Announcement: ASR Will Be Measuring Speakers!

[amirm]

Hmm, we readers seem to have countless number of detailed measeurement/analysis requests! Modern mesurements and FFT-based analysis of measured impulse response make this possible.

The "problem" is actually the modern aspect of the measurement system. As with REW, Klippel system uses the so called Farina Chirp stimulus. When converted to an impulse response the distortion nicely separates itself in time domain. Problem is, each harmonic distortion has a long decay in a real room so they run into each other and generate incorrect results. Here is a nice graph from Klippel which shows this problem:

The system has a neat feature where you can take a normal measurement (like a distortion one above) and take out the room reflections from it. Once there, the effects of the room is taken out and the results become accurate:

The user interface is not intuitive and it took me a few hours tonight to figure out how to get it to work. Will have to try it tomorrow.

 

[Blumlein 88]

The "problem" is actually the modern aspect of the measurement system. As with REW, Klippel system uses the so called Farina Chirp stimulus. When converted to an impulse response the distortion nicely separates itself in time domain. Problem is, each harmonic distortion has a long decay in a real room so they run into each other and generate incorrect results. Here is a nice graph from Klippel which shows this problem:

View attachment 45669

The system has a neat feature where you can take a normal measurement (like a distortion one above) and take out the room reflections from it. Once there, the effects of the room is taken out and the results become accurate:

View attachment 45670

The user interface is not intuitive and it took me a few hours tonight to figure out how to get it to work. Will have to try it tomorrow.

So this explains why when I use REW for checking several spot frequencies using their RTA/FFT display on a continuous tone I get much better distortion results than when I use the sweep. The way it varies isn't so different, but the levels of distortion are. Probably also explains why sweep distortion always seems to elevate around crossover points with multi-way speakers.

 

[Juhazi]

So this explains why when I use REW for checking several spot frequencies using their RTA/FFT display on a continuous tone I get much better distortion results than when I use the sweep. The way it varies isn't so different, but the levels of distortion are. Probably also explains why sweep distortion always seems to elevate around crossover points with multi-way speakers.

Partly, I guess. Taking measurements in nearfield eliminates some issues, but high spl is not possible because mic starts to distort. A diyer with unlimited time and no need to publish results (like me) can do various measurements with various settings, environments, and analyses. It takes time and learning to get repeatable and logical results.

Anyway, it is easy to see changes in distortion of a diy dsp-controlled multiway speaker, when settings are changed, eg. high D around 2kHz when tweeter has too shallow xo slope.

https://www.diyaudio.com/forums/software-tools/338511-howto-distortion-measurements-rew.html

Distortion peaks in sweeps a are mostl likely mechanical resonances. Then play sinewave of that F and listen to it, then press your finger on side/back panel and listen to the sound changing! It may be a loose wire inside rattling too etc. Measuring only at some sine waves won't reveal this kind of quality issues, that are very common in low price category or diy!

 

[celroid]

I am excited for the reviews, someone please send him Dynaudio Emit M20. Please!

 

[Rick Sykora]

Hi Amir and all,

Am new to ASR, but have been doing audio as a hobby for over 40 years. In the 80s, had some higher end equipment, but with raising family my time and money went to them. As a substitute for more audiophile pursuits, I moved into speaker design (mainly subwoofers), using Bassbox Pro, REW and XSim. As I also tinkered with quad in the day, so have moved into home theater as well.

Am in the process of downsizing now, but my main system is Epique CBT24 fronts, Mirage OMD-5 surrounds with a couple of Titanic subwoofers powered by a Yamaha TSR-7810 and an Emotiva XPA-100 for the subwoofers. I like Amir’s more objective approach to testing equipment and am really interested in seeing the speaker results! I do have a spare OMD-5 that am willing to submit for testing, but Amir’s call as they are NLA.

Aside from my audio hobby, I have a part-time computer consulting business. My longest term customer is my very non-technical wife (Winkle). So know you know the origin story for my account name. Am still catching up on ASR history, but look forward to contributing more to the forum eventually .

Regards,

Rick (aka Winkleswizard)

 

[mhardy6647]

in re: https://www.audiosciencereview.com/...-will-be-measuring-speakers.10725/post-303864

So... with some trepidation, I am gonna tip my hand a wee bit re: quantitative analysis -- with the disclaimer that I have made my livin' as a quantitative scientist (analytical biochemist in biotech/biopharma) for nigh-on four decades. I also have four-plus decades of training and experience that lead me to be very skeptical (although, I hope, in a constructive way!).

I thoroughly believe that everything important about hifi components, including loudspeakers, can be measured.
I just as thoroughly believe that we don't necessarily know how to do it.

Bear with me for a second, 'K?

I say this because, from my perspective, the quantitative analyses of (in the present case) loudspeakers uses a set of what I would call surrogate markers in an attempt to unambiguously and objectively quantify loudspeaker performance.

One can certainly envision a designer or manufacturer gaming the system (so to speak!) and come up with a loudspeaker that measures perfectly in the test(s), but is perfectly lousy in any real-world assessment outside the parameters of the test(s).

Recall, e.g., the emissions test mode of the VW Diesels not so long ago

Besides my long-time day job, I have a rather passionate interest in education, and served on the School Committee (school board) in our erstwhile home town in Massachusetts for seven years. During our time in MA, the state implemented standardized testing as a metric of performance/accomplishment that would be (was, and AFAIK, is) required for graduation from MA public secondary schools. Our little town had (and still has) very good schools, but the "best" schools in MA were pretty surprised when "we" came out on top, statewide, the first two years of the standardized testing. Our facilities were modest, our per-pupil expenditures were middle of the pack, but "we" took our mission to educate very seriously. We came out on top because the schools did a good job -- but, as time passed, other schools invested in very deliberately (and explictly) altering their curricula to teach to the test, with the predictable result that their scores climbed. That said, I would opine, that teaching to the test turns out students that are very good at scoring well on standardized tests, but not necessarily capable of performing at a vaunted level in the real world.

I think (!) that this is still a pervasive problem in hifi as well (so to speak).

Now, I don't mean to suggest that the standardized testing should be abandoned. Far from it! The best thing to do, I'd say, is to gather ever more test data and correlate it, as best possible, with the squishier aspects of sound reproduction. This can (and should!) result in better understanding of the predictability of the surrogate markers, and indeed can result in improvements of those objective tests.

This site/forum offers just that opportunity -- and I am looking forward to watching as the "learnings" develop!

whew! sorry for the diatribe!

Now, back to the regularly-scheduled topic.

 

[Ilkless]

https://www.audioholics.com/room-acoustics/room-reflections-human-adaptation

I suggest you read this. The general principles/trends toward smooth dispersion and response are consistent with known psychoacoustics. There is room for variation but it is a space bounded by these general principles, and not "teaching to the test". Your argument is apologism for subjective waffle and the cult of personality towards dismal components.

in re: https://www.audiosciencereview.com/...-will-be-measuring-speakers.10725/post-303864

So... with some trepidation, I am gonna tip my hand a wee bit re: quantitative analysis -- with the disclaimer that I have made my livin' as a quantitative scientist (analytical biochemist in biotech/biopharma) for nigh-on four decades. I also have four-plus decades of training and experience that lead me to be very skeptical (although, I hope, in a constructive way!).

I thoroughly believe that everything important about hifi components, including loudspeakers, can be measured.
I just as thoroughly believe that we don't necessarily know how to do it.

Bear with me for a second, 'K?

I say this because, from my perspective, the quantitative analyses of (in the present case) loudspeakers uses a set of what I would call surrogate markers in an attempt to unambiguously and objectively quantify loudspeaker performance.

One can certainly envision a designer or manufacturer gaming the system (so to speak!) and come up with a loudspeaker that measures perfectly in the test(s), but is perfectly lousy in any real-world assessment outside the parameters of the test(s).

Recall, e.g., the emissions test mode of the VW Diesels not so long ago

Besides my long-time day job, I have a rather passionate interest in education, and served on the School Committee (school board) in our erstwhile home town in Massachusetts for seven years. During our time in MA, the state implemented standardized testing as a metric of performance/accomplishment that would be (was, and AFAIK, is) required for graduation from MA public secondary schools. Our little town had (and still has) very good schools, but the "best" schools in MA were pretty surprised when "we" came out on top, statewide, the first two years of the standardized testing. Our facilities were modest, our per-pupil expenditures were middle of the pack, but "we" took our mission to educate very seriously. We came out on top because the schools did a good job -- but, as time passed, other schools invested in very deliberately (and explictly) altering their curricula to teach to the test, with the predictable result that their scores climbed. That said, I would opine, that teaching to the test turns out students that are very good at scoring well on standardized tests, but not necessarily capable of performing at a vaunted level in the real world.

I think (!) that this is still a pervasive problem in hifi as well (so to speak).

Now, I don't mean to suggest that the standardized testing should be abandoned. Far from it! The best thing to do, I'd say, is to gather ever more test data and correlate it, as best possible, with the squishier aspects of sound reproduction. This can (and should!) result in better understanding of the predictability of the surrogate markers, and indeed can result in improvements of those objective tests.

This site/forum offers just that opportunity -- and I am looking forward to watching as the "learnings" develop!

whew! sorry for the diatribe!

Now, back to the regularly-scheduled topic.

 

[Nightlord]

Amir - don't know if the question has been raised before (and answered). Any thoughts on how to deal with speakers where the wall boundary is part of the speaker design? Is it possible to put a wall piece, or possibly to place a pair of speakers back-to-back?

Some examples: the (old) Carlsson speakers, the current Larsen, the Guru Audion Q(M)10 - I'm certain there are non-Swedish ones too...

 

[Juhazi]

My suggestion is both NFS and room measurement.

Actually also many floorstanders with reflex ports down low should be measured in a normal room too. That is what Stereophile's Atkinson often does. Still, room measurements are specific only to that specific room, that must be reminded of.

 

[pjug]

@amirm If we send new speakers for testing, do you want us to play them for some time before sending? Or do you do a break-in before testing new speakers?

 

[LTig]

@amirm If we send new speakers for testing, do you want us to play them for some time before sending? Or do you do a break-in before testing new speakers?

Please do not play them in. We need to know whether this has any effect at all.

 

[mhardy6647]

I just had an interesting thought. Well, OK, maybe not interesting... but interesting to me.

Would the presence of other loudspeakers in the test garage lab have any impact on the measurements? Or would their effects be 'canceled out' by the testing scheme used by the Klippel gizmo?

 

[pma]

@amirm , will you be measuring speakers, or just toys, as you have been doing till now? I am not sure if measuring toys has much value.

 

[Krunok]

@amirm , will you be measuring speakers, or just toys, as you have been doing till now? I am not sure if measuring toys has much value.

You can easilly change that by sending your "speakers" to him. Or you think it is reasonable to expect him to buy $1500 speakers to measure them on a weekly basis?

 

[LeftCoastTim]

@amirm , will you be measuring speakers, or just toys, as you have been doing till now? I am not sure if measuring toys has much value.

There was some consensus to measure many inexpensive (<500 USD) speakers early to find high value items and get a sense of "what's out there". So far the most expensive item was Amir's personal Revel C52 (2500 USD), and the second most expensive was the KH80 (500 USD).

I am very interested in what "non-toy" speakers you plan to send to Amir to test!

 

[mhardy6647]

My non-toy speakers, unfortunately, are too big to spin around (in either sense of the words spin around)
They'd be a little pricey to ship coast-to-coast, too, I fear.

 

[NTK]

@amirm , will you be measuring speakers, or just toys, as you have been doing till now? I am not sure if measuring toys has much value.

These are just training exercises. I am sure @amirm can't wait to publish reports on "audiophile grade" speakers too.

Given the flak he is already receiving, I certainly won't blame him until he has gotten a totally air tight story behind his measurements before posting negative reports on expensive speakers. Many people are going to have their feelings hurt, and not like it.

 

[amirm]

@amirm If we send new speakers for testing, do you want us to play them for some time before sending? Or do you do a break-in before testing new speakers?

If you are buying new, it is best to have it drop shipped to me so we save in shipping. I usually play them some in the process of setting up the system prior to measurements. The measurements also take an hour or more so if the thing is going to change, it will and it will average out.

 

[amirm]

@amirm , will you be measuring speakers, or just toys, as you have been doing till now? I am not sure if measuring toys has much value.

My Revel C52 speaker which I reviewed is not a toy. I have a few hi-fi speakers that I plan to test.

 

[daftcombo]

What's great is: since ASR is measuring speakers, nobody is interested in DAC/amp measurements anymore and we are now focusing on the real thing, where the $$$ should go. Instead of "upgrading" for diminishing or best said: inaudible returns.

Next step will be room reviews though.