Announcement

Collapse
No announcement yet.

Overnight Sensation - Klippel Analysis @ Erin's Audio Corner

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • tktran
    replied
    Well to be frank a speaker can be hand tweaked to a
    +/- 1dB response for the on axis, but to get it replicated on dozens if not hundreds of pairs of speakers, ( regardless of DIY or production ) it’s really hard, if not impossible to do, without lots of re$ource$

    Leave a comment:


  • LOUT
    replied
    Originally posted by Chris Roemer View Post
    Do I think that a "designer" (getting no compensation) should spend time/money on a "redesign" over time as drivers change? No I do not.
    Most of you can tell (approx.) how old designs are by their post dates.
    More often than not, older designs are still being built by newcomers to our hobby.
    Many of them will give an "opinion" as to whether or not the design lived up to their expectations.

    Mr. C. HAS (more than most peeps) tried to stay "aware" of driver changes (I think one of his tweeters ended up with the polarity markings swapped - from the factory).
    He's also swapped out (NLA) tweeters for those more available for many of his most popular designs.
    That's really all that can be expected of us, right?
    Assuming this is all just from the current drivers not sounding/measuring like the older ones of the same model, then at least it means there aren't any popular mistakes being made anywhere else, like during the measuring process (except maybe by PE from what it sounds like by your ND105 experience). That's kind of a silver lining at least.

    As far as expectations, I suppose it depends on whether or not the company or person is using those measurements as advertising, how far from accuracy they end up, which aspects they're charging money for, etc.
    A few examples of my (potentially misguided) expectations:
    -The OS design is freely shared and it doesn't appear like the measurements have changed very drastically (PaulC's earlier comments suggest it's still roughly following his preferred voicing, and there isn't a flood of user complaints or anything like that). Maybe a little frustrating for the designer and any golden-eared users, but mostly harmless.
    -When a modest or large drop in accuracy happens to a paid design that's advertising accuracy or a useable FRD, then the customer isn't getting what they paid for. That can (should?) hurt sales and make future customers a little wary of the brand.
    -If even a freely shared design suffers a drastic enough change where the measurements are way off and/or a flood of user complaints happen, then it might hurt the designer's reputation somewhat if they leave it as-is. I don't know how common drastic changes are though...for all I know it doesn't happen at all.

    Leave a comment:


  • Chris Roemer
    replied
    I designed my "Nano Neos" back in 2011, using Dayton's (then new) ND105-4 4" neo woofer.
    It's advertised T/S parms (at the time) were Fs = 54Hz, Qts = 0.55, and Vas = 0.16 cu.ft.
    (That indicates an approx. F3 near 40Hz in a volume of about 0.3 cf.)

    I had just gotten my WT3 at the time, and the ave. for the drivers I got was: Fs = 63, Qts = 0.49, and Vas = 0.13.
    That indicates an F3 in the mid 50s in a volume closer to 0.2 cf.

    In 2015 Dayton updated their specs to: Fs = 79, Qts = 0.61, and Vas = 0.07.
    That puts the woofer in the high 50s in about 0.15 cf.
    These are still the T/S parms that Dayton goes by to this day.

    Oddly enough, they'll still model an F3 close to 40Hz in 0.3 cf (which is their "recommendation" from "Box Pro" - or whatever software they sell).
    Was that ND woofer ever "redesigned"? I don't think so.
    Were parts/materials ever resourced over the the last 10-12 yrs? It SEEMs likely.
    Hard to believe that the FR-curve hasn't changed some ...

    Do I think that a "designer" (getting no compensation) should spend time/money on a "redesign" over time as drivers change? No I do not.
    Most of you can tell (approx.) how old designs are by their post dates.
    More often than not, older designs are still being built by newcomers to our hobby.
    Many of them will give an "opinion" as to whether or not the design lived up to their expectations.

    Mr. C. HAS (more than most peeps) tried to stay "aware" of driver changes (I think one of his tweeters ended up with the polarity markings swapped - from the factory).
    He's also swapped out (NLA) tweeters for those more available for many of his most popular designs.
    That's really all that can be expected of us, right?

    Leave a comment:


  • djg
    commented on 's reply
    If anyone's that worried about "woofer creep", perhaps they should buy a new B4N(?), find an old one and do their own measurements. Or, make some new OSs and "klippel" them into submission. Maybe someone over at ASR would want to do this.

  • LOUT
    commented on 's reply
    I wasn't sure if shipping the new OS to PaulC and asking him to measure and compare oldVSnew results was too much of an ask (I have no idea how much work this is for his space and setup), but I agree that it sounds like a potentially faster way to narrow down what's causing unexpected differences.
    On the positive side (in this method's favor) I'd think the 0-degree/on-axis gated measurement would be plenty enough...I don't think the high-resolution bass-response nearfield/groundplane? measurements would be super important for this, nor any off-axis results.
    I'd think either the 500hz-10khz gated on-axis closely follows the old results (on PaulC's website?) or it closely matches Erin's Klippel on-axis results (being the same speaker that was used and all), or it does something completely different from either (the spooky result).
    Similarly (if not already the case) it shouldn't hurt for Erin to also try and grab his own 0-degree gated measurement to further sanity-check the Klippel results...once again, if it's not a PITA request...mostly in case of the "spooky result".

  • DeZZar
    replied
    Originally posted by LOUT View Post
    Maybe the woofer from PaulC's measurement is noticeably different than the version the company ships today?
    I believe this right here is a significant consideration.

    The risk for kit designers is they really have no control over quality - no opportunity to adjust design as the various runs of units are made which is something a manufacturer does (should) have control over. As times goes on there is no way to adjust the kit for any change that may occur in the components and no way of making a final measurement on the assembled unit to ensure its performing correctly as it goes out the door.

    What I would be most interested in seeing is if ErinH was gracious enough to send Paul Carmody the sample unit, and for Paul to re-run his own measurements on that unit and provide a comparison to the original design. Particularly if Paul still has the original OS speaker on hand to re-measure as well.

    If the newly made kit and the original speaker measure the same - then its a difference in measurement results etc which can be explored further if anyone cares enough to put the time in.
    If the newly made kit and the original speaker measure differently - then its a change in manufacture of parts and highlights the issue of independent third party supply of parts that are supposed to faithfully re-constitute an original design

    Leave a comment:


  • joshshetter
    replied
    Originally posted by tktran View Post
    Josh,

    By level of dispersion I mean, do you want more or less dispersion in the top octave? And can we say that categorically edge diffraction is always harmful and makes the speaker worse?
    Or just different? (and more expensive to build- which incidentally takes away from resources that may be directed better elsewhere)

    So in your example, does the spinorama tell you which speaker you might prefer? eg. the 8030C, 8030A or 8331?

    The fellow who came first in InDIYana 2022 used polar measurements to simulate the crossover, in fact cobbled up with crossover minutes before the event. In his words " one of the first passives I've built", using simulation only.

    CTA2034A is a tool. But you need also need science, art, resources, opportunity, experience and interest/passion.

    I'm not trying to be provocative, but I've heard both the 8040, 8340A or 8341 and I can tell you which I'd rather buy... and it ain't the most expensive one with the coaxial driver and woofers hidden behind the waveguide, although I was ready to drop my cash down on a pair...

    I would prefer the 8331, it has the least amount of issues. I do have the 8030c, they are nice but they sound small, which to me shows up in dispersion, small speakers tend to go omni earlier. Not sure the 8331 would fair better in that regard,

    I do think diffraction is pretty much always bad and the speakers that have the best imaging IME are the ones with the least amount of it, I.E. genelec/ large roundovers/waveguideded tweeter which basically don't have any.

    As far as dispersion level, I want it even. I don't like blooms in dispersion at certain frequencies because that would send mismatched energy into the room in relation to direct energy.

    I think what one can pull from measurement data depends on their own experience. I have a ton of experience with many different speakers and analyzing them, I think when you can hear things in person and have info to correlate what you hear with, you gain insightful knowledge on how to interpret the data. I've had enough speakers to know that I like neutrality and on the whole I think most people do. I've found my preferences very much lie with Toole's research on listening preferences.

    Personally I don't find speaker design to be much of an art, it's total science for me. You're just doing your best to mate drivers, all the art for me comes from the music the speaker plays.

    Leave a comment:


  • LOUT
    replied
    Originally posted by Paul Carmody View Post
    I don't quite understand why my measurements look so different than his. Also, I'm pretty sure this speaker has been measured by other parties around the internet and not seen anything quite this striking.
    Originally posted by ErinH View Post
    The real problem is that so many DIY designs are based on insufficient methods. It's not necessarily even that resolution; it's the improper merging of NF/FF, simulation + FF or just simply not having a proper FF measurement to begin with. It's more common that people would like it to be. But hopefully when I post data and people say "well, *my* data doesn't look like that" they'll take the extra steps to understand why rather than brush it off.
    Originally posted by fpitas View Post
    Your measurements look pretty different than Paul's. Until we understand that, I'm not sure what to think.
    Would it be possible (making some big assumptions that you both have the free time and the desire) for PaulC and ErinH to discuss or private-chat together about their measurement techniques and tools down to the fine details with each other, then come back here and post if they find some likely causes for any differences between different measurements which shouldn't be present?
    I think ErinH has mentioned doublechecking some Klippel results against his own gated+NF+groundplain measurements which both confirmed the Klippel could match those style of measurements and that he's familiar with performing them. And I think PaulC is very familiar and experienced with all these microphone measuring techniques.
    If there's a way for you two speaker-nerds to sleuth out where some of the more major differences are happening (the ones which aren't explained by resolution differences nor room interactions, since some of these are too wide-Q to hide in low-resolution and I'm assuming the idea of NF+gated is to avoid room interactions), then your findings could help many of us avoid making mistakes.

    Maybe there's an easy-to-miss refinement in NF+gated measurement technique that we can learn about?
    Maybe there's something in the Klippel that would benefit from a patch?
    Maybe the woofer from PaulC's measurement is noticeably different than the version the company ships today?
    Maybe it's none of the above, but something completely different and important?

    Leave a comment:


  • tktran
    replied
    Josh,

    By level of dispersion I mean, do you want more or less dispersion in the top octave? And can we say that categorically edge diffraction is always harmful and makes the speaker worse?
    Or just different? (and more expensive to build- which incidentally takes away from resources that may be directed better elsewhere)

    So in your example, does the spinorama tell you which speaker you might prefer? eg. the 8030C, 8030A or 8331?

    The fellow who came first in InDIYana 2022 used polar measurements to simulate the crossover, in fact cobbled up with crossover minutes before the event. In his words " one of the first passives I've built", using simulation only.

    CTA2034A is a tool. But you need also need science, art, resources, opportunity, experience and interest/passion.

    I'm not trying to be provocative, but I've heard both the 8040, 8340A or 8341 and I can tell you which I'd rather buy... and it ain't the most expensive one with the coaxial driver and woofers hidden behind the waveguide, although I was ready to drop my cash down on a pair...

    Leave a comment:


  • joshshetter
    replied
    Originally posted by tktran View Post

    Only one.

    but let’s remind ourself that what the NFS is a measurement tool. What it does is allow faster and more accurate measurements in 360 degrees, both horizontally and vertically.

    The designer still has to do everything else, pick the tradeoffs he’s willing to live with.

    I think we are jumping the gun thinking that the NFS CTA 2034A output will give us complete insight into what separates great speakers from the very best ones. It can tell you a good from a bad speaker, and show you why, better than a single on axis measurement, or a sound power measurement.

    It does not do that. There’s still a lot that the CTA2034A doesn’t tell us- like non linear distortion, level of dispersion, timing, diffraction, location of radiators etc

    The designer has to make all those choices.
    Klippel shows you dispersion, not sure what "level of dispersion" means, diffraction would show up there. The 8030c review on ASR shows a 3d view of the speakers radiation that does show you where drivers are located.

    Leave a comment:


  • tktran
    replied
    Originally posted by djg View Post
    What currently available kit/design has been developed with Klippel analysis?
    Only one.

    but let’s remind ourself that what the NFS is a measurement tool. What it does is allow faster and more accurate measurements in 360 degrees, both horizontally and vertically.

    The designer still has to do everything else, pick the tradeoffs he’s willing to live with.

    I think we are jumping the gun thinking that the NFS CTA 2034A output will give us complete insight into what separates great speakers from the very best ones. It can tell you a good from a bad speaker, and show you why, better than a single on axis measurement, or a sound power measurement.

    It does not do that. There’s still a lot that the CTA2034A doesn’t tell us- like non linear distortion, level of dispersion, timing, diffraction, location of radiators etc

    The designer has to make all those choices.

    Leave a comment:


  • invaderzim
    replied
    Originally posted by Paul Carmody View Post
    <snip>
    I think the hope of the Klippel spinorama has been "FINALLY we can measure what makes some speakers sound great. We can finally quantify all of it." But I don't think it's achieving that goal. <snip>
    I think Erin gets closer to doing this but typically still the measurements are done to show what is supposedly wrong with a speaker instead of looking at what makes it appealing sounding. I think most of that is because the ones spending the huge money to buy this equipment are more interested in products that measure well than looking at why people might like some products with questionable measurements. The designers that aim for a sound aren't going to spend 6 figures to try to remove that sound from what they design.
    I do have to say that I enjoy Erin's work because he doesn't put people down for liking things that don't measure well. But it is pretty clear that some people don't enjoy 'perfect' as much as others and many aren't willing to tolerate unpleasant reproduction even if it is accurate.

    I've moved enough speakers into different rooms and places in those rooms to know that one that sounds good in one place can sound bad in another. And while Erin has said that the Klippel removes the room so you can then extrapolate what the speaker will do in your room, I haven't seen anything on how to do that step. I don't have a dedicated listing room; I listen to music in our living room. I'm not going to cover our walls with sound absorbing panels; I just don't like the way they look for everyday living so my room will always affect the sound. The room correction I tried made the music sound, perhaps, cleaner but I didn't listen as often or as long with it in place. Ironically, for the first 30 seconds or so of listening I was impressed so in a quick comparison I would have picked the corrected sound, but I have to go on what makes me want to listen more.
    How do in-room measurements compare back to the Klippel? Obviously if my room has a peak where the speaker has a dip or vice versa then a non-perfect speaker may sound better to me.

    Leave a comment:


  • JRT
    commented on 's reply
    Some of that depends on what you get accustomed to in listening to playback of recorded music, and the type of music.

    If you listen to live music from unamplified acoustic instruments performed in a small venue, piano, trumpet, drum, clarinet, etc., (eg. some live small group modern jazz) and if you soon after listen to well engineered recordings of similar music, then you may begin to dislike unatural sounding playback.

    If instead the recordings are not well engineered and/or the music has a lot of synthetic content and/or is processed with various DAW plugins, then the recording is not realistic, so likewise any playback, and any EQ or other processing in playback is just tailoring the resulting sound to personal preference, and there is nothing wrong in preference.

  • djg
    replied
    What currently available kit/design has been developed with Klippel analysis?

    Leave a comment:


  • tktran
    replied
    I think there's some cognitive dissonance going on here. And I don't mean that in a negative or pejorative way. I just think you're trying to reconcile what you're seeing from Erin's measurements which A) contains full 360 degree polar measurements B) has a razor sharp resolution of under 2Hz due to Klippel NFS, with what you're used to seeing, which is 1. on-axis measurements, and/or a few off axis measurements, with a resolution of around 200Hz and 2) what you're hearing, which sounds good, if not great, for a small speaker.

    So how can something that measures so irregularly still sound good, you're asking yourself?

    Maybe the measurements are wack. Maybe the built speaker is wack. Maybe the drive units are wack (off spec after a decade)

    But I bet they do sound good. And I don't think Erin is saying they sound terrible. And he's not saying they should be flat
    (what is flat anyway? is +/-3dB flat, or +/-1 dB flat, or +/- 0.1dB "flat".
    And are we talking on axis, off axis, listening window, power response, in-room response etc)

    As a long time follower of your work Paul, I respect everything you say.

    What I'm trying to say is that measuring and designing with a full 360 degree field is like.. well, remember when we all went from designing speakers WITHOUT measurements, to designing WITH measurements?

    It was a Eureka! or total lightbulb moment? "OMG I see now" "Before I was in the dark, now I can see the light"
    "I can't believe I've been doing it without measurements?" (I know I can't, but I know people who can/did eg. Wolf)

    Well with 360 degree measurements, it's the next big step IMHO.
    Everything you say about designing with measurements as well as ears and making sure it translates, and experience matters etc...
    well it only gets better/easier/faster/more intuitive from here with 360 degree measurements....

    I think you have nothing to lose but to try it!

    Leave a comment:

Working...
X