Klippel Near Field Scanner on a Shoestring

I don't know how good of a velocity microphone it makes, but some years back I was able to remove the circuit board from the back of a small electret capsule and run a tiny wire from its back contact to its fet follower mounted below. Had to put a metal screen around it all to prevent hum pickup, also went through 3 capsules till I managed to get one going without breaking it. But they're absurdly cheap so it was mostly just time and aggravation.

But I was able to put this into a horn, with diaphragm facing perpendicular to the axis and get a pretty strong null with careful positioning. I was seeing if I could see effects of HOMs, and did get some interesting ripply response peaks. I may have the butchered mic downstairs somewhere still. I wondered whether there might be some clever way to calibrate it, but never looked very far into it.
 
...velocity with two mics is done all the time in intensity probes. It is a very common technique...

I know it is commonly used in intensity probes but it seems to be sub optimal for this application, as shown in the link I provided.
Even intensity probes show problems at not-very-low frequencies.
Prof A. Farina tests a Brüel & Kjær unit >here< and it's textbook at 1kHz but close to useless by 125 Hz.
If even the very expensive B&K has problems with phase and amplitude match at 125 Hz then Aaron's affordable microphone will need care.
I think this supports his surmise that Klippel move one mic rather than use two so as to minimize mismatch.
It reinforces my desire to find the analysis that is most robust to real world errors.

... but some years back I was able to remove the circuit board from the back of a small electret capsule and run a tiny wire from its back contact to its fet follower mounted below...

That sounds like a fun project, there was a discussion of velocity mic calibration on Micbuilders forum if you want to follow it up.
Velocity mics are not easy, that's why Microflown practically has a monopoly despite the >$10k price.

Best wishes
David
 
Dave - I am well aware of the limitations of two mics for sound velocity, but solutions are as simple as just spacing them wider apart - if we need velocity at LFs. I am not sure that we do. Maybe two sweeps for low and high frequencies. In my system I just fit the LFs to a radiation model, but this is problematic in the general case. As I said however, I would not start doing this problem in the general case. Start simple, work up resolving problems as you go.

PS. The Zernike functions are an orthogonal set in a circular aperture, but that is not sufficient. The functions in the aperture must also have an orthogonal set for the radiated sound. It may be possible to get these from the Zernike functions (but maybe not), however, they are already know for the Bessel ones.
 
...I don't think the Zernike functions are readily available. I don't have code for them. I had not even heard of them until I looked them up. Even in the classic text "Fourier Optics"...

I only learned of them from Aarts and Janssen's 2009 paper where they extend D.B. Keele's work on near field measurement.
Aarts was in the optics team on the development of the CD before he moved to acoustics and Zernike functions seem to be common for optics analysis, probably made them a natural choice for him.
The paper measures at different points on the speaker axis rather than on a surface. So they only do the radially symmetric modes, I am not sure if this addresses your concerns.

...I am well aware of the limitations of two mics for sound velocity, but solutions are as simple as just spacing them wider apart

But when they are wider apart then the linear approximation is less accurate, that's more or less my point, and shown in the reference I linked.
The B&K intensity probe does have different spacers that can be swapped, as I expect you know.
It would be easy to make spacers practically any size but the maximum available is only 50 mm, so I assume they felt that 50 mm was the limit for acceptable accuracy.
B&K are fairly smart so I take their number as a reasonable hint.

... if we need velocity at LFs. I am not sure that we do.

We can already do nice HF measurements with a time window so isn't LF exactly where we need to optimize?

Maybe two sweeps for low and high frequencies.

This ties back to the Klippel patent that sparked my concerns in the first place because it tries to optimize the scan and avoid multiple sweeps.
You have been somewhat dismissive of that patent but I take it seriously.
Partly because of the circumstances around it - there was no established player in the market with patents that needed to be circumvented.
So presumably they felt they had solved a real problem - to justify the cost, time and paperwork.
And that is supported by an actual examination of the patent.
So I would like to understand this better to see what problems they encountered and learn from their work.

Start simple, work up

Absolutely, that was my advice to the OP too;)
I just want to look ahead a bit, to be sure that the simpler version is extensible.
And usually the effort to understand the universal case provides a better perspective on the special cases.

Best wishes
David
 
Patents are seldom written to elucidate some topic...

Sure, plenty that are a waste of time - and life is short.
That's why I provided my rationale why I think this one is of interest.
But the real test is simply whether the actual content is useful.
I realize you contribute to quite a few threads, so have you had the time to read the actual patent?
If so then any comments on the content?

...I prefer to learn from academic papers

I hope there was sufficient academic cred. in my earlier link to a paper by Dr. Manuel Melon, Fellow of the AES, full professor at a reputable university, etc. etc.;)
Also on the academic side - I just read Putland's doctoral thesis and learned some useful stuff.

Best wishes
David
 
I realize you contribute to quite a few threads, so have you had the time to read the actual patent?

Best wishes
David

No, I have not read the patent. But its not because I post here, its because I no longer study acoustics as a result of my other passions (glass art and quantum field theory.) They don't leave much time for anything else. I've done my stint in acoustics, time to move on. I'll advise and answer questions, but I won't go and study anything new.
 
I am certainly interested in making a mockup- proof of concept but I have to admit the math is over my head the way it is presented in the documentation. I am skilled at matlab and at acoustical measurements but I haven't seen "the light" yet concerning the math involved here. Actually I am following this thread with great interest, hoping that someone will explain the math concept in such a way that it clicks in my head.

Kind regards,
Kees
 
...but I have to admit the math is over my head the way it is presented in the documentation. I am skilled at matlab and at acoustical measurements...

Hi Kees
I emailed the start of an explanation to Aaron and planned to expand it and post it here.
That way any mistakes can be corrected and I can be sure it is clear.
I just wanted to be sure I understood it properly myself so I have hesitated while I wait for a copy of "Fourier Acoustics".
But your interest will prompt me to finish it, first part should be in a day or two.

Best wishes
David

Or if you don't want to wait then just email the address in my profile.
 
Last edited:
Thanks, nice to see someone else is interested.

Best wishes
David

I am not totally uninterested, I am just not interested in what Klippel is doing (just as he seems uninterested in what I am doing.) I already have a system that meets most of my needs and Klippel's is just too elaborate and expensive to be appealing to me. His system is for professionals, not DIY. DIY needs what I do, not what Klippel does.

The one thing that I want to add to my system is a calculation of the vertical lobbing that results from non-coincident drivers. This should be possible from a simple knowledge of the spacing without taking any vertical measurements or at most one or two. That interests me.
 
I am not totally uninterested...

My comment wasn't a swipe at you at all.
On the contrary, I appreciate any comments and if people take the time to read any links or references then that's a bonus, it's a free, voluntary forum after all.
I meant "someone else" apart from you, there hasn't been a lot of other responses to Aaron's question.

Best wishes
David
 
I can also envision another application; being able to seperate the direct soundfield from the reflection would enable one to "zoom in" to reflections in the application of room acoustics, which is partly my job (I design recording studios and concert spaces for a living, doing loudspeakers as a hobby).
It would, even from an academic point of view, be very interesting to "look" at low frequency fields in smallisch rooms.

Best regards,
Kees
 
I am sure there are lots of other areas to look at, but what I am most interested in is a simple way of getting good polar data that can be put into a data-base or form by which we can all access the data for comparison purposes. I have tried to do this with my polarmap program, but not being available for others to use to actually take and create the data has been a big hindrance. To me, the purpose of this thread was to do that, to create a public domain analysis program that can take and create a polar map of all field points and frequency points to be viewed at will. This goal does not seem to be moving very far. Complex scanning systems would be dead on arrival for DIY.
 
But when they are wider apart then the linear approximation is less accurate, that's more or less my point, and shown in the reference I linked.
The B&K intensity probe does have different spacers that can be swapped, as I expect you know.
It would be easy to make spacers practically any size but the maximum available is only 50 mm, so I assume they felt that 50 mm was the limit for acceptable accuracy.
B&K are fairly smart so I take their number as a reasonable hint.
It makes me wonder what the design intent of their probe is. If it’s focused on higher frequency, maybe that’s why they only have spacers up to 50 mm. But then again, from that paper you linked, it looks like performance goes downhill below 1/8 wavelength spacing, that could mean the scan layers would need to be close to a half meter apart, and that’s a bit inconvenient.

It feels like it might be useful to make up some kind of test that we can perform to give us an idea of the best way forward.

I am certainly interested in making a mockup- proof of concept but I have to admit the math is over my head the way it is presented in the documentation. I am skilled at matlab and at acoustical measurements but I haven't seen "the light" yet concerning the math involved here. Actually I am following this thread with great interest, hoping that someone will explain the math concept in such a way that it clicks in my head.
I’m glad to hear that! And it makes me feel better to know that I’m not the only one who needs help with the math.

Hopefully you’ll be able to lend some of your skills to this project. :)

His system is for professionals, not DIY. DIY needs what I do, not what Klippel does.
DIY does indeed need what you do. And I would like to have that, but in addition, I like to see as much of the ‘picture’ as possible. Hence my interest in being able to do something likes Klippel.

Although being able to make a full range measurement without an anechoic chamber is something anyone building speakers can use.

The one thing that I want to add to my system is a calculation of the vertical lobbing that results from non-coincident drivers. This should be possible from a simple knowledge of the spacing without taking any vertical measurements or at most one or two. That interests me.
Would that be similar to calculating an interference pattern?

…there hasn't been a lot of other responses to Aaron's question.
I’d like to say that I have been very grateful for all the answers that I have received from the two of you. It’s pushing this idea closer to reality for me.

The fact is that few understand the need for considering directivity. Fewer understand how it is measured, and even fewer care about doing it as efficiently as possible. This small group appears to be the three of us.
I think that is very gradually changing, though. Twenty-five years ago when I was starting to get interested in audio, acoustic measurements were rare in publications and in the DIY community, and when they were present, they were a single on axis curve. Usually with 1/3 octave resolution. But now it’s not uncommon to see off axis measurements, and although they are indeed still very low in angular resolution, it is progress.

And even if it’s just the three of us for now, having the tools available can encourage more people to become aware of the need.

I can also envision another application; being able to seperate the direct soundfield from the reflection would enable one to "zoom in" to reflections in the application of room acoustics, which is partly my job (I design recording studios and concert spaces for a living, doing loudspeakers as a hobby).
It would, even from an academic point of view, be very interesting to "look" at low frequency fields in smallisch rooms.
It’s interesting that you bring that up because it’s an application that has crossed my mind too. After all, knowing the timing, direction, and intensity of a sound allows for a much deeper analytical (and useful) analysis of what the room sounds like and what can be improved.

To me, the purpose of this thread was to do that, to create a public domain analysis program that can take and create a polar map of all field points and frequency points to be viewed at will.
That is roughly the purpose that I had in mind in starting it, and in addition, to provide a tremendously useful tool for the helpful individuals who have taken to measuring speakers, subwoofers, and raw drivers for the dedicated enthusiast or DIY’er.
 
It makes me wonder...

I think it is possible to explain the maths in physical terms.
The B&K probe is an all-purpose intensity probe.
There are inevitable physical constraints, if they increase the spacers for low frequency sensitivity then they lose hi frequency accuracy, and conversely.
We have extra information, we know the source is inside the scan surface and the echoes are outside so we can use the fact that the solution should be in the form of a Bessel (or, equivalently Hankel) function.
So we can fit the data to a Bessel function rather than use a linear approximation.
This is more accurate, as the reference shows, we have a better trade-off between hi frequency accuracy and low frequency sensitivity so some of the limits of the B&K probe are not relevant.
So we can exceed the 50 mm limit of the linear approximation but 500 mm is probably excessive.
This is essentially what the Klipppel patent is all about.
It is complicated by reflections off the speaker itself, which break our assumption.
It is not clear to me yet how much of a problem this is, Earl says he never noticed this but it may be that it simply wouldn't be evident with his method.

We also have another technique that is unavailable to the B&K probe.
We can time window and handle the anechoic hi frequency information separately.
This results in an optimization that tends in the same direction as the previous point, towards increased distance.

Best wishes
David
 
Last edited:
The fact is that few understand...it as efficiently as possible...

Efficiency is where I have a conceptual problem at the moment.
To fit a uniformly spaced set of data is not much more difficult that a Fourier transform.
It is not clear to me how to optimize measurement points when the practical interest is much more in the forward direction.
And once we have optimized but uneven spaced data I am not sure exactly how to fit it.
IIRC Earl Williams actually wrote that Fourier Acoustics would be fairly simple if it wasn't for real world issues like this, and truncation effects, numerical stability and so on.

Best wishes
David
 
Last edited:
I can also envision another application..to seperate the direct soundfield from the reflection would enable one to "zoom in" to reflections in the application of room acoustics...

This may not be so easy, one could say that as we "zoom in" then any errors in the data are expanded too.
Or, as mathematicians would say, "the inverse problem is ill conditioned".
I don't know if this applies to exactly your idea but my intuition is that there will be serious problems with data sensitivity to noise, measurement inaccuracy etc.

Best wishes
David