Measurements: When, What, How, Why

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
No Arguments here Loren, but I suspect that it will be possible to find two completely different speakers which measure almost identically but which will (and maybe not to all) sound completely different on certain types of music.

I highly doubt this, esp given your scenario (ie from listening, being able to articulate how they sound different, then reversing to the measurements and hone in and find the differences..'no slam for rock, fine for cello tends to suggest we hone in on the bass/upper bass region. Just yabbing here). If they measured 'identically' with sufficient depth of measurements then they would sound identical. Surely?


I have my own theory, as to why different people will have very different prefferences when listening to a bunch of speakers, and it has less to do with what the speakers do well, and more to do with what they don't do well. Everyones hearing is different (as SY has pointed out), and my proposition is that different people are more or less sensitive to the different "errors" that any particular loudspeaker might have. Whether they be frequency response aberrations, phase errors, distortion, compression or what have you. So whilst it might be possible to measure all of these things, the results are going to mean different things to different people, depending on which particular thing it is that they are more sensitive to.

I like the way you are going here, it has been something like this simmering in the back of my mind for a while now.

Seems to me there are two basic (but differing) interpretations of this. The most basic is 'measurements don't tell us anything' from the completely subjective crowd, well we can dismiss that one straight off.

I THINK why they might say that is this slight variation, 'If measurements told us all then WHY does speaker A which measures well sound bad to my ears?' (or amp A etc etc)

To me that heads straight into the territory you have alluded to just now.

What we *have* only are the measurements of the speaker/equipment, what we DON'T have is the measurements of the listeners ears/preferences. As pointed out, not only do we have different ears and different amounts of hearing acuity (we don't need SYs extreme example of his wife's lesser hearing in one ear, it's a fun thing to do when a group of audio mates get together...I set my system up so the image was exact dead centre for me, but to others it was six inches left or right, a fun and interesting experiment we can all do), we also have (surely) differing levels of tolerance for distortion (say).

I might be fine on 1% level of third harmonic distortion (making this up as I type), but my ears could bleed if I have that along with 0.5% at second harmonic...and you could be the complete reverse. (ie along the lines that earle has found with ratios of distortion in amps)

So whilst it is all well and good to have all these objective measurements, I suspect until we get down to the personal measurements of likes and dislikes etc it will be hard to close the gap between 'measurements can tell is all' vs 'measurements tell us nothing'. Quite obviously as it stands both extreme statements are incorrect, I feel by including that missing set of measurements (the personal preference 'file') we would get very much closer to the truth.
 
Could you post a few links please?? I know JBL does on all the Pro Stuff and some of their consumer products. I would also like to know who else makes the measurements as well. Especially in consumer products where it's generally not considered an issue.

Rob:)
"...where it's generally not considered an issue." Can not agree with that, Rob. Maybe it shouldn't happen, that's all. When designing and choosing a speaker/driver you need/want to know.
Beyma does "Power Compression Losses" graph in their general specs sheet.:)
By the way, also learned the same thing from Mr. Geddes (tests).
 
Last edited:
From someone locally in the speaker business: "When we make speakers we need to know what the customer preferences are, which may vary from location to location..."

From that, I gather than in order to make money in the business, you just have to give the costomers what they want.
 
Hello Inductor

"...where it's generally not considered an issue."

Did you miss Consumer Products in the beginning?? I would like to see manufacturers Power Compression measurements on speaker systems like Infinity, Dynaudio, Wilson Audio, and so on

In the Pro Sound world it's a given. I consider it an issue. My DIY speakers are all on the large side using high power high efficiency drivers except for a couple of the bass drivers with lower sensitivity.

Rob:)
 
I have a basic question for the guys that have and use measurement set-ups.

Asside from defining what a meaninful set of measuremets are what do you look at now?? We can use that as a starting point.

I look at the on axis frequency response at a couple of measurement distances. I also like to look at multiple measurements in the Listening Window and then average them. I use MLS Sine and Stepped Sine. Usually do both and MLS and a Sine and compare them for similarity for each measurement.

Look at the CSD, ETC, Step Response.

Have not done polars will one of these days.

Rob:)

I spend time learning about the CSDs I create. I enjoy looking through Zaph's measurements and learning about different drivers.

I'm also Spending more time learning about Wavelets. Thanks to Elias and Micheal on this site, oh and LeCleche (sorry for the spelling).

In the end I could study measurements for 5 years and still know nothing about drivers and speakers. I need to get the drivers and listen so that I can correlate that sound back to the CSD, IMD and polar plots.....Its practically impossible to figure it all out.

but its fun for me because I enjoy learning.

Right now Im trying to figure out which compressions drivers sound okay to my ears and match that up with measurements. So far compression drivers are not remotely as good as my Ribbon designs in the sweet spot. No doubt the directivity is there but there is a certain "rough edge" to most CDs.
 
Last edited:
Hello Inductor

Did you miss Consumer Products in the beginning?? I would like to see manufacturers Power Compression measurements on speaker systems like Infinity, Dynaudio, Wilson Audio, and so on

In the Pro Sound world it's a given. I consider it an issue. My DIY speakers are all on the large side using high power high efficiency drivers except for a couple of the bass drivers with lower sensitivity.

Rob:)
Yes I did, but it's your fault. Reread it again (the way you said it, you mention only JBL/#16).:) Unless I need my glasses.:cool: Fanny the way some people, like you, try to change their opinion after another poster doesn't agree with. Not to mention bad manners from others, nowadays...:forbiddn:
 
Last edited:
That's completely due to the horn they are mounted to. Horns are very complex and most are poorly executed, so the resulting sound is bad.

sure but this is where the measurement are needed and show little difference between bad and good. Augerpro did some serious horn measurements so this isnt taking on a whim. Including a Geddes OS waveguide.

I believe its more then the horn. Geddes has supported this when saying its less about the OS waveguide and more about the complete design. I think Dr. Geddes also believes CDs mostly sound the same (he could be refering to after proper XO design), I have 6 different CDs and they do sound different on the QSC waveguides.
 
I'm sorry Earl (if you're still following this thread :eek:).

I believe you may have misunderstood my attempted point.

I don't (nobody? does) expect you to give a treatise on measurement.

I admit the "very good" can be misread, or read both ways.

The "alluded" to applies to this forum. You have mentioned many, many, aspects of measurement over time, and their significance as you see them, but possibly, if cookbook statements of some sort arose out of this thread, they would be welcome. Particularly arcane, little known, little understood issues, as is already happening.
This need not go to the extent of violating intellectual provenance.

There is Joseph D'Appolito's "Testing Loudspeakers" book from Audio Amateur Press. I've read thru it and Joe gives a lot of detail about the process - formulas and charts galore. However, what's lacking big time is how to interpret the measurements.

The intent of this thread was to attempt to gather information "systematically" (just the way I am :)), especially interpretation.

Certainly, this is going on in the Cables thread, but OT.

If you are interested, I'm sure you could continue to help us with significance, interpretation, importance, and critique, in particular.

It was certainly not my intention to be so stupid as to get you off side.

David
 
I have a basic question for the guys that have and use measurement set-ups.

Asside from defining what a meaninful set of measuremets are what do you look at now?? We can use that as a starting point.

Impedance plotted on the complex plane (that gives LOTS of information!), frequency response on axis and as many off axis points as I can get, nearfield response of woofers, in-room response, distortion, accelerometer response when I see pigtails in the Z plots. I'd love to do more, but my measurement setup is limited (anyone who wants to get me a Klippel for my birthday...?).
 
Just another Moderator
Joined 2003
Paid Member
Hi Doug and Zilch, I am in no way trying to undermine any findings of either John (zaph) or Earl. What I do find amuzing is that John puts a very strong emphasis on distortion as being a very important measurent, Earl (if my memory serves me correctly) thinks it is largely irrelevant (unless I have confused him with someone else).

You will also find other industry experts who will have other oppinions, and disagree on various aspects. Surely they can't *all* be correct on *every* aspect????

Did I say that measurements were meaningless? NO.
What did I say?

1. That in my oppinion, it is possible that the measurements that we can and do do, don't necessarily tell you the full story.

2. That in my oppinion, some people may be more sensitive to some sorts of faults than others. Yes the measurements may show these faults, but to say that if fault A is present in speaker B, that all people will equally dislike that speaker I don't think will reflect reality. Yes findings may show that in general everyones preferences are *similar* and that speakers of a certain quality level will be preferred over a lower quality level, but within a specific quality range you will probably find that you will not get an unanimous prefference for speaker A over speaker B.

I'm not in a position to provide any testing evidence that two speakers that measure similarly could sound different, I'm basing this on the assumption that whilst the two speakers may behave very similary for things that typical tests measure, like freqency response tests, transient response, phase etc, that when presented with a very complex waveform to reproduce (which lets face it, most music with more than a single voice or instrument is) that they may not equally be able to do so. When you think about it, the fact that transducers CAN do what they do as far as creating the illusion that there are multiple different sound sources each with their own very specific characteristics, that for all intents and purposes sound like they are separate is pretty remarkable.

Maybe I'm just unaware of these tests, or how the tests I *am* aware of can predict this, but I don't take the approach that everything is known, and that there is nothing new that can be discovered.

I'd also like to see the full context of the quotes at the begining of the thread. As quotes in Isolation can be interpreted very differently to how they are if they are in the original context. For instance In Earls quote, is the You directed at a particular person, or is it a collective "You" ie meaning that no-one yet knows how to do some as yet to be discovered measurement...

Tony.
 
Last edited:
Here is a bit of a "literature search" (borrowing from the "Cables" thread), a bit of a review of the reviews, a bit of synthesis, with obvious acknowledgment to Earl Geddes.
Hopefully an invitation to all to fill in the gaps for novices like me.
With some aspects, the isue seems to be the best way to make the measurement (e.g. a clean gated impulse response). With others the issue may be what matters (e.g. types of distortion). With others it may be the correlation (e.g. thermal compression and dynamics?).

This is the multi-way speaker forum, but we don't want to ignore the room:

Gedlee

Basically, in a well designed system, the speakers (and let's not forget the room) are really the only impact. By this I mean that swapping out anything but the room and the speakers with other "well designed" pieces will yield "no perceptible change" (in a DBT of course). If this is not the case then it is a straightforward task to find out why this has occurred and to identify which piece is "correct" and which is not.

Swapping rooms and speakers is always such a major effect that it is difficult to determine the "why"...

...Imaging is more complicated because it involves the room and how the loudspeakers interface with the room. Just looking at an axial response is not going to tell you anything about how the speaker will interface with the room, you need to look at the entire polar response. It needs to be well controlled both on and off axis AND fairly narrow.

"Spaciousness" is clearly another property that is more the room than the loudspeaker so there aren't many loudspeaker measurements that are going to tell you much about this, although InterAural cross correlation has been shown to be highly correlated with the perception of spaciousness.

Specific to speakers:

We obviously want an even on-axis respomse, however...

Gedlee

Basically the two things that are usually measured Axial FR and THD are probably the two least useful measurements that I can think of. No one could tell a good loudspeaker from a bad one with any confidence based on these measures...

Off-axis response. It seems we want an even polar response in the horizontal, but how narrow and why? (There are CD waveguides that cover anything from 90 X 90 to 60 X 40). What about the importance of the vertical plane and its contribution to the overall power response (e.g. d'Apollito MTM)?

Impedance:

SY

Impedance plotted on the complex plane (that gives LOTS of information!), frequency response on axis and as many off axis points as I can get, neaarfield response of woofers, in-room responmse, accelerometer responmse when I see pigtails in the Z plots...

...Most often, manufacturers use a Klippel system to measure power compression and nonlinearity... (power compression) is pretty easy to measure and is done routinely by every competent driver and speaker manufacturer.

Who uses accelerometers? How useful are they (or is box construction OK with rule of thumb)?

Distortion measurements:

Gedlee

There are some fringe issues in perception that have not been resolved- "dynamics" comes to mind as not fully understood, but we have learned a lot- like the insignificance of nonlinear distortion in loudspeakers, etc.
... "thermal compression" which I see as equvalent to "dynamics". There are no standard tests here, but I have done several tests of the thermal properties of different loudspeakers and they are highly variable and significant. This needs to be standardized and better quantified, but its not black magic.

So would 4" voice coils on woofers compared to 2", assuming a satisfactory inductance, be an advantage in home hi-fi for example?
As has been pointed out, Geddes, Linkwitz, and Zaph, to take three, may not agree on all the distortion issues?
An invitation for comment and elaboration.

Phase:

B&K have (expensive) stand-alone phase measuring equipment. Do the computer set-ups that DIYers have access to give enough information, accurately enough? What phase response details matter?

Perhaps, most salient to this thread, what information that other techniques provide is important, or are they just ways of expressing the same data? CSD, wavelets, spectrograms, integration time (MLS v's swept sine wave v's stepped sine wave etc.)

For example, does the combination of the phase response and step response (acknowledging that they are both aspects of the impulse responmse) tell us enough about time alignment or group delay? What should we aim for anyway?
 
Who uses accelerometers? How useful are they (or is box construction OK with rule of thumb)?

Anyone with $2 to buy one.:D

Very useful. If memory serves, Stereophile used them in their speaker measurements as well, so there should be some good illustrations of their utility.

What phase response details matter?

That's a matter of great controversy! d'Appolito has a very accessible treatment of excess phase in his book.
 
What I do find amuzing is that John puts a very strong emphasis on distortion as being a very important measurent, Earl (if my memory serves me correctly) thinks it is largely irrelevant (unless I have confused him with someone else).

amuzing? I find it extremely irritating, disturbing

I'd also like to see the full context of the quotes at the begining of the thread.

the quotes are purely marketing statements of "I am an expert, I know my (secret and arcane) art and what I am doing and they don't, therefore trust me, buy my ideas/expertise/products" etc.

best,
graaf
 
It seems to me there are a few really conflicting issues in this thread. One group of issues is "What is the purpose of the speaker" and another group of issues is "What are the measurements of importance?"

Some background- Let us imagine a performance hall. There is the sound heard at the podium by the conductor, there in the sound heard by the person in row 3 center section, there is the person in row 15 center section, and there is the person sitting under the balcony in the back row 45 center. These people all have different valid listening experiences and in their private reproduction systems (which mimic there performance experiences) have very different goals for reproduction. Not going into the details of this but it is very clear the sound at these different positions are very different leading to very different goals for loudspeaker design and measurement.

Measurements are about what the designer believes is important very similar to the seating arrangements above. What the builder of a loudspeaker measures has a lot to do with which person above the loudspeaker is designed to please. I can say with no doubt the speaker targeted for the guy under the balcony is not going to satisfy the conductor standing at the podium and what the conductor wants to hear. Yet both are valid listening experiences and both are valid loudspeakers and both are impossible to build into one loudspeaker without massive amounts adjustments to be made by the user. Many receiver manufacturers have attempted such with DSP effects like "concert hall" and "rock" and "mud room" setting for their surround sound systems. All a failure in my opinion but someone else may like them and that is fine. It is clear a single loudspeaker will not do.

So what is important to what we hear? If we all wrote down the 20 things that were important in a loudspeaker or reproduction system in the order of importance pretty much 15 of the 20 items would occur on everyone's list. However the importance of any particular item would reflect that individuals personal priorities. Every list is valid with of 1x10e10 difference in order available. Again there is no point to go into the details of this because the point is very clear and the problems of measurement and design become self evident.

In the end look for guidance and products by people who have goals similar to the goals of you. It is your personal goal that is being attempted to satisfy so find others with similar goals and happiness :) may result in the end. For me, I like the sound the conductor hears at the podium: fast; intimate; detailed; perfectly balanced; crystal clear; low distortion; high signal to noise ration with wide solid clearly placed in space performers. This sound is absolutely not the sound heard by the guy under the balcony. If you want that sound follow my post and target my responses. That array on the left is the ultimate for me with DSP controlled signals for each driver and a constant radiation angle vertically and horizontally. It really winds my watch. On the other hand if you are that person under the balcony you may want to look at planar loudspeakers like magnepans.

So who wants to answer this question to for the one size fits all? Only the fool. My size fits the conductor and studio engineer. Look for the one in your size and hopefully a fit may be found.
 
Hi Doug and Zilch, I am in no way trying to undermine any findings of either John (zaph) or Earl. What I do find amuzing is that John puts a very strong emphasis on distortion as being a very important measurent, Earl (if my memory serves me correctly) thinks it is largely irrelevant (unless I have confused him with someone else).

You will also find other industry experts who will have other oppinions, and disagree on various aspects. Surely they can't *all* be correct on *every* aspect????

Why does it matter who is correct? BOTH are correct. The goal we have is to take all valid data and learn from it. I think there is too much generalization over Geddes opinion on distortion because we can definitely tell a BAD driver from a GOOD Driver when certain order harmonic distortions are high. Remember Geddes has never posted its completely irrelevant. I suspect he has some assumptions about the quality of the driver used and from that POV distortion isnt as meaningful as other parts of the overall speaker design.

Measurements that even are in audible can matter to some. Its important that people like Zaph do many measurements to help out DIY communities. Sadly he is know working in the industry (building and selling drivers and speakers) so free measurements are no more :(



Did I say that measurements were meaningless? NO.
What did I say?

You more or less are posting that you can hear something and you believe it can not be measured which I believe is false. Soundwaves are 100% measureable. Everything your ear gets is measureable.

What you should be probably discussing is how your brain (and ours) processes what it hears and the conclusions it makes.

btw, if you are not controlling your listening test its pointless to discuss what you heard because your brain will use other variables outside of the signal from the ears to make conclusions.

1. That in my oppinion, it is possible that the measurements that we can and do do, don't necessarily tell you the full story.

2. That in my oppinion, some people may be more sensitive to some sorts of faults than others. Yes the measurements may show these faults, but to say that if fault A is present in speaker B, that all people will equally dislike that speaker I don't think will reflect reality. Yes findings may show that in general everyones preferences are *similar* and that speakers of a certain quality level will be preferred over a lower quality level, but within a specific quality range you will probably find that you will not get an unanimous prefference for speaker A over speaker B.

This is all just subjective preference and has little to do with speaker measurements, speaker accuracy, etc. Its no more then person A likes the color red and person B likes the color Blue. Also a huge majority doesn't know better, they have never been trained properly, they have never educated themselves on what the most accurate sound will sound like.



Essentially everyone should have an in room EQing system and they should EQ to their specific needs.

Its like buying a car....we pick the brand we like then we pick the colors at purchase time. The EQing is picking the colors.


I'm not in a position to provide any testing evidence that two speakers that measure similarly could sound different, I'm basing this on the assumption that whilst the two speakers may behave very similary for things that typical tests measure, like freqency response tests, transient response, phase etc, that when presented with a very complex waveform to reproduce (which lets face it, most music with more than a single voice or instrument is) that they may not equally be able to do so. When you think about it, the fact that transducers CAN do what they do as far as creating the illusion that there are multiple different sound sources each with their own very specific characteristics, that for all intents and purposes sound like they are separate is pretty remarkable.

Maybe I'm just unaware of these tests, or how the tests I *am* aware of can predict this, but I don't take the approach that everything is known, and that there is nothing new that can be discovered.

I believe the illusion is more from your brain then from anything the speaker/room does. The brain is extremely powerful in terms of suggestion. You want to believe there is a live show in front of you and you convince yourself its true when many of you close your eyes. I do not close my eyes because Im a realist about audio, Im not going to seduce myself into some fake illusion. If I want live, I travel to live shows.
 
Last edited:
Off-axis response. It seems we want an even polar response in the horizontal, but how narrow and why?

Distortion measurements:

As has been pointed out, Geddes, Linkwitz, and Zaph, to take three, may not agree on all the distortion issues?

If the polar response is at or below 90 then the speakers can be pointed such bthat the first wall reflection is neagted. Hence 90 is a minimum, but below that is not necessarily a big advantage. A narrower directivity tends to delay the start of the rooms reverberation which enhances the perception of image.

Show me a single study that has correlated distortion measurements to subjective perception - I will show you three that shows that there is no correlation. Floyd Toole, Sean Olive, Alex Voishvillo, Lauri Fincham and I are all in agreement that nonlinear distortion in a loudspeaker is not a significant factor (unless it is broke).
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.