evaluation. Measurements or structured listening have very little value to most hobbyists. But, they are essential to designers and pro reviewers.

-k

I've always felt that measuring without listening, as well as listening without measuring, are both fighting with one arm tied behind. Don't we measure speakers to see all their abberations and then listen to rank order these faults and see which are above a threshold of perception?

The approach I've been happy with in design, and I suspect your approach is similar, is to rely heavily on measurements early on, while roughing out a crossover, until the driver passbands are about right and the various sections are adding well, then to switch to listening for the fine balancing of a dB here and a dB there.

Its not the only approach (equal parts measuring and listening) but it is the fastest, and I think the surest. I always thought it would be interesting to design a system purely by listening, not a single measurement. I think I could do it, but I wonder if I wouldn't get stuck in a blind alley. You know, converging on some euphonic curve with every trouble area pulled down to hide it. Plus what about all the early stages where drivers sections aren't adding, poor phase blend, etc. How long would it take to stumble across a crossover point, crossover order and phasing that worked? (Those that don't design systems for a living may not appreciate these details but they are a major part of the process.)

I am frequently bemused (annoyed) to see a Stereophile review where some new company has designed a seriously flawed system because they have no technical background or ability or interest in measurements. You know, the woofer to tweeter crossover that sucks out 20dB because having the woofer and tweeter the same polarity "is the right way to go". (And what about all of these turntables that cost $10s of thousands with no specs for wow flutter or rumble because "traditional specs don't matter" Don't get me started!!)

I guess the real question is, when we know something measures badly, do we toss it out without listening and move on to something that is objectively better? Have we passed up some great sounding combination by assuming that bad measurements correlate with bad sound? Perhaps, but my experience (and prejudices) tell me, no.

Just my $.02

David

[dingus]

I've always felt that measuring without listening, as well as listening without measuring, are both fighting with one arm tied behind. Don't we measure speakers to see all their abberations and then listen to rank order these faults and see which are above a threshold of perception?.......

......I guess the real question is, when we know something measures badly, do we toss it out without listening and move on to something that is objectively better? Have we passed up some great sounding combination by assuming that bad measurements correlate with bad sound? Perhaps, but my experience (and prejudices) tell me, no.

Just my $.02

David

thats the beauty of being in my position, i dont have to worry about measuring, the work has already been done for me. in my view, measurements and specifications are essential towards meeting any design goal, but need to be verified by listening.

"If it measures good and sounds bad, -- it is bad. If it sounds good and measures bad, -- you've measured the wrong thing." - Daniel von Recklinghausen

[Zilch]

Now, there are legit arguments to be made about the benefits and weaknesses of both objectively based assessment (which certainly can be listening-centric itself), and purely subjective approaches. A lot of it, I think, comes down to the type of person doing the evaluation. Measurements or structured listening have very little value to most hobbyists. But, they are essential to designers and pro reviewers.

That depends upon the nature of the hobby. If it's DIY, which comprises an increasing majority of audio hobbyists, measurement capabilities are an essential tool. That doesn't only mean building stuff; as systems become more sophisticated, it also plays a significant role in in-room optimization. It doesn't require a huge stretch to appreciate the potential benefits of applying the same principles and technologies to simpler installations, as well, and with the widespread availability of inexpensive measurement tools, interest in how it all works and behaves in our own spaces is expanding exponentially:

http://www.classicspeakerpages.net/IP.Boar...?showtopic=5736

I always thought it would be interesting to design a system purely by listening, not a single measurement. I think I could do it, but I wonder if I wouldn't get stuck in a blind alley. You know, converging on some euphonic curve with every trouble area pulled down to hide it. Plus what about all the early stages where drivers sections aren't adding, poor phase blend, etc. How long would it take to stumble across a crossover point, crossover order and phasing that worked? (Those that don't design systems for a living may not appreciate these details but they are a major part of the process.)

I have what I consider to be an excellent article on this subject saved in one or another of the computers around here; the answer is that, even in the hands of a capable designer, the result will conform to their specific tastes and preferences only, which, as we all know, are subject to change as soon as the next time they listen. A competent head-banger will build something very different than a classicist will. With measurements and knowledge of the differences, as you suggest, it's far easier to build either or both, as desired, and mighty good ones, as well.

Bottom line of this thread for me is that Tim's experience with stacking Advents was seen as an opportunity to dump on objectivism in general, and the Zichster, in particular. Fact is, it does not invalidate the principle, as stated, in any respect....

[dingus]

...Bottom line of this thread for me is that Tim's experience with stacking Advents was seen as an opportunity to dump on objectivism in general, and the Zichster, in particular. Fact is, it does not invalidate the principle, as stated, in any respect....

and i saw it as an attempt to turn the enjoyment of stacking Advents into Zilch of an experience by using strict objectivity. fact is without subjective assessment there would be no need to bother listening.

[Zilch]

and i saw it as an attempt to turn the enjoyment of stacking Advents into Zilch of an experience by using strict objectivity. fact is without subjective assessment there would be no need to bother listening.

Bah.

A statement of principle, and a simple recommendation as to how to mitigate the consequences: turn off one of the tweeters in each stack, and lowpass one of the woofers in each.

Speaker Dave subsequently reiterated the point, and suggested an MTM alignment.

Try either or both; you might just like one of these alternatives.

[Perhaps other options will come to mind, also.... ]

[dingus]

Bah.

A statement of principle, and a simple recommendation as to how to mitigate the consequences: turn off one of the tweeters in each stack, and lowpass one of the woofers in each....

fair enough. i'll allow that i may be reading into your post's (and i definitely am taking into consideration your post's in other threads) and wrongly assuming that you believe your recommendations are the only reasonable option for everyone.

....Speaker Dave subsequently reiterated the point, and suggested an MTM alignment.

Try either or both; you might just like one of these alternatives.

[Perhaps other options will come to mind, also.... ]

i have no doubt i would enjoy them... quite possibly more than an un-modded stack, but with my ability and lack of desire to modify speakers, it will remain a mystery to me unless i happened across a readied pair.

[kkantor]

Sure... but even those hobbyists who measure things tend to use it in support of what they like, rather than as a standard. (This could be said of "pros", too, but that's a much longer and involved philosophical debate...)

Also, it takes years to learn how to make and interpret speaker measurements in a meaningful way. Buying a Behringer mic and D/L-ing some freeware is fun, but it is not going to yield much in the way of design chops or theoretical framework. Most of the measurement stuff I see on the nets is drivel. (This is not directed at you, or the handful of other hobbyists who have gone pro, and taken their work seriously.) For one thing, it invariably fails to confront the >tradeoffs< in any measurement method or design approach, no matter how objective. Speakers, in particular, are complex and subtle systems, and nobody has come close to successfully designing them by algorithm. The best designs are simply the ones that make the fewest and best compromises. The weighting of these compromises must be related to human perception, either through a perceptual model, or structured listening or just market acceptance.

And all this, anyway, assumes the end user has access to source material that meets some set of analytic criteria.

-k

"Meters don't kill speakers. Hobbyists with meters kill speakers."

[kkantor]

thank you sir for summing up this point so succinctly.

Thanks! Well, I guess that settles it then... no more objective/subjective debates on the web, eh?

-k

[kkantor]

I've always felt that measuring without listening, as well as listening without measuring, are both fighting with one arm tied behind. Don't we measure speakers to see all their abberations and then listen to rank order these faults and see which are above a threshold of perception?

The approach I've been happy with in design, and I suspect your approach is similar, is to rely heavily on measurements early on, while roughing out a crossover, until the driver passbands are about right and the various sections are adding well, then to switch to listening for the fine balancing of a dB here and a dB there.

Its not the only approach (equal parts measuring and listening) but it is the fastest, and I think the surest. I always thought it would be interesting to design a system purely by listening, not a single measurement. I think I could do it, but I wonder if I wouldn't get stuck in a blind alley. You know, converging on some euphonic curve with every trouble area pulled down to hide it. Plus what about all the early stages where drivers sections aren't adding, poor phase blend, etc. How long would it take to stumble across a crossover point, crossover order and phasing that worked? (Those that don't design systems for a living may not appreciate these details but they are a major part of the process.)

I am frequently bemused (annoyed) to see a Stereophile review where some new company has designed a seriously flawed system because they have no technical background or ability or interest in measurements. You know, the woofer to tweeter crossover that sucks out 20dB because having the woofer and tweeter the same polarity "is the right way to go". (And what about all of these turntables that cost $10s of thousands with no specs for wow flutter or rumble because "traditional specs don't matter" Don't get me started!!)

I guess the real question is, when we know something measures badly, do we toss it out without listening and move on to something that is objectively better? Have we passed up some great sounding combination by assuming that bad measurements correlate with bad sound? Perhaps, but my experience (and prejudices) tell me, no.

Just my $.02

David

Agreed. I need all the hands I can get.

-k

more later

[Doug G.]

Well, this thread has certainly evolved into a very interesting discussion and I agree that a happy marriage between objectivity and subjectivity is the best approach.

I get a kick also out of this fairly recent attitude among "modern" audiophiles that traditional specifications don't mean a thing. They are perfectly willing to rely on a traditional design approach in most of the equipment they utilize but have disdain for it at the same time.

Thanks guys.

Doug

[speaker dave]

Thanks! Well, I guess that settles it then... no more objective/subjective debates on the web, eh?

-k

How about we tackle evolution/creationism next?

David

[Doug G.]

How about we tackle evolution/creationism next?

David

OK, I'll start. God created man so he could create the word "evolution."

Doug

[speaker dave]

Here's a trick that will get rid of all the treble lobing problem of a stacked pair of Advents. Maybe Zilch can accomodate us with some measurements.

If you stack a pair of Advents and series connect them then they share the voltage evenly since (presumably) their impedance curves match. Each system gets half the voltage and radiates 6dB less than if it got the full voltage. At the listeners position, if they have been phase alligned as suggested, they will add to double the pressure or +6dB. Net level difference with 2 systems in series is zero.

Now the problem is getting phase allignment, especially at high frequencies where a little lateral shift or an elevated listening position will throw the HF badly out of phase.

If we bypass the treble around one speaker with a capacitor, the full treble voltage goes to the second system. The 6dB loss, 6dB gain steps are bypassed. Since the full treble voltage now goes to one system it matches the level of the series connected low frequency pair. Flat response is maintained and lobing is reduced or eliminated.

The only trick is to find the right size cap to shunt around one system, with sufficient rolloff to reduce the lobing.

Series connecting won't bother any amp. It becomes more likely to voltage limit than current limit, but a bigger amp should have plenty of voltage swing.

Who's going to try it???

David