Ideas for more meaningful speaker measurements

[Rick Sykora]

Feel I should add that would like to keep the focus here on improvement ideas.

Some ideas may or may not be feasible, but if you want to discuss a particular solution, please start another thread.

Thanks!

Rick

 

[tvrgeek]

Hope I did not miss it in the previous 6 pages, but seems measurements are focused on a few power levels in steady state.

How does output vary with power? Drivers are not terribly linear once very far away from center. How much? Does it matter? We could measure @ 200mW, 1W, 5W, 20W etc.

Where does distortion start to increase? What frequencies, what SPL? In testing drivers, I have found as a general rule, very general, at about 1/2 X-Max, the distortion skyrockets. Of course, X-max is an ill-defined value. But from a system standpoint, looking at the lower range of each driver, some objective measure could be done. I suspect, but have no way of testing, there are similar points with regards to breakup. For instance, if a well designed system did not reach these corners until over 96dB, ( or about 86 for that matter) I could care less, but someone else might.

How easily does a driver "start"? We can see in the CLS it's resonances and damping. What we don't see is the inverse. The action on the first cycle from rest. This may not matter and can be interpreted from existing measurements, but I wonder. Drivers are a mass on a spring. Second and third derivative and all that.

 

[Holmz]

Hope I did not miss it in the previous 6 pages, but seems measurements are focused on a few power levels in steady state.

How does output vary with power? Drivers are not terribly linear once very far away from center. How much? Does it matter? We could measure @ 200mW, 1W, 5W, 20W etc.

Where does distortion start to increase? What frequencies, what SPL? In testing drivers, I have found as a general rule, very general, at about 1/2 X-Max, the distortion skyrockets. Of course, X-max is an ill-defined value. But from a system standpoint, looking at the lower range of each driver, some objective measure could be done. I suspect, but have no way of testing, there are similar points with regards to breakup. For instance, if a well designed system did not reach these corners until over 96dB, ( or about 86 for that matter) I could care less, but someone else might.

How easily does a driver "start"? We can see in the CLS it's resonances and damping. What we don't see is the inverse. The action on the first cycle from rest. This may not matter and can be interpreted from existing measurements, but I wonder. Drivers are a mass on a spring. Second and third derivative and all that.

I thought that was captured in the compression numbers?

And motor linearity has been captured in some of Erin’s older data.
It is more germane to a driver than to a speaker though, which explains why I saw it only on drivers.

 

[tvrgeek]

If it pertains to the motor, it pertains to the system as I know of no way that can me midigated in the crossover. No feedback.

 

[Holmz]

The distortion measurements at a fixed SPL have the distortion.

That is the motor, the cabinet, the port, etc… The whole enchilada.

 

[tvrgeek]

The distortion measurements at a fixed SPL have the distortion.

That is the motor, the cabinet, the port, etc… The whole enchilada.

Excepts it is not fixed. It varies with displacement.

 

[Holmz]

Excepts it is not fixed. It varies with displacement.

… and with the position that the motor is in.

However a pink noise signal, with a fixed SPL, should be a pretty good benchmark for it.
Where in NC are you?

 

[tvrgeek]

Hillsborough. Just north of the triangle.

 

[tvrgeek]

Ordered a new mic so I can get back to data. I was hoping the upgraded MiniDSP mic would go above 20K, but seems not. One that does is $600.
I was playing with REW and seeing how well I did no the crossover for my RS125/XT25 desktops in-situ. Not bad. The Vifa sure shows it's colors for being well behaved. I scrapped trying to use them on my new build and ordered some more Seas. I need a little better off-axis response in that use-case. Anyway, I will do some testing to see if I can measure the lower frequency distortion corners in system.

 

[fineMen]

Feel I should add that would like to keep the focus here on improvement ideas.

Since You ask this way ...

amplitude response does not need finer resolution
- grand total is summed up in- and direct sound, will vary wildly in different setups anyway
- positioning of phantom sources in the stereo panorama will depend on direct, but that isn't critical

directivity does not need better than spinorama, predicted response can be omitted
- rooms spread wildly in their properties, predictions are nonsensical
- better encourage people to measure what a certain setup actually delivers, promote a service that would possibly take care of this
- encourage appropriate equalizing
- still keep high standards for smooth directivity, but relax a bit, perfection kills

come to consider non linear distortion
- IM needs a reasonable concept, multitone is very often misunderstood and misused, IM destroys music
- HD is not that critical due to anyway very prominent overtones in any relevant program material
- but please equalize before measuring, especially in the bass

Ordered a new mic so I can get back to data. I was hoping the upgraded MiniDSP mic would go above 20K, but seems not. One that does is $600.

The Panasonic WM something of Linkwitz fame can still be found for purchase. Last summer I recorded bats in my backyard using it. After frequency transformation from about 45kHz (!!) to 2kHz with audacity I could listen to it. I couldn't determine the species, though. But, that wasn't a prob of the mike. They were not after food, it seems, just navigating, I was told.

 

[tvrgeek]

Not calibrated.

 

[hedrick]

I’ve been experimenting with speakers in my living room. I have tried 3

1. Paradigm Studio 20, a moderately recent speaker that measures fairly flat, and is well regarded. 2 way 6,5 in ported.
2. ADS L500. An excellent speaker from the 1970s. However I suspect the crossover may have drifted. 2 way 8 in sealed.
3. Altec 95, a mediocre 3 way 8 in sealed speaker from a few decades ago.

All show as fairly flat in tests for reviews or (for the Altec) factory specs. The Altec is the worst, with a 3 db or so peak at 3k but otherwise fairly flat.

I can change the timbre by adjusting bass and treble, but the speakers have character that seems pretty independent of those adjustments. I’ve wondered what it is that generates that character.

 

[tvrgeek]

Very small differences in eq, 1/2 dB in a 1/3 octave range can have huge effects.
Distortion and breakup, Again, energy where it is not intended
Polar response, diffraction How each interacts in the room, more differences in energy.
Differences in timing, mids to treble, or across the crossover region which happens to be in out most sensitive zone.
How the drivers mass effects their acceleration.
Complicated.
Even if all of that is as close as modern DSP can get it, they will still sound different. Not as much as trying to use tone controls.

So, with better measurements, they still have to go into a totally unknown environment, your room.

IMHO
The Paradigms are really terrible in the tweeter breakup. I too got suckered into buying a set due to the great press reviews.
A peak @ 3K will cause voices to be too sibilant and cymbal crashes to he harsh. 1dB is enough.
Tone controls are great and we should all have them, though not to eq the speakers. To adjust if an album is a bit hot or not. I wish we still had "loudness" controls, but the magazine "purists" convinced the high end world they are evil because you hear them ( the main idea actually) so to be high-end, they removed them and raised the price. I am surprised music servers do not have them as an option.

 

[fineMen]

Very small differences in eq, 1/2 dB in a 1/3 octave range can have huge effects.
...
So, with better measurements, they still have to go into a totally unknown environment, your room.

So You say, it is not critical to have smaller deviations in the response of the naked speaker, but what counts is the in-room response?

 

[sigbergaudio]

So You say, it is not critical to have smaller deviations in the response of the naked speaker, but what counts is the in-room response?

Above schroeder we increasingly hear the direct sound as frequency rises, so the native speaker response surely matters a lot. Dispersion characteristics also affect how the speaker interacts with the room. So I suspect that's not what he's saying.

 

[thewas]

I am afraid he really means that, see also

How to make quasi-anechoic speaker measurements/spinoramas with REW and VituixCAD

Yes, but as noted above it usually isn't practical. It adds extra variables, especially if measuring indoors. The two main issues are that 1) if measuring indoors, the distance to walls changes as you move the mic, and 2) you would have to be very careful to not change the distance from the...

www.audiosciencereview.com

 

[fineMen]

Above schroeder we increasingly hear the direct sound as frequency rises, so the native speaker response surely matters a lot. Dispersion characteristics also affect how the speaker interacts with the room. So I suspect that's not what he's saying.

What about?

(A) measurement of a ready made speaker
- on the customers site
- on the reviewers site
(B) measurement during development
- ...
- ...

All in-room versus anechoic, except for (B)'s directivity (decidedly anechoic) and distortion (anechoic, as to harmonize equalisation).

As long as Toole's theory is well defined, and it then also holds, the diffuse field determines the spectral balance, while the direct radiation determines the location of phantom sources in the stereo panorama. (As long as You put Your head into a bench vice, precisely positioned and fixed.)

Any caveats here? Remember that Harman preferes a monophonic setup for quality assessment, PIR and so on.

Only if not, what about (sic!) the sensitivity of stereo sound field impression to ripples in the direct amplitude response--if they are corrected in the diffuse field? Is +/-1dB just o/k, or would +/-3dB be pretty much sufficiant? How does such an estimation correlate to vastly varying sound balance in live performances, even during one event? The instrument may be moved ever so slightly and due to its chaotic (!) directivity the sound changes noticeably.

Not so easy

 

[Holmz]

Hillsborough. Just north of the triangle.

I am near Ashville this week.

Very small differences in eq, 1/2 dB in a 1/3 octave range can have huge effects.
Distortion and breakup, Again, energy where it is not intended
Polar response, diffraction How each interacts in the room, more differences in energy.
Differences in timing, mids to treble, or across the crossover region which happens to be in out most sensitive zone.
How the drivers mass effects their acceleration.

I doubt it. How would the play high notes if they were limited by mass?

Complicated.
Even if all of that is as close as modern DSP can get it, they will still sound different. Not as much as trying to use tone controls.

Would phase come into it?
(But I like step function s and impulse response measurements.)

So, with better measurements, they still have to go into a totally unknown environment, your room.

We hear the direct sound, and we can usually identify it separately as being in a f’ed up room.

IMHO
The Paradigms are really terrible in the tweeter breakup. I too got suckered into buying a set due to the great press reviews.
A peak @ 3K will cause voices to be too sibilant and cymbal crashes to he harsh. 1dB is enough.
Tone controls are great and we should all have them, though not to eq the speakers. To adjust if an album is a bit hot or not. I wish we still had "loudness" controls, but the magazine "purists" convinced the high end world they are evil because you hear them ( the main idea actually) so to be high-end, they removed them and raised the price. I am surprised music servers do not have them as an option.

 

[tvrgeek]

I am near Ashville this week.



I doubt it. How would the play high notes if they were limited by mass?




Would phase come into it?
(But I like step function s and impulse response measurements.)


We hear the direct sound, and we can usually identify it separately as being in a f’ed up room.

Give the Highland Brewery a visit. One of 67, but I find their ales to be the best in the state.

A lot of folks have answered for me and most are close.
Yes, in-room, in your listening position is what matters in the final evaluation. What you hear.
How each speaker will interact with your room varies.
The better behaved the speaker is, the easier it is to deal with in your room.
EQ needs to deal with frequency, phase and time.

All that is technical, measurable, tunable. What is not is how your brain inherent, genetic, and training, processes what you hear to decide if it is "musical" or not. And was alluded to, when you walk into a room, your brain takes a pattern from the ambient and biases what you hear with what it expects. So, a perfectly flat response in a very live or very dead room will confuse your poor brain. How well it separates the environment from the source varies. So, tough on speaker makers, big advantage to the DIY. We can play with the room ( under the rules of WAF) play with the crossovers, cabinets, positions and electronic processing that the OEM's can't. All hail Owens Corning 705.

Our brain used frequency, time and phase for localization clues. When they are not coherent, it takes a bit of subjective pleurae from it.

It is often confused between steady state frequency response of a sprung damped mass with how it handles inertia. Yes, impulse response can be revealing. But one needs to look at non-linear distortion as well as linear. Not the most important, but it valid within the laws of physics. Yea, mass is the limiting factor in ability to play high notes. That is why we have multi-way speakers.

 

[hedrick]

Very small differences in eq, 1/2 dB in a 1/3 octave range can have huge effects.
Distortion and breakup, Again, energy where it is not intended
Polar response, diffraction How each interacts in the room, more differences in energy.
Differences in timing, mids to treble, or across the crossover region which happens to be in out most sensitive zone.
How the drivers mass effects their acceleration.
Complicated.
Even if all of that is as close as modern DSP can get it, they will still sound different. Not as much as trying to use tone controls.

So, with better measurements, they still have to go into a totally unknown environment, your room.

IMHO
The Paradigms are really terrible in the tweeter breakup. I too got suckered into buying a set due to the great press reviews.
A peak @ 3K will cause voices to be too sibilant and cymbal crashes to he harsh. 1dB is enough.
Tone controls are great and we should all have them, though not to eq the speakers. To adjust if an album is a bit hot or not. I wish we still had "loudness" controls, but the magazine "purists" convinced the high end world they are evil because you hear them ( the main idea actually) so to be high-end, they removed them and raised the price. I am surprised music servers do not have them as an option.

The Soundstage measurements of the Studio 20s don’t show any significant distortion until 95 db. I wouldn’t think I’d be getting that. My trusty Radio Shack meter shows peaks in the 60 db range. The Stereophile measurements suggest possible woofer breakup that affects highs.

Incidentally, among the speakers listed, the Studio 20s seem to be the best. As noted, I think the L500 are no longer in optimal condition.