Any abandoned nuclear power plants near you? The scale of this whole installation amuses me.
https://www.stereophile.com/content/nwaa-labs-measurement-beyond-atomic-level
NWAA Labs' "turbine room," which is 650' long, 350' wide, and 80' high—large enough to hold four NFL football fields—is a true free-field space. . . . it's so large that there's no need to gate impulse responses in this space because reflections are so attenuated after 160ms.
. . .
John was fascinated by the room's huge, curved microphone array, which contains 19 matched Earthworks M30 (ANSI Type 1) measurement mikes. The array is 4.1m in radius and is suspended 15' above the floor. Microphones are mounted within 1/16" of a central point, one every five degrees. Each mike is electronically compensated for distance to one sample at 48kHz.
. . .
John asked whether at a 4.1m distance it was possible to measure a large speaker and get proper integration of drive-unit outputs. Ron replied, "We know what our limits are. The spacing of drivers cannot exceed 48". . . . The big problem with loudspeaker measurements in general is you need to be further away than the largest dimension of the loudspeaker. That's fine for an LS3/5a but impossible for a significantly larger loudspeaker. . . . We can measure subs all the way down to 25Hz when we support them above our 5'-tall wedges.
You are safe from pages and pages from me given I will be out this evening. Wouldn't you be better off looking to zoom with those with a more positive interest in the project rather than those chatting on a thread they find interesting because of bafflement. There is a whole forum dedicated to single driver speakers so my being baffled by what some DIYers value highly likely doesn't mean much.
Typical cost for engineering labor hours would be about $200/hr. To have lab work done, you also have to rent the lab, which can vary all over the place. At one extreme, renting a transonic / supersonic wind tunnel can be $50k per run or more. At the other extreme, having IR temperature surveys done on a small portable electronics device is so cheap it would be just rolled into the labor cost of the technician. Klippel NFS is obviously in between those two extremes. I think Patrick Bateman's estimate is about right, assuming the organization doing the testing intends to pay competitive salaries/benefits and be somewhat profitable.
j.
There is another reason and that's called marketing.
If they even would be even able to sell this for 1/10 of the price, people probably wouldn't be interested.
That's just the weird psychology of selling a product.
There are seriously interesting books written about this, a must read for everyone to understand and mostly be baffled about the world around us.
$100k is for a company like Tymphany, Sony, Sonos or Samsung absolutely exchange money.
If the goal is to understand building speakers better, I would rather advice spending that amount on courses or hack even a university program in a relevant field. Depending the country you're in, that can even be done for less than half that amount.
Because I find bafflement is better cleared up through in-person discussions, rather than sniping from a keyboard.
That might be true for a luxury items like watches, handbags, etc. I just don't see it for a highly technical product like this.
The system costs what it does because the company has to recoup the significant R&D costs from the relatively few worldwide customers, and it has to do it relatively quickly before technology advances enough that a competitor can offer something similar for half the price.
Exactly as Patrick Bateman described....
a) I was under the impression we were talking about accuracy in the 300-1000Hz, or maybe a bit lower, say 200Hz minimum.
b) Are you suggesting NF is not good enough for subs or anything between 25 and 200Hz? (yes: also depends om driver size, but for any 10-12")
{c) I do not build floorstanders anymore, but that is irrelevant for the discussion}
That be the measurement boggy-man! .... How to get into the far-field and not need to gate.
Breaking that traditional need to measure speakers in the far-field in a reflection free environment, is imo, what's new about Kippel's NFS.
The NFS can take enough precise measurements in the near-field, where inverse-square law distance attenuation has yet to take place, and where phase has yet to settle into place.....to be able to mathematically extrapolate them into far-field response.
This extrapolation to the far-field is not the simple up close merging of driver sections' gated measurements....as best i understand the Klippel interview vid with C. Bellmann. If anyone who has watched the entire vid can correct my impression that Klippel is doing something entirely new. please do so, and i will say big thanks!
Ime, very close gated measurements of driver sections merged together do not give the same results as far-field measurements.
They make look very close in the frequency domain, but not so in the phase domain. And the phase domain goes on to dominate how well xovers then work.
When i compare such up close measurements against outdoor far-field ones, I end up with a speaker that measures one way indoors up close, and another way outdoors in far-field.
Anyway, I really think NFS is very cool and quite an advancement in measuring.
If someone had one within a half day's drive that could handle a large synergy horn, I'd probably haul one over, and be willing to spend up to $300-500, just to get some kind of verification/comparison to the outdoor spinorama measurements i make.
For pricing ideas, I don't know how many opt for it but ascend acoustics has an option on their Sierra LX speakers to run the ones you buy on their Klippel for $250 each. They say each one takes about 6 hours.
Greetings all,
since the content has already deviated form the question asked, can someone explain the reason for the measurements?
As I understand it based on the discussion, perhaps incorrectly, the Klippel measurement (i) mitigates the influence of the environment (room?) and (ii) provides a far field data.
A DIY person designs a loudspeaker, implements the design and optimizes parameters for one's listening environment, e.g., based on measurements. Assume that the DIY person obtains accurate Klippel measurements as described in the first paragraph. So, I understand that the measurement may be beneficial at the implementation phase, e.g., deciding on relative location of the drivers, crossover frequency and slope, and the like.
But, how do the measurements help with what I termed the parameter optimization, as now the DIY person is faced with the unknown of the listening environment and, consequently, different loudspeaker response.
Am I missing something?
Kindest regards,
M
since the content has already deviated form the question asked, can someone explain the reason for the measurements?
As I understand it based on the discussion, perhaps incorrectly, the Klippel measurement (i) mitigates the influence of the environment (room?) and (ii) provides a far field data.
A DIY person designs a loudspeaker, implements the design and optimizes parameters for one's listening environment, e.g., based on measurements. Assume that the DIY person obtains accurate Klippel measurements as described in the first paragraph. So, I understand that the measurement may be beneficial at the implementation phase, e.g., deciding on relative location of the drivers, crossover frequency and slope, and the like.
But, how do the measurements help with what I termed the parameter optimization, as now the DIY person is faced with the unknown of the listening environment and, consequently, different loudspeaker response.
Am I missing something?
Kindest regards,
M
That's what I also used to think.
But I like to approach things on a objective way and there has been plenty of research done on the field of marketing.
"Technical products" work on exactly the same way.
The best known example of this of a guy who had a very unique and great idea of a technical medical device.
Couldn't get it sold at all, until he raised up the price significantly.
We always assume that tech people are so special and unique.
Will never fall for such things.
Again, please dive into the science, I would like to hear and see proof otherwise.
In fact, I bump into many engineers on a very regular basis that really believe to have to buy that new LeCroy scope.
The emphasis is here on believing, since the majority of those engineers don't measure anything substantial.
Let alone using the features of such a scope.
I get it it, toys are fun, even more so when they look fancy and well engineered.
But often you will get the same results with a lot less.
THAT's a legitimate use-case for the NFS. (..though the designer also has to remember that cardoid (or dipole) below about 200 Hz in most rooms only extends a few inches from the baffle.)
Also of course Erin's is where a well defined repeatable testing platform that can handle full polar measurements pretty much un-attended is as well.
1. Sufficient Absorption (Ironically "looser" cheap fiberglass batting is better at lower freq.s - you just need enough thickness and that can include a thin air-gap between batts): walls, ceilings, and floors (with the loudspeaker elevated to accomidate the floor absorption).
2. MLS
3. "Tailing" the low-end.
4. Modest expectations. (..in search of useful, not perfect.)
That's one of those times I would NOT opt for the NFS (even if I had access to it), I just don't think the NFS could properly measure a large deep horn.
Have been chewing on this thread for some time. For Spinorama type full scale system evaluation, the advantages of the Klippel NFS are obvious. No discussion there.
But for system design I am more in doubt. What is lacking in current diy measurement systems in accuracy in the 200-1000 Hz range. That will require a large hall or outdoors measurements. Or is that assumption already wrong?
But apart from indeed cardiodide systems, why do we need to take measured diffraction into consideration? The harm is then already done: the prototype is built. Furthermore, the degrees of freedom are very imited with conventional box type systems. I.m.o. advanced modelling software before the actual build looks more helpful for multiway system development. DonVK's experiments and Mabat's AT4H thread really show the way: and so do Augerpro's own waveguides.
And then this: To what extent may off axis modelling (200-1000Hz) based on an on axis measurement, driver size and diaphragm characteristics etc, offer a viable alternative? I would like to hear your views on these matters.
But for system design I am more in doubt. What is lacking in current diy measurement systems in accuracy in the 200-1000 Hz range. That will require a large hall or outdoors measurements. Or is that assumption already wrong?
But apart from indeed cardiodide systems, why do we need to take measured diffraction into consideration? The harm is then already done: the prototype is built. Furthermore, the degrees of freedom are very imited with conventional box type systems. I.m.o. advanced modelling software before the actual build looks more helpful for multiway system development. DonVK's experiments and Mabat's AT4H thread really show the way: and so do Augerpro's own waveguides.
And then this: To what extent may off axis modelling (200-1000Hz) based on an on axis measurement, driver size and diaphragm characteristics etc, offer a viable alternative? I would like to hear your views on these matters.
The other side of the coin is, what resolution of measurement do you feel is accurate enough.
I use 1/24th per octave as standard in my crossover designs, but consider 1/12th octave as a bare minimum, which is the same as the equal temperament scale.
Now, between 500Hz and 2000Hz is only 1500Hz mathematically, but a full 2 octaves. (each doubling of frequency is an octave)
If I want
1)1/24th octave resolution that is 1500 / 24 /2 = 31.25Hz.
If I want a 2) 1/12 octave resolution, that is 1500 / 12 / 2 = 62.5 Hz interval.
1) 31.25Hz needs a 32ms time window (T = 1/f)
2) 62.5Hz needs a 16ms time window.
With mic and DUT @1m, that means you need to be 1) 6m or 2) 3.22m from any surface.
Reference: https://mehlau.net/audio/floorbounce/
So yes, get your speakers and microphone up high up above the ground. 10ft is good, but 6 metres is better. Now do your spinorama. Try not to drop your speakers, or worse, hurt yourself.
Last edited:
You ask for proof from your counterparts and offer an anecdote as ‘fact’.
Being so apparently immersed in ‘the science’, how about you offer up some proof on your position first?
Last edited:
That is precisely why I think that advanced modelling would be an easier way to go than circus type 5-degree up-in-the air measurements.
There is a merging of sorts of gated measurements for the high frequencies in most cases as Sound field separation is increasingly difficult as frequency rises and requires exponentially more measurement points to keep the error down.
A quote from the DIY Klippel thread, the reports in the first page detail a lot of information on how this all happens.
"SFS and time gated impulse measurements complement each other very well. One works well at low frequencies, and the other works well at high frequencies. Klippel recommends a dividing line of 2-3 kHz."
3D modelling with ABEC/AKABAK prior to building can reveal a lot about the directivity of the proposed speaker and is a good way of comparing different concepts. You could take it further and model the inside and any ports to see if the layout is presenting any issues. Most of this could be worked out from first principles to avoid the biggest mistakes but not everyone knows all the things to look for. Even with that level of modelling odd things can happen that weren't expected and these seem to have a funny habit of showing up in the lower mid frequencies where short gating has reduced accuracy. As they tend to be resonances or other unexpected interactions close field measurements around the box, burst decay and impedance measurements as b force suggests can be used to see if any of those nasties are hiding there. If those measurements are clear then there is no real need to worry about getting a higher resolution look because anything else is very predictable.
Erin measured an SH50, that is a pretty big deep horn, not quite as big as Mark's, the results were accepted by Danley so I see no reason to doubt their efficacy.