Well, it's serious math to me 🙂
My undergrad math minor is long forgotten, and a couple of courses in differential equations is as far as i got in calc.
No, i do think one can begin to make measurements with the precision of the Klippel.
The key is precise location, precise distance increments....necessary for the precise math/programming 😉
Again, i suggest watching the first vid in #21 all the way through.....
The key is time vs convenience.
It's not hard to make a very simple DIY jig to be within ±1-2mm and a degree or so precision.
More than enough for these kind of things.
But in the end the question should be, how much more you really gain with it in making a better sounding speaker vs a good looking graph? I ask and say that even from a professional point of view, since I had many of these conversations as well among other professionals. Incl many who VERY seriously looked in buying a Klippel system.
If that is even a goal!
Because the majority here seems to forget that in most commercial cases there are other things (much) more important than that.
Mostly it's form (or looks) over function. Even for many DIY'ers this is quite important (without them even knowing).
Or in other words, the form is already set in stone, and you just have to make the best out of it.
Hey all, has anyone ever taken a complete set of polars, 360 degrees, outside on a spinorama?
Both horizontal and vertical?
In the far-field ? (which IS the valid distance for traditional anechoic speaker measurements)
I haven't. I take far-field polars, but only for a limited listening window. So I'm missing alot of room response.
Complete 360 polars, devoid of reflections are just too tough...besides, it's so damn hard to keep a far-field test setup adequately steady.
Both horizontal and vertical?
In the far-field ? (which IS the valid distance for traditional anechoic speaker measurements)
I haven't. I take far-field polars, but only for a limited listening window. So I'm missing alot of room response.
Complete 360 polars, devoid of reflections are just too tough...besides, it's so damn hard to keep a far-field test setup adequately steady.
You only need 0-90 degrees front as well as 0-90 degrees back (by flipping the speaker) ?
The 90 degrees can be even skipped by one of the two, although is very handy to use as a reference.
The rest is copy-paste.
Unless your design is (very) asymmetrical and unpredictable.
Which wouldn't be my advice to begin with.
Vertical is a different story. That being said, there are a lot of unknowns and discussion when it comes down to vertical directivities.
The truth is that you don't even need that many angles at all to get a good sense of the directivity of a speaker and how constant that directivity is.
We take the information and hopefully model a device to house/hold our ideas and selection of materials and test out the DIY ideas. If a more accurate test produces a better result I'm all for it. The question is how many times do you have to test the same idea? I liken better testing to better results.
I build a cabinet, I choose my drivers, wire, XO parts and I build a second pair of speakers too. Do I send that pair in for testing too? Do you get where I'm coming from? The point of the second test is to confirm the first? I think the quality of the questions we ASK are very important as well. Why am I doing all this testing after the fact? Did I goof and can I hear an issue that is the Cabinet/speaker? That would be the only reason for me to run additional (hard core) testing.
Getting from here to Mars is just 1000 corrections away. I'm sure there is a straight shot to get there, I'll put my money on the 1000 corrections along the way.
If I was producing commercial speakers how many times would I retest a product? You put that product in 10 different rooms with 10 different setups. If the room is 50% of the sound you can't build that into the cabinet or speaker design. We can make the design more versatile though.
I don't want the perfect speaker. I want the speaker I can make perfect in my rooms. That is when I finally quit looking.
Perfect for me is a turn of a dial for my wife. She will move a speaker, what do you do about that?
I been using tape for 35 years. She moved that to, being a smarty pants.
I build a cabinet, I choose my drivers, wire, XO parts and I build a second pair of speakers too. Do I send that pair in for testing too? Do you get where I'm coming from? The point of the second test is to confirm the first? I think the quality of the questions we ASK are very important as well. Why am I doing all this testing after the fact? Did I goof and can I hear an issue that is the Cabinet/speaker? That would be the only reason for me to run additional (hard core) testing.
Getting from here to Mars is just 1000 corrections away. I'm sure there is a straight shot to get there, I'll put my money on the 1000 corrections along the way.
If I was producing commercial speakers how many times would I retest a product? You put that product in 10 different rooms with 10 different setups. If the room is 50% of the sound you can't build that into the cabinet or speaker design. We can make the design more versatile though.
I don't want the perfect speaker. I want the speaker I can make perfect in my rooms. That is when I finally quit looking.
Perfect for me is a turn of a dial for my wife. She will move a speaker, what do you do about that?
I been using tape for 35 years. She moved that to, being a smarty pants.
Yes. If the service was available, I would pay $500 + shipping to measure my recently completed project. Although I would actually have my textreme system done first, and since it is a small 2-way midrange tweeter system, I would expect it would be a bit cheaper.
a big part of me resonates with some of what you say, which i take as, we need less to discern things than we imagine....
That said, i seriously question the measurements we routinely make as DIY'ers........are all that good.
As best as i can tell from what folks post, I've taken far more outdoor quasi-anechoic measurements than most...waay far more really.
And with the most careful best practices I've been able to come up with, mine still vary a lot.
Well, so what that they vary? What is their purpose anyway?
For me, the measurements are what I can tune to....what minimum phase driver/box response variations i can correct, and achieve smooth directivity summing post xovers.
The directivity stuff is pretty easy really, because that is predominantly a function of the acoustic design, and not subject to much tuning other than picking xover points and orders. Especially simple for me, since i use steep linear phase xovers and can cross where ever directivity smoothness demands.
The minimum phase stuff, on a driver by driver basis, is not so simple. It's damn hard for me to get totally repeatable measurements...even in the far field outdoors.
Averages become the reference, simply because i don't know which meas to tune to...
And again, so what?
Well, my experience is the measurement I end up using to correct, matters to the point of magic found vs no cigar.
So personally, I'd love to know the repeatable truth.
Not that i think the Klippel NFS could provide the measurements for what i like, large synergies...i don't think they big synergies can fit...
However, that's just a limitation of the NFS machine's size/capacity.
But my point in all this is, i think you are too casual in assuming the DIY crowd can, and even more so does, make good quasi-anechoic far-field measurements to tune by. It ain't easy!
And despite what you seen to think, i think Klippel has something new with the NFS...again, how about watching the vid.
If you do so, I'd be interested in your further take on things ...but really, only if you do so...
Last edited:
Winner winner chicken dinner.
Finally we have a response in the “I see value in the anechoic-chamber quality measurements without needing anechoic chamber”
@bikinpunk
1. Donation coming. Please advise of (Transfer)Wise account, otherwise Paypal eats up 3%
2. Klippel Near Field Scanner designed Purifi PTT6.5X-04-NAA (Aluminium cone) 2-way stand-mount speaker is coming. In easy to build cabinet for DIYers.
Design plans available by donation.
100% of proceeds to Erin Hardison & family.
Last edited:
@mark100
Keep in mind that not everyone here is just a "DIY'er"
Second to that, a lot of professionals do very similar things.
Trust me, I have seen quite some "kitchens" by now.
Some produce and manufacture some very well measuring loudspeakers.
I don't know why your measurements aren't repeatable.
There must be something wrong with the gear or so.
As you can read in my previous posts, I am very well aware of the capabilities of NFS system for many years, since I had the opportunity to discuss this system many many times with other professionals as well as clients.
Waaaaay back we even had people going to their facility for other type of courses, when they were still busy developing all of this.
Keep in mind that not everyone here is just a "DIY'er"
Second to that, a lot of professionals do very similar things.
Trust me, I have seen quite some "kitchens" by now.
Some produce and manufacture some very well measuring loudspeakers.
I don't know why your measurements aren't repeatable.
There must be something wrong with the gear or so.
As you can read in my previous posts, I am very well aware of the capabilities of NFS system for many years, since I had the opportunity to discuss this system many many times with other professionals as well as clients.
Waaaaay back we even had people going to their facility for other type of courses, when they were still busy developing all of this.
One thing I don't think has been mentioned plainly yet is that the math they are doing isn't just a straightforward distance/SPL calculation. Since they are not in the far-field, in many cases the output from different drivers has not fully integrated yet at the microphone position. So they have to predict how the potentially non-stable sound field will develop at different distances.
Also, in the youtube video they talk about using the spaced spherical scans to determine the direction the sound is traveling, so reflections from outside can be removed.
Last edited:
I believe you. With a great deal of effort, I can achieve a 10 ms gate window on full polar measurements, but only with a small speaker I can elevate quite high. A 10 ms gate provides resolution down to 100 Hz in theory, but only in a 1 octave resolution. This is not very useful. 1/3 octave resolution at 300 Hz, and 1/6 octave at 600 Hz. I consider 1/6 octave to be about the minimum to be useful.
When the near field technique is used to measure a woofer to get the 2-pi response, the resolution can be quite good at low frequencies, but there is an upper limit. With a typical 8" woofer, the upper limit is about 700 Hz.
So if I can achieve 10 ms gate window, I can merge the near field LF response with the gated response at ~ 600 Hz, and I will have acceptable resolution across the band. At high frequencies and low frequencies I will have very good resolution, but the region from 300 - 1200 Hz I will have a barely acceptable 1/6 octave resolution.
Now what happens If I am using a 12" woofer? now my near field LF measurement becomes invalid at 440 Hz. I have to make my merge at no higher than 400 Hz. My 10 ms gated measurement gives me a 1/4 octave resolution at 400 Hz.
What if my speaker is too large to lift 250 cm into the air? What if the best gate I can get is 5 ms? At my 400 Hz merge, I have a resolution of 1/2 octave, which does not provide much confidence. In a situation like this, the merge process can become a judgment call, an educated guess on how to stitch together the near field response and the overly smoothed gated response.
So yes, I see a great deal of value in a true anechoic measurement down to low frequencies. An anechoic chamber is one way to accomplish this. Klippel NFS is another way.
j.
To know the advantages of doing something differently, one needs to understand the limitations of the current methods or processes.
Well articulated Jim.
Well articulated Jim.
Ok, I watched the video in post 21. While taking the spinorama in the near(ish) field will certainly reduce the importance of the room effects, the Klippel indeed seems to do more to "remove the room". The trick is that they move the microphone, but they do not move the speaker. This means the interaction between the speaker and the room is always the same. It seems this allows the Klippel system to build a model of how the speaker interacts with the room, and subtract this from the data in order to "remove" the room.
All in all, the Klippel really is "just" a combination between a robotic system to precisely position the microphone around the speaker, and a bit of software that processes the data. I don't think what is going into making such a system is worth 100 k$, and it would be possible to DIY such a system. However, I don't have the motivation, need or time to go this way. As long as there is no competition, Klippel will be able to sell it at whatever price the big players are willing to pay.
..realizing this thread's purpose is moot given the real cost of an NFS.
It looks like coordinate cepstral analysis/editing based on models.
Unless you have a particular use-case the NFS isn't really needed IMO.
Editing like this is all about *reducing (not truly eliminating) ripple caused by the room that contaminates the result - and we have that with various limitations already. At the DIY-level it's all about knowing those limitations and working with them to achieve similar results (and perhaps arguably better results at higher freq.s).
*notably as the mic moves closer to the source ripple (typically diffraction) caused by the loudspeaker itself becomes more prominent (and varying based on position). I sometimes look at certain loudspeakers measured properly with the NFS and think to myself :that doesn't look quite right between 500 and 1.2k: and I wonder if it is more correct or less in the context/comparison of a good/anechoic far-field measurement.
It looks like coordinate cepstral analysis/editing based on models.
Unless you have a particular use-case the NFS isn't really needed IMO.
Editing like this is all about *reducing (not truly eliminating) ripple caused by the room that contaminates the result - and we have that with various limitations already. At the DIY-level it's all about knowing those limitations and working with them to achieve similar results (and perhaps arguably better results at higher freq.s).
*notably as the mic moves closer to the source ripple (typically diffraction) caused by the loudspeaker itself becomes more prominent (and varying based on position). I sometimes look at certain loudspeakers measured properly with the NFS and think to myself :that doesn't look quite right between 500 and 1.2k: and I wonder if it is more correct or less in the context/comparison of a good/anechoic far-field measurement.
Last edited:
Most anechoic chambers don't go that low in freq. with a high absorption coefficient.
So to wrap up: apart from the accuracy and convenience argument, in room the Klippel is substantially more accurate in the indeed tricky 200-1000 Hz range, furthermore taking diffraction into consideration. Those are valid points, but for me the nagging question is whether there ways to achieve similar results in e.g. ARTA, given the outrageous pricing. I recall mbrennwa also struggled somewhat in the Monkey Box design in the 300-1000Hz range.
And yes, I am well aware of the shortcomings and traps of diy measurements: to my eye also much of what I see is way too good to be true. No dispute there about the Klippel advantages: it certainly looks foolproof. Yet the price is prohibitive.
For me the main -yet unanswered- question remains: is it a software issue, or do we (only? need much bigger spaces for accurate 200-1000 Hz measurements (and much, much more time) with the usual measurement suspects?
And yes, I am well aware of the shortcomings and traps of diy measurements: to my eye also much of what I see is way too good to be true. No dispute there about the Klippel advantages: it certainly looks foolproof. Yet the price is prohibitive.
For me the main -yet unanswered- question remains: is it a software issue, or do we (only? need much bigger spaces for accurate 200-1000 Hz measurements (and much, much more time) with the usual measurement suspects?
I am very surprised but the world would be a lot less interesting if we all valued things the same way. You are aware that for that sort of money you might be able to get your measurements from a properly accredited institution if you fitted in with their schedule. They probably wouldn't book, build up, measure, take down for that price but a few might agree to piggy back your job on a similar one. Whether they knew you or not is likely to be relevant though so perhaps I should keep quiet.