it was fun to follow how this thread has been slowly but inevitably turning into "Loudspeakers and room as a system vol. 2" or even "The Objectives of a Loudspeaker in a Small Room vol.3"
years are going by, and we are still groping, though some of us have their minds made up on the issue
I wonder why - after reaching their end of the road - they still want to join discussion? What for? To preach?
and they seem to forget that this is NOT scientific discussion group
this is a forum for amateur hobbyist and sounds good to me is essential to this hobby
years are going by, and we are still groping, though some of us have their minds made up on the issue
I wonder why - after reaching their end of the road - they still want to join discussion? What for? To preach?
and they seem to forget that this is NOT scientific discussion group
this is a forum for amateur hobbyist and sounds good to me is essential to this hobby
Last edited:
Of course I mean "no smoothing"... not even 1/10th octave smoothing.
Certainly you are correct with regard to most commercial manufacturers... but I have seen many others post significant amounts of data with absolutely no smoothing applied.
Our hearing is certainly not 1/10th octave averaged.
I haven't heard your speakers and I'm sure they are really quite good compared to many others if not most others.... but certainly your speakers are not "perfect" unless we should rely on 2 graphs with 1/10th octave smoothing and some subjective evaluations.
- The warts would be stuff that shows up in a free-field un-smoothed frequency response plot vs 1/10th octave smoothing.
- More data would be group delay and distortion measurements.
For uninhibited objective evaluation, comparison and discussion of course. 🙂
yours is the perspective of the groping but there is just no point in discussing anything for people who have their minds made up on the issue many years ago
By groping I hope you mean, "showing or reflecting a desire to understand, especially something that proves puzzling" and not the more common definition, "moving or going about clumsily or hesitantly; stumbling" 😱
In my relatively limited experience, the un smoothed data can show anomalies that are otherwise hidden and these are especially telling with regard to cross over design and potential improvements that could be made within that region.
100hz is nearly 1/2 octave at and around middle "C". Very important region I'm sure everyone would agree.
Last edited:
I clicked on "Abbey" above and got a very confusing web page. The upper graph has three curves on it, with nothing telling me the difference in the three. Toward the right there's an instruction to click and drag the mouse and something will update. I couldn't get that to work (I have a 3 yr old PC with Windows XP sp2). No answer to either of my questions... 😕
actually - both, the pun was intended
I was alluding to bentoronto's post
but for those with their minds made up on the issue we are certainly "moving or going about clumsily or hesitantly; stumbling"
Having a look at that page poses a couple of questions for me.
The most obvious of which is the screen shot of frequency response and polar plots shown for the Abbey at the bottom of the page do not match the frequency response and polar plots from the interactive applet linked just above it on the page - not even close.
The screen shot version shows a response taken 16 degrees off axis that is tapering down by 6dB from 200Hz to about 12Khz. And yet if I chose the same angle in the applet the response is more or less flat from 200Hz to about 8Khz, and very different to the screen shot.
So which is correct, and why such a big discrepancy ? Different measurement environments or different prototypes ?
The second point is that the flattest response is 16 degrees off axis, (conveniently chosen in the screen shot version) with the on axis response having about a 3dB peak at 2.5Khz, and a 6dB dip at about 5.5Khz, a pretty huge error for such a small change in angle.
The accompanying description says "The pronounced "hole" on axis does not impact its high performance at 15-30 degrees off axis - it's optimal listening angles". Well I guess that makes it alright then! 😉 Just don't listen to it on axis!
For someone who seems to be so strongly focused on "measurement is king and flat (on axis) is right", trying to dismiss such a poor on axis response in that manner seems unusual to say the least.
Either the irregular on axis response is deemed to be an acceptable tradeoff in that particular design and application, (I'm guessing it's a sound reinforcement speaker, not an audiophile one) or the design is flawed.
What also seems apparent from the graph is that the on axis notch in the response from the wave-guide is far worse than the summed response suggests - presumably the black line on the interactive graph is the wave-guide response measured by itself - if so the on axis drop off from 2Khz to 5.5Khz is a staggering 15 dB.
To try to make up for this the woofer is not completely rolled off, (again assuming the red line represents the woofers individual response) but is instead shelved off to -12dB at 600Hz+ so that the woofer still has sufficient output in the treble to help fill in the hole. This means in the mid treble region the woofer is the primary source of on axis treble.
I hate to judge a speaker from graphs alone, but I honestly can't see this sounding good at all. At the very least imaging is going to be very poor and inconsistent due to the incomplete rolloff of the woofer and the huge hole in the wave-guides response allowing the woofer to "show through" at a frequency range which it is ill equipped to perform well.
Not to mention the huge upper mid/lower treble response variations with small angular changes, and the fact that because the woofer is only 12dB down over most of the midrange/treble there will be significant interference and lobing between the drivers, which whilst not showing up obviously in the summed frequency response, will allow any cone resonances etc of the woofer (which are surely present on a 12" at those frequencies) to be clearly heard.
Very odd indeed...
Last edited:
The graphs and "user interface" on the linked page are actually just a screen shot. To use the interactive simulator on the link above it (from which the screen shot is taken) you need to have Microsoft .net installed and use Internet Explorer.
Thanks for the vote of confidence! 😉
Of course they aren't the whole story, isn't that what this thread is about?
It might be a good time for those who follow this thread to look at the upper left hand corner of this page. What does the masthead say? DIY Audio.
Not diy Super Hifi, or Professional Loudspeaker Designer Lounge, nor Acoustic Engineer's Clubhouse. DIY
Projects here range from "Stuff a few drivers in a box and party" to "I want a world class system - price, effort and time no object." Many here do not have the skills or equipment needed to get all the measurements right. So some by ear tweaking is the most practical approach for many members of this forum. And guess what? It actually works.
I have nothing against measurements, I use them every day. I've helped a number of forum members learn how to measure. Over the years I have heard a lot of home made speaker projects, many of them suck. The builders would have done well to learn measurement techniques to at least get in the ballpark. Then tweak by ear.
If someone has a technique that works for him, fine. Want to share it here? Great! But to claim that there is only one right way to achieve good results? Well, that dogma will bite you.
amen!😉
problem is that the dogma has some avid followers here and that they never miss any occasion to express it - I mean to spit it in the face of the groping
In Earls defense, I don't think he has ever said that. Yes he is very trusting of measurements, but my understanding is that he does NOT feel that on axis is all that important. There is much more going on than that.
I guess my point was, in Earls book measurement is king and tweaking by ear (or voicing as it's generally called) is invalid.
And yet the amount of actual change applied from measured "flat" that would typically be done on a good design when "voicing" it, is in the order of 1dB or so, possibly less, and far less than the response anomalies in that particular design, making his claim of voicing being invalid a moot point.
There is so much more than that and the axial response is not very important - the listening axis is somewhat important, but even that is not everything.
In the top curve the black line is the DI and the red line is the power response. The power response will not change with angle, but the DI is always referenced to the choosen angle because that is the way it is defined. A "good" loudspeaker has a near flat and smooth listening axis response and a DI that is likewise. Since my speakers are always optimized about then 15-25 degree axis this is where the best combination of these curves will reside.
There are differences in the data sets shown in the Abbey page and the applet. They are different speakers taken at different times and use a slghtly different analysis. The applet is the more recent data set and analysis technique.
As to smoothing, the poster is incorrect in that the ear does have about a 1/10 octave bandwidth resolution at higher fequencies and as Markus pointed out it is wider than that at LFs. The use of constant bandwidth FFT data is not something that I am ever going to post because it is wrong. The ear just does not hear that way and there is no point in showing measurements that are knowingly incorrect.
As to group delay, I will agree that a view of the actual impulse response should be shown as this will show any effects of group delay. But since numerous studies (Moore, etc.) have shown that group delay is inaudible, simply ploting it as a function of frequency is not meaingful either.
As to "distortion" I assume the writer means nonlinear distortion. There is no one in this business that believes that things like THD or IMD or whatever tell us anything about the sound quality. If the writer believes otherwise then I suggest that he read the literature as this fiction has been dispelled a long time ago.
There is one set of data that I have not gotten arround to showing and that is transient thermal measurements (aka "dynamics"). I have done some of this but not enough to be confident that the data is as meaningful as it appears.
So basically I contend that the current data presentations are as complete as is currently done anywhere that I know of and I will stand by them as they stand today. I see no reason to post lessor accuracy data just because thats what someone wants to see. Those plots contain analysis and data presentation that is far beyond anything else that I know of and until such time as I am convinced that a better technique is available I will stick with what I have.
Last edited:
That is a really nice tool... the Behringer shows very nice measurements. I wonder which one has been measured?
I also wonder which Orion version was was measured on the backside?
Don't you just love it when someone offers you information, BUT, you've got to download and install some program, and have microsoft.net and microsoft browser installed and active (I use Mozilla so fewer spams and virus type stuff)... The trouble is, you never know what some additional program is going to do to your computer. I've learned the hard way to avoid those. If someone can't show and/or explain their argument in a simple way.... no thanks.
I went back to the Geddes website again to see if I had missed something. Nope. It's as bad as I thought it was. He's not very good at presentation/communication. I think the reason his speakers MAY be pretty good sounding is because they apparently have controlled directivity, which is good for retaining imaging info from the original recording, rather than stimulating the listening room acoustics (especially side wall reflections) to create an often effective but false sense of spaciousness. I also wonder if his plastic cabinets don't emit energy that fills the room with delayed de-corrolated energy that may also create a sense of spaciousness.
I have no idea what you are saying. Do you care to clear it up for me?
The speaker is not "voiced", but it is not flat either. There is, as we have discussed at length in this thread and others, a gradual falloff in the high end response. This was done after several years of comments and investigations and it is now felt by almost everyone in the industry that a flat CD speaker will always sound bright. Call this "voicing" if you like, but it was not done based on my listening - it was quite analytic.
A look at my comparison data will show that there are no loudspeakers out there that are flat to 1 dB. I do not know whose data you are looking at but mine says otherwise.
Thanks - its a very complex data analysis that I want to write-up someday, but there are some people who are interested in using it and so I am going to wait to see where that goes. As I said, no one has ever done analysis like that before.
The Behringer is the same one that was used in the Orion vs. Behringer listening test. In fact both of these speakers are the exact speakers as used in that test. The Orions were directly from and made by SL. I am sure that someone knows the Behringer model, I just don't have it handy.
And yes, I agree the Behringer is top notch. And it gets very good reviews (even from SL himself). My only problem with it is the limited dynamic range.
There are many ways to solve any problem, that is quite correct, and I am sure a monkey could "tweak" a loudspeaker to become a masterpiece given enough time, but there certainly is only one optimum way to get to this result and I do not believe that it involves listening tests.
Its certainly NOT the fun DIY way, thats for sure. It's tedious and data intensive, but it yields high quality results the first time virtually every time. Its not for DIY, no, but we have been down that road before.
- Status
- Not open for further replies.