Zaph's Challenge

[---k---]

I'm surprised I haven't seen anyone post about Zaph's Challenge that he posted to his blog yesterday.

http://zaphaudio.com/blog.html

Quick summary, Zaph has built two pairs of speakers that are as identical as he can, but one with high distortion drivers and the other with low. He's willing to ship them to people for listening if they pay to ship them to the next person.

Pretty cool. I know Brandon and a few others have tried similar things. It will be interesting to see how it works out.

[Notorious_AK]

I guess Zaph doing it in response to GedLee's research.
But Geddes did mention importance of high quality construction and presence of shorting ring in a driver.

I would be very interested to see 3in voice coil drivers (Morel or Hi-Vi - good built quality, high power handling and so-so distortion measurements) vs much lower distortion seas or scanspeaks.

[---k---]

I'm not going to guess why Zaph is doing it. He usually not shy of stating his opinion or hiding his motives, so I'll let him say. I'm not really looking forward to the internet debating, as I can already predict much of it. Still, I think it is pretty cool that he is willing to do this.

I'm not sure my ears and opinions are worth much by themselves. If some in the Chicagoarea to want to participate (Chris and Paul), I would be more than willing to host a day and the shipping costs.

[Chris7]

This is probably the most exciting thing to happen in the DIY community in a while. (In my view, the only thing worse than not measuring is measuring but then arbitrarily ignoring the results, assuming they don't matter without listening.) Thanks to Zaph for undertaking this.

[benchtester]

If some in the Chicago area to want to participate (Chris and Paul), I would be more than willing to host a day and the shipping costs.

I am always up for a listening party, but in Zaph's blog, he mentions:

"People will see the importance of non-linear distortion information for themselves with some relatively long term and in-depth listening."

I have a thought for the metropolitan areas. If the speakers could be scheduled to be within driving distance from listener to listener, this would save shipping and reduced risk of transportation damage. For logistically support I would be willing to receive the speakers in the San Francisco Bay Area and sheppard them around to the scheduled listeners, then afterwards I would ship them to the next area. It would also help me to meet some of the local people and would help to ensure that the speakers stay on schedule. Perhaps you could do the same for the Chicago area.

After the unbiased results are in, maybe Zaph would let them be demo'ed at gatherings like Burning Amp (shameless plug - check it out) and listening parties?

[Jim Holtz]

I hate to play devils advocate but if one pair has Revelators and the other pair has Dayton classic drivers, it won't prove much. I'll wait to see what the drivers are before proclaiming this an eye opening test.

Now, if one pair has woofers with .1% distortion and the other pair has woofers that have .5%, then it'll be a test to see how audible slight differences in distortion actually are.

Jim

[benchtester]

I hate to play devils advocate but if one pair has Revelators and the other pair has Dayton classic drivers, it won't prove much. I'll wait to see what the drivers are before proclaiming this an eye opening test.

Now, if one pair has woofers with .1% distortion and the other pair has woofers that have .5%, then it'll be a test to see how audible slight differences in distortion actually are.

Jim

Just an observation followed by wild speculation:

In his last clearance, Zaph put up a pair of SB Acoustic 5" drivers for sale; then they disappeared from the list. I assumed he thought up a project for them. Is this the project????

[---k---]

"People will see the importance of non-linear distortion information for themselves with some relatively long term and in-depth listening."

I agree with this 100%. I've tried to do listening tests before. I think in quick tests, it is very hard to pick up much beyond tonal, voicing differences. But, a group get together is easiest.

When Zaph gets closer to having the details worked out, we'll go from there.

[Jed]

Quick test... grab a pair of BOSE/store bought speakers and put them next to your Mission Accomplished speaker.

Though, I'm sure the differences would go beyond just harmonic distortion. I'd assume both pairs would have to have linear FRs to make a valid comparison- among other things to isolate the listening experience to the differences in harmonic distortion.

[Undefinition]

I'm not sure my ears and opinions are worth much by themselves. If some in the Chicagoarea to want to participate (Chris and Paul), I would be more than willing to host a day and the shipping costs.

I'm definitely down for it.
But sure, we could take it a step farther. You could use the pair for a while, then I'll take them, then pass them on to Chris.
Either way, I'd like to participate.

[---k---]

Zaph will have to put grills on the speakers, so we won't know which is which, until we cheat.

[JSG]

While interesting, this test appears very close to uncontrolled to me. It appears that the probability will be high that drivers can easily be preselected to be audibly different.

In addition, different drivers are likely to have different polar responses, and different breakup modes. The latter is particularly an issue for harmonics that depend strongly upon breakup modes.

A more realistic test that might get around some of these difficulties would be to use identical drivers and crossovers, along with analog circuits to generate nonlinearities, or better yet, the same speaker pair with the capability to switch these in and out. This would allow the user to select their own sources of music or audio, and allow the aforementioned difficulties to be controlled. Of course, it has its own difficulties. One is still preselecting the type of distortion, and it may be difficult to replicate the types of distortion typical of drivers.

Or, if you have low distortion speakers, you could predistort sound files in a more or less arbitrary manner, and play them. It's still not double-blind, though perhaps there is a way to make it so.

John G.

[ttan98]

Go to this site:

http://www.diyaudio.com/forums/showthread.php?s=&postid=1760919#post1760919

for Gedlee's comment, right of the press.

[mazurek]

Hehehe, I suppose my dad will be evaluating this soon. He encourages me and sometimes buys my systems. I will be delivering to him a system with a very similar crossover network to his previous one with the main difference being lower distortion drivers at higher volumes. He very much likes his old system, hopefully he doesn't play it at volumes where he becomes unsatisfied with it relative to the new system, and enjoy them both. I feel I do notice a difference, but sometimes once you drop the stressed out nitpicking mode of listening, its all good. I suppose the main requirement is to make distortion low enough that you don't actively listen for it.

I think there are a lot of things with loudspeakers that are important if you make them important. For example, phase. Small differences won't make a big deal if you deal with it sloppily in the crossover, but can snap things into place if you are almost there. Once you start paying attention to system distortion and trying to uniformly lower it, I believe that makes a difference too. Even electronics to some extent, but usually the most noticeable thing with quality electronics is low frequency extension I think. I guess whats important is not to go crazy off into the deep end of imaginary-land, but to make sure to form testable conceptual models and then investigate (making sure not to assign undying faith to one model). Unfortunately my quest for knowledge has emptied my pockets.

[AJINFLA]

Using something like a DEQ2496 and eqing each speaker to as flat as possible say 80-15k @ 2m on axis outdoors, then listening in a 2m triangle...if the ambient noise level is sufficiently low. Maybe even using software eq and burn individual WAVs for each. Of course, switching will be an issue, similar to when Brandon was going to do his comp with cone materials. Controls for something like this can be difficult. Sounds like fun

cheers,

AJ

[lunchmoney]

Go to this site:

http://www.diyaudio.com/forums/showthread.php?s=&postid=1760919#post1760919

for Gedlee's comment, right of the press.

Safe to say even Zaph would agree that it isn't perfect science... and that absolute truths aren't meant to be derived from it and carved into the side of a mountain somewhere.

But that doesn't mean it isn't worth doing.

[kevinp.]

Yeah, I would love to partake if only to understand what I should be listening for, but I doubt my name is well-known enough.... :cry:

[wigginjs]

Could the same test not be created by taking a pair of speakers with very low non-linear distortion and adding non-linear distortion components through DSP (probably a computer program)? Can non-linear distortion be added via FIR or IIR filter? Seems simpler and more accurate than going through the process of building these two pairs of speakers. You could also control the amount of distortion added to compare a different levels.

[dlr]

Could the same test not be created by taking a pair of speakers with very low non-linear distortion and adding non-linear distortion components through DSP (probably a computer program)? Can non-linear distortion be added via FIR or IIR filter? Seems simpler and more accurate than going through the process of building these two pairs of speakers. You could also control the amount of distortion added to compare a different levels.

This is the difficulty, I think, in trying to use a single system. How do you generate non-linear distortion equivalent to that in a motor? The fact that it's non-linear means that it will vary with the signal level and in essence random when considering music content. It's going to change with every change in signal content and magnitude.

I suspect that zaph has done is get two versions of the same drivers, one with copper caps and one without. There will be differences in sensitivity I expect in this case, but if the Fc is done identically and they are level matched as he claims, I can't see any better way to do this sort of test. You can't "switch in and out" non-linear, random distortion.

The testing for distortion seems to be related to adding in known amounts of linear distortion to test for audibility of some known distortion. That may be the only real-world method of evaluating distortion in a way to provide some level of confidence required for documented testing purposes.

I still don't know how much importance to apply to distortion. It may be that I don't listen at very high levels often enough for it to be as important to me.

This will still be an interesting test and I'd do it if I felt that I could dedicate the time. Just don't expect certain posters at "the other forum" to do anything but scoff at it. That I predict with a high level of confidence.

dlr

[Chris7]

Geddes' current position as I understand it is that low-level distortion (e.g. from an amplifier or source) is potentially audible but high-level distortion from transducers is not. So you couldn't really falsify his view with a source-only test. (Of course his position really seems counterintuitive.)

As for his attitude towards loudspeaker distortion, he's basically adopted a position that's unfalsifiable: any reasonable loudspeaker's nonlinear distortion is inaudible, but if we find a loudspeaker with audible distortion, it's not a reasonable loudspeaker. Ultimately this is an anti-scientific position.

I appreciate Zaph's empirical approach here.

[wigginjs]

I appreciate Zaph's empirical approach here.

I hate to nitpick because I like Zaph's approach too, but it isn't exactly empirical is it? The reviews generated will be completely subjective.

[Chris7]

I hate to nitpick because I like Zaph's approach too, but it isn't exactly empirical is it? The reviews generated will be completely subjective.

Empirical just means experimental. It doesn't have to mean "objective" in an abstract sense. e.g. Every experiment in psychology is empirical; it's just not necessarily objective.

[alias2]

One possibility would be to build a 3rd pair with different drivers but also low
distortion along exactly the same principles, I do not know if the drivers are
available to do this. The gist here being the "differences" between the two
low distortion versions (or lack of) should show the validity of the approach
in qualifying that the essential differences are due to distortion differences.

Alas I'm not in CONUS, /alias2.

[Rick Craig]

Interesting comparison but difficult to pull off. It would also be best to test them blind; otherwise, personal biases can affect the outcome.

[Jed]

Interesting comparison but difficult to pull off. It would also be best to test them blind; otherwise, personal biases can affect the outcome.

I'm pretty sure the speakers will look identical. The listener will not know which is which.

[benchtester]

I hate to nitpick because I like Zaph's approach too, but it isn't exactly empirical is it? The reviews generated will be completely subjective.

The point that Dr. Geddes missed is that this is not a scientific experiment. Zaph's goal is not to test a theory or hypothesis; he isn't trying to prove anything.

This is basic engineering. We have a goal of building a good sounding loudspeaker (so the product performance is inherently subjective). We start with a productive definition, then a conceptual design, measuring of components, selection of components, detail design, assembly, tuning and final test and then to put the loudspeaker into service.

The goal for the engineer is to produce a loudspeaker that is pleasing to listen to. (The goal of the listener is enjoyment.) The purpose of the measurements is to guide the design. This exercise is an attempt to understand how predictive the measurements are to the final performance of the loudspeaker. This is a process improvement project.

Some measurements correlate very well with the final performance. If you show me a frequency response graph with a big spike, I can describe the sounds the speaker will make without listening to it. One of my correlation exercises consisted of running a metal driver, with a big breakup node, unfiltered; it was very educational. Zaph is trying to do the same type of thing with non-linear distortion. His goal is much more challenging, since he needs to control a number of variables to isolate the target parameter. I am certainly looking forward to it.

[---k---]

I think the big thing with this test, that makes it different than all the other BS you see on the internet, is that it is going to be a real test.

While there are numerous other ways a test could be done, no one seems to do it! Tons of internet bandwidth are devoted to discussing the best way to test, but very few have actually even attempted to do the testing. (I'm not even sure people who actually do listening any more - how could they with as much time as they spend online.)

However imperfect this test may be, it will actually get people to do some listening. It will be extremely easy for the tests. Zaph is building the speakers and shipping them to the first person. All that person has to do is hook the speakers up and actually listen to their usual mix of music. The test may not answer every question, but it will answer some. I think it will serve as a good foundation for future discussion.

[dlr]

His goal is much more challenging, since he needs to control a number of variables to isolate the target parameter. I am certainly looking forward to it.

This is so often the fly in the ointment of speaker comparisons. How does one create a system to test one difference and have "all else equal"?

The more auditioners that zaph can get involved, the better it's going to be. Certainly it will be single blind as long as no one drags out their measurement system to check the distortion.

Thinking about that, I would ask that zaph measure both systems using his standard set of measurements and more if time permits, but the caveat being with the crossover in place. It won't show nearly as much to use only raw driver data.

dlr

[benchtester]

This is so often the fly in the ointment of speaker comparisons. How does one create a system to test one difference and have "all else equal"?

dlr

We need to rely on another engineering principal:

"Close enough for all practical purposes"

(There is an off-color joke distinguishing engineers from scientist, with this as a punchline. I'll PM to anyone interested.)

[lunchmoney]

We need to rely on another engineering principal:

"Close enough for all practical purposes"

(There is an off-color joke distinguishing engineers from scientist, with this as a punchline. I'll PM to anyone interested.)

Yes please. I'm a designer who works with a lot of engineers. More ammo is always needed. :

[dlneubec]

I've not read Geddes study, so I could be completely wrong here, but I'm guessing that there will be at least one big difference, the audience. In this case, Zaph will probably be using speaker designers who are used to critical listening and ferreting out problems in loudspeakers.

I would bet Geddes participants were a more general audience who were not used to ciritical listening. I'll also bet that Geddes work was with very small segments of listening, rather than hours or days, in which case the higher distortion might have to be much higher to notice, since listening fatigue is not part of the equation.

[HeatMiser]

As for his attitude towards loudspeaker distortion, he's basically adopted a position that's unfalsifiable: any reasonable loudspeaker's nonlinear distortion is inaudible, but if we find a loudspeaker with audible distortion, it's not a reasonable loudspeaker. Ultimately this is an anti-scientific position

My thought when I read the linked post was that the good Doctor is applying a rather too formal logic to this topic where he talks about "burden of proof". He seems to think he can dismiss any findings contrary to his own, having found one or more counterexamples. The problem here is that as far as I know, nobody is claiming that nonlinear distortion always and without fail implies inferior sound reproduction, as if the two were a mathematical equivalency. If such a claim were being made, Dr. Geddes might well say that he has disproved the thesis merely by demonstrating a single case where it isn't true. But I don't think that's the claim at all, and we might all be better served by taking experiments like this as an opportunity to reveal a few more data points that may suggest some useful correlations.

[Rick Craig]

I'm pretty sure the speakers will look identical. The listener will not know which is which.

Hmm - what drivers fit that description? I know of a few coming out that may look exactly like their previous versions but are different behind the front of the cone. Based on something I posted earlier Krutke may have to eat some crow. :rofl:

[fbov]

I think the big thing with this test, that makes it different than all the other BS you see on the internet, is that it is going to be a real test.
...

Actually, that depends on us - how well we let our ears be the deciding factor and ignore non-aural cues like appearance, expectation, prejudices, etc. I have high hopes that Zaph did a good job on system design and appearance, but it's up to us to try, as much as possible, to leave our preconceptions at the door and let our ears do the judging. Yes, long term listening is required for some factors like fatigue, but not exclusively; some differences are easier to sense with short time delays between changes.

I wish you all the best of luck and look forward to reading your results. I also hope to hear the comparison someday, as the easiest way to train your senses is by example.

Have fun,
Frank

[benchtester]

Give the guy a break. Making different drivers performance similarly except distortion; and you want them visually identical? :E

That's what I call an "over-constrained product definition". (Use some grill cloth if necessary; taken from the same bolt for uniform FR of course. :lol: )

[---k---]

I'm not sure I want to get off topic discussing Geddes. But, according to the his AES paper, he used 15 second clips from Phantom of the Opera. The study was sponsored by the Eastern Michigan University, so I would assume unpaid students. I counted 21 dots in the graphs, so maybe 21 students.

I'm going to leave it at that, and not add my opinions.

I always like to listen to different speakers, because I always learn something. The more I listen, the more I learn. I still have a long ways to go. So, I'm looking forward to this.

[mazurek]

I'm not sure I want to get off topic discussing Geddes. But, according to the his AES paper, he used 15 second clips from Phantom of the Opera. The study was sponsored by the Eastern Michigan University, so I would assume unpaid students. I counted 21 dots in the graphs, so maybe 21 students.

I'm going to leave it at that, and not add my opinions.

I always like to listen to different speakers, because I always learn something. The more I listen, the more I learn. I still have a long ways to go. So, I'm looking forward to this.

That is a good point. I am really excited to see tests by this group. There is a lot to be said for people who have the patience, concentration and time to listen. For example, my old room mate had moved out and visited after a year. He insightfully told me that my new speakers were better in the upper midrange. The old version was not bad, maybe only a dB or so off in FR, and slightly higher HD. The point was that this individual knew how to characterize sound during extended listening, which allowed him to have a memory and make comparisons. Its cool that this experiment will enable people to take all the time they need to characterize it for themselves.

[jkrutke]

Hi all, I'll just make one quick set of replies and comments on this. I was going to post yesterday, but then a can of silly string exploded in my kitchen, caking the walls with pink goo. Heheh, don't ask. Anyway, here's what I have written up.

Quick summary, Zaph has built two pairs of speakers that are as identical as he can, but one with high distortion drivers and the other with low. He's willing to ship them to people for listening if they pay to ship them to the next person.

Yes, except that they aren't built yet. I'm slowly working on that.

I guess Zaph doing it in response to GedLee's research.
But Geddes did mention importance of high quality construction and presence of shorting ring in a driver.

Go to this site:

http://www.diyaudio.com/forums/showthread.php?s=&postid=1760919#post1760919

for Gedlee's comment, right of the press.

Though I'm not a big fan of what goes on over at DIYaudio, I just took a look. Some things never change. I stopped reading that forum a year ago because it seems to have devolved into Audio Asylum. Earl Geddes fits right in though. This has nothing to do with Earl in particular, though he seems to have a lot to say about it and I see he's very quick to shoot it down before it has even started and before any details have even been worked out. Doesn't bother me at all, let him say what he wants. To be clear, I'm not interested in Earl Geddes' professional opinion on anything, nor am I interested in having a debate with him. Now if Wolfgang Klippel or Sigfried Linkwitz wanted to chat with me, I'd be happy to listen.

I hate to play devils advocate but if one pair has Revelators and the other pair has Dayton classic drivers, it won't prove much. I'll wait to see what the drivers are before proclaiming this an eye opening test.

Now, if one pair has woofers with .1% distortion and the other pair has woofers that have .5%, then it'll be a test to see how audible slight differences in distortion actually are.

The internet is full of people playing devils advocate. Everyone's got something to prove, and everyone's got a hidden agenda. It's obvious to me that some folks are already worried that I might actually "prove something" and are starting a defensive campaign early. I'm not saying you're in that group Jim, but guys like Earl Geddes certainly are.

The drivers I've selected have substantially different non-linear distortion profiles, and I've chosen them that way on purpose, along with the requirements that I be able to get the response curves, both on axis and off axis, to be close to the same. The differences in non-linear distortion can't really be summarized with a simple percentage number. This is about isolating the whole package of non-linear performance. The goal is not to see if people can notice minor differences, rather it's just to help people link what can be seen in non-linear distortion testing to subjective listening. I believe that link is the weak part in the DIY community right now.

Just an observation followed by wild speculation:

In his last clearance, Zaph put up a pair of SB Acoustic 5" drivers for sale; then they disappeared from the list. I assumed he thought up a project for them. Is this the project????

Holy cow, you guys don't miss anything. I had to yank those drivers quickly before anyone bought them because I don't actually have them. I had forgotten that I borrowed them to Mark K to be run through his style of testing.

Zaph will have to put grills on the speakers, so we won't know which is which, until we cheat.

Interesting comparison but difficult to pull off. It would also be best to test them blind; otherwise, personal biases can affect the outcome.

Yes, the systems will have identical dark brown grills which are non-removable. The frames will be designed to have a low impact on the response curve. Here we will have 2 pairs of speakers that look the same, and measure the same as far as frequency response.

While interesting, this test appears very close to uncontrolled to me. It appears that the probability will be high that drivers can easily be preselected to be audibly different.

In addition, different drivers are likely to have different polar responses, and different breakup modes. The latter is particularly an issue for harmonics that depend strongly upon breakup modes.

This is a comment that makes me wonder what you guys think my goals are. I re-read my blog entry and I think it's pretty clear. The drivers chosen for the systems are audibly different because their non-linear distortion profiles are different. That's what this is all about. The systems use drivers that are the same size and are laid out with the same center to center distance and baffle locations. That combined with the same filtered driver response curves and a somewhat lowish crossover point means the polar responses will be the same for the most part.

I hate to nitpick because I like Zaph's approach too, but it isn't exactly empirical is it? The reviews generated will be completely subjective.

Yes, when the end test method is ears, it is indeed completely subjective. You could call it an empirical test based on subjectivity.

Thinking about that, I would ask that zaph measure both systems using his standard set of measurements and more if time permits, but the caveat being with the crossover in place. It won't show nearly as much to use only raw driver data.

Yes, I'll be doing individual filtered and system measurements.

On to some other general comments: Both woofers are metal. Both have breakup nodes that happen to fall within 1kHz of each other. Both tweeters are 1" fabric domes. The acoustic center offsets of the drivers are fairly close. The boxes will be small and sealed, mostly for ease of shipping. Bass extension isn't a concern, except along the lines that the extension be similar between the systems. All linear distortion issues will be removed from the picture. All non-linear distortion issues will be measured. Folks will hear it first, then see what they have been hearing. I have not decided if I will withold the non-linear measurements until after evaluation.

Initially, there will be no group listening of these systems, only one at a time. This is to encourage people to form their own opinions without being subjected to anyone else's. I think a week is enough time for a person to avoid the typical evaluation issues associated with short term listening. I might consider shipping to areas where there are a dense concentration of listeners and the speakers can just be dropped off at the next guy's house, however I would require the promise that there will be no discussion about the results. As far as selecting people, I'm going to favor those who have not "taken sides" and just want to hear for themselves. I'll probably limit the number of people to about 6 or 8, to complete the rounds before the end of the summer.

This "test" if you could call it that, is less about proving something and more about education. I think it will be fun and interesting, and yes, for some people it will be eye opening. It's a ways off however, and I'll chat more about it in the future.

[Jed]

Hmm - what drivers fit that description? I know of a few coming out that may look exactly like their previous versions but are different behind the front of the cone. Based on something I posted earlier Krutke may have to eat some crow. :rofl:

Grill cloth.

[Johnloudb]

I've not read Geddes study, so I could be completely wrong here, but I'm guessing that there will be at least one big difference, the audience. In this case, Zaph will probably be using speaker designers who are used to critical listening and ferreting out problems in loudspeakers.

I would bet Geddes participants were a more general audience who were not used to ciritical listening. I'll also bet that Geddes work was with very small segments of listening, rather than hours or days, in which case the higher distortion might have to be much higher to notice, since listening fatigue is not part of the equation.

I completely agree, experience really does matter and so does extended listening. Geddes was being pretty closed minded I think.

[Jim Holtz]

Quote:
Originally Posted by Jim Holtz
I hate to play devils advocate but if one pair has Revelators and the other pair has Dayton classic drivers, it won't prove much. I'll wait to see what the drivers are before proclaiming this an eye opening test.

Now, if one pair has woofers with .1% distortion and the other pair has woofers that have .5%, then it'll be a test to see how audible slight differences in distortion actually are.

Originally posted by John Krutke
The internet is full of people playing devils advocate. Everyone's got something to prove, and everyone's got a hidden agenda. It's obvious to me that some folks are already worried that I might actually "prove something" and are starting a defensive campaign early. I'm not saying you're in that group Jim, but guys like Earl Geddes certainly are.

The drivers I've selected have substantially different non-linear distortion profiles, and I've chosen them that way on purpose, along with the requirements that I be able to get the response curves, both on axis and off axis, to be close to the same. The differences in non-linear distortion can't really be summarized with a simple percentage number. This is about isolating the whole package of non-linear performance. The goal is not to see if people can notice minor differences, rather it's just to help people link what can be seen in non-linear distortion testing to subjective listening. I believe that link is the weak part in the DIY community right now.

Hi John,

Thank you for responding. Also, thanks to you, Jon Marsh, Mark K., Jed, AugiePro and anyone else that has performed driver testing and made it available to the DIY community. I and many, many others make your testing the 1st stop when considering a design.

To clarify, I don't have a dog in the fight or an axe to grind. If I could think of any more cliches I'd throw them out but I can't. The point is, my devils advocacy was simply a thought I had in my head about comparing drivers with large differences.

I've subjectively felt for a long time that different cone materials have signature sound qualities that transcend the motor, crossover design, frequency response etc. I've also seen drivers dismissed as unworthy for what I considered minor increases in distortion over the one that is lowest. Yet, when I actually listen to the super low distortion driver, it sounds bland and doesn't resemble the live performance which is in my opinion the ultimate comparison. However, when I listen to other super low distortion drivers, I think "WOW" that sounds real. The typical response by many is, you just like to hear distortion. That theory doesn't seem to have much value since I like designs based on some super duper drivers and not others. I can only deduce there's more to driver design than distortion alone. Yes, its very important but I submit, it's not the only factor to consider.

I have zero to gain or lose from any out come of your testing, nor do I have anything to prove, I'm just curious. I am looking forward to seeing the results of your comparison tests. I think it'll be very interesting and maybe it'll shed some light on the observations I've made over the years. Regardless it'll be fun! :T

Thanks for putting forth the effort and investment to make it happen!

Jim

[DancesWithBeers]

I see John's experiment sort of like restaurant reviews I read all the time before going out to eat. Although all the information is subjective, and a lot of the people's opinions might be way off base because they don't understand the style of food, or might have their own issues that might interfere with having a good experience, you can, from reading enough of the reviews, see trends and many points of agreement that go beyond mere chance.

I think these sorts of opinions are a good thing. Remember when you just had to make a guess as to where to eat? I used to eat at as many bad as good restaurants. These days I rarely have bad food and that is because, for the most part, there are objective, shared components to things we experience as good. There have been studies showing this for physical beauty, music (certain progressions of chords sound better than others, etc), and a host of other things that were thought to be purely subjective. We, as humans, agree about common tastes far more than we disagree.

As with all things, there are going to be people who enjoy things that most of us would have to be forced to experience. Humans can adapt to and learn to enjoy almost anything. But for the rest of us, unless John overestimates our ability to hear distortion, I think we'll be able to add some auditory commonalities to our body of knowledge that didn't exist before - as long as the reviewers exhibit the same honesty that John does when he reviews drivers.

Looking forward to the results.

-D

[Amphiprion]

There is a large DIY group in Austin that I'm sure would be interested in this. I would be happy to front shipping costs from the group in Austin to the next individual. The Austin crew could probably make a DIY day out of it instead of having our usual poker gathering. Several of guys in the group are builders/designers with everything from IMP/M to Praxis and everything in between. It's pretty hardcore group; two members used to run a for-profit custom loudspeaker and HT install shop that closed about a year ago.

I used to be the bass list admin back in the good old days, if that gives me enough street cred to be part of the test.

[---k---]

Mark,
I don't believe for a second that you would be able to resist breaking out your measurement gear and spoiling the whole test. I think you and your crew may have too much street cred. :P

[Dennis H]

I don't think Earl Geddes has bothered to learn anything about Zaph and his distortion testing or he wouldn't have reacted so negatively. They both agree that THD as a single number isn't a good metric of audible distortion. Geddes's papers strike me as formalizing mathematically what most of us already believe, that higher order distortion is more objectionable than lower order.

Quoting his first paper, his equation for a metric of audible distortion has the following characteristics:

• It should be more sensitive to higher order nonlinearities than lower order ones.
• It should be weighted towards greater values for nonlinearities at lower signal levels.
• It must be immune to changes in offset and gain (first order slope) since these are inaudible effects.

Of course Olson was saying the same thing 40 years ago and he was building on studies done in the 1920s. You can read all about it in this masters thesis written a couple of years before Geddes's paper.

A New Methodology For Audio Frequency Power Amplifier Testing Based On Psychoacoustic Data That Better Correlates With Sound Quality