There are no explanations given, in fact most of them are being ignored and/or unanswered.
As for the second part; no.
It's not hard to ask further what somebody's meaning is, especially when there are already so many thick layers of misreading/miscommunication because of text only, difference in perception, difference in language and difference in culture.
Which is not only frustrating when people just assume certain things, it's also extremely distracting from the original conversation.
There are many many many things you or other people say or do, that is very disrespectful to other people.
Just please think about that for a second next time, because you don't seem to be always aware of this.
Which I am totally okay with, that is just the nature of communicating.
I say that with no offense at all. 👍
To go back ontopic instead of going further with this non-sense.
Literally the only thing I am doing is investigate if there aren't smarter methods.
It's extremely obvious that throughout this topic that there has been only very little investigation on this (if anything at all).
This is done from an OBJECTIVE point of view, NOT my subjective view of what is better or not.
I think you just really hit the most important point here; having a fixed measuring setup!
I think this is the biggest benefit of a Klippel system.
Not because how unique the system is, but because it's a fixed setup with fixed dimensions.
To explain this on a practical way; if we would just use the hardware and only measure gated measurements, this will already result in a very great system.
But again, that is is basically going for plain convenience!
Which is 100% fine, I never said that there is anything wrong with convenience.
I was just talking about expectations.
One very big improvement that a Klippel system is lacking, is the fact that the microphone moves and not the speaker itself.
I have done both, and in my experience a moving speaker (and therefor a very fixed setup) works much more reliable.
I personally think that the whole Klippel trick can be done this was as well, or at least to a great extend.
Which makes it a lot less convoluted
I explained this in a previous post, the time you win with automatic steppers and measurements vs just doing something by hand is extremely minimal in sense of work and time.
With a fixed setup, I can measure an entire speaker within 30 min.
So I find the amount of work/time a not so very strong argument to be perfectly honest.
For convenience, aka I don't have to move my butt to turn a speaker, that's a different story.
The strongest point is;
- getting more resolution at the lower frequencies (below say roughly 500Hz)
- being able to measure from about 50Hz (needed for multi-way systems)
I still have a lot to learn, but I think the reflections of the speaker in the room have to be consistent in order to calculate them out. By rotating the speaker, the reflections by the room change.
That's true, but the question is not if they change, but at what frequency they stay consistent enough.
Because below a certain frequency there isn't any/very little change in reflections anymore.
In addition, because these reflections change, you can use that as well to average the results.
I feel that is a bit of a chicken-egg story.
If you rotate the speaker together with the microphone and take measurements, reflection might very well average out. That is quite similar to keeping the rotation/orientation constant, but change the distance.
The direct signal has to stay the same. You can't change the angle between microphone and speakers. That dependency is what you want to measure.
Rotate microphone as well as the speaker?
Not sure what you mean by this.
For just measuring the angle; you can either move the mic or the speaker, that's 100% the same thing.
Moving both doesn't make sense since you will always stay on-axis?
But to make things already a step smarter IF we also want to move the microphone.
You could maybe move (turn) both in the opposite direction.
That way you only have only ever move 90 degrees.
This already make the entire mechanical assembly a lot easier
Yup. The Weinreich / Klippel method requires the sound fields of the speaker and the echoes to remain the same with all measurement steps. In other words, the speaker and the reflectors (walls, furniture, etc.) must not move relative to each other. The speaker must not rotate.
@b_force It's not about averaging things. It's about disentangling the direct sound of the speaker from the echoes. Just read the Weinreich paper I posted a few days ago. It's also not necessarily about saving time, what you can do manually, or being sensible. The boys (and gals?) seem to be interested in avoiding boring routine work, making things more precise, improving resolution, DIYing cool stuff that would cost 100k as a commercial offering, and have fun. Just let them play
The near field to far field conversion can be done with a moving speaker. The Sound field separation cannot be. For me the real trick is sound field separation. I ran every option I could think of to shortcut it past the people I know who understand the math. I really hoped that was not true but I have now accepted that it is.
One benefit of an automated turntable is something Earl touched on earlier. Angular resolution. The code Tom wrote to control a stepper motor through ARTA at my request is capable of accurately positioning at less than 1 degree of rotation. Depending on the gear setup there can easily be 9 steps on the stepper motor per degree of movement.
There was a paper I read and can longer find that tested how much angular resolution was needed to avoid inaccuracies from interpolation in balloon data. The conclusion was measurements at 2.5 degree steps or less were needed.
My test platform takes less than 3 minutes to make a 180 degree measurement set at 10 degree steps with ARTA using 3 short sweeps similar to the NFS.
3 minutes vs 30 is probably irrelevant for a one off measurement session but the accuracy and repeatability of the setup is nice.
I am aware of that, but that is such a mouthful!
I already said that before as well!It's also not necessarily about saving time, what you can do manually, or being sensible. The boys (and gals?) seem to be interested in avoiding boring routine work, making things more precise, improving resolution, DIYing cool stuff that would cost 100k as a commercial offering, and have fun. Just let them play
Nothing wrong with that either, but it's important to set the right expectations right from the beginning!
I was just reading this above, and I am a bit confused about angular resolution in what context?
(which is probably a misread from my end).
- For directivity purposes?
In that case such extreme resolutions are not needed to estimate a proper well behaved directivity.
- For the disentangling trick?
In that case I am missing how much is needed as function of frequency?
The angular steps of the measurements taken for creating a polar map e.g, 5,10,15,20
To avoid introducing errors from interpolation high angular resolution is needed. It then becomes a question of how much error is acceptable. See Earl's quote.
Not specifically, as it is not clear to me yet how much angular resolution is needed for any given frequency for sound field separation.
I have read that before, but I am sorry but to get a good indication of directivity we don't need uber micro precision in angles, because there are very little reasons to believe this directivity (and therefor polar plots) will all of a sudden totally change per (very) small increments.
Or another way of saying this, is that super resolution is only necessary when there is a high reason to believe that directivity will change drastically at these small increments.
If that is acoustically even possible (which I find extremely unlikely, since that's not how things naturally and physically work), the next question is if the human hearing is even perceptible for such teeny tiny changes.
Or like I said before, I am misreading things?
I have been responding to you in irritation, and I apologies, and I appreciate you taking it gracefully. Language can indeed be a barrier to communication, even when two people are speaking the same one, and then added problems come when intent appears ambiguous.There are no explanations given, in fact most of them are being ignored and/or unanswered.
As for the second part; no.
It's not hard to ask further what somebody's meaning is, especially when there are already so many thick layers of misreading/miscommunication because of text only, difference in perception, difference in language and difference in culture.
Which is not only frustrating when people just assume certain things, it's also extremely distracting from the original conversation.
There are many many many things you or other people say or do, that is very disrespectful to other people.
Just please think about that for a second next time, because you don't seem to be always aware of this.
Which I am totally okay with, that is just the nature of communicating.
I say that with no offense at all. 👍
You define objectively working smarter as follows:
There are two concepts being discussed in this thread: modal analysis and sound field separation. Applied to my speaker building and design process, using modal analysis would drastically reduce the number of measurements I would need to make and there would be a significantly improved data set compared to what those measurements would get me with "traditional" methods. Using sound field separation, I would not have to merge two data sets, all my major acoustic data would be captured at the same time, and owing to the nature of how the data would have to be gathered, the measurement system would be at least partially automated and modal analysis would be inherently included.
That sounds like objectively working smarter to me.
Are those two concepts the only way to work smarter? Perhaps not, but for me, those two concepts seem like they are worth pursuing. And you are of course free to bring up other novel ways of working smarter, I'm certainly not opposed to investigating and hearing about novel methods, just look at the discussion on the beamforming method earlier in this thread.
Earlier you asked why the devotion to these methods in this thread, and I think I misunderstood the intent behind it. When it comes to modal analysis, I don't think there is another way other than not using math and taking a measurement ever 2 degrees. And as for sound field separation... I don't know is there is another way, but it seems like enclosed measurement surfaces is the only way to go. After all, to disentangle the sound source from reflections would need some kind of map of the sound field. Weinreich and Klippel came to that conclusion as did several other researchers, and sound intensity measurements use it too. Are there other ways to get anechoic data in an echoic space? There may be other ways (a sound field microphone, or some kind of beamforming might work too), but I find this one interesting and am focusing on it.
At the stepper motor shaft I agree with the angular accuracy. But at the end points of a measuring system I think that our plus or minus 3mm will be the norm. Having made many jigs and fixtures for a great many years I can tell you that rigidity is the main problem. Even the aluminum extrusions that Klippel uses are flexible at the connecting points. So I have a few ideas to maintain rigidity that are fairly simple. It's called tension and compression. Utilization of thin inexpensive steel cable. You can keep a relatively light swing and and boom rather rigid when it is in tension. Think cable suspension of a radio tower. A radio tower will not hold itself vertically, but with the aid of the cables it is just fine. Compression is the other method. You can draw tight a long wood arm by means of looping a thin cable through it and tightening via a threaded bolt embedded at the correct point. Fairly easy to do. I think everyone on this thread is reasonably handy.
If you guys want, I can make short videos to get the creative ideas flowing.
I agree. But we need to start somewhere. And everything that I am considering will be easy to turn into Numeric control.
When a polar response goes through a radiation null, and these happen a lot with angle, then the rate of change can be extremely high and without modal analysis these get missed unless very detailed data is taken. That's the whole point. It does make a difference.
It is true here that there are two separate tasks being looked at here. Modal analysis and sound field separation (SFS). The second requires a rotating mic system, the first does not. So unless one is on board to assemble the full 3D hardware then finding a way to measure LF response with a fixed mic is at odds with SFS. I am of the camp that a fixed mic and rotating the speaker is far more practical for DIY and rotating mic only makes sense for the pro designer. Hence it may be that the two approaches are not compatible and maybe should be separate discussion threads. AT least we need to be clear on which approach we are taking.
For example Bill Waslo did something several years ago that is intriguing. He used statistical forecasting to predict what the impulse response would be if there weren't reflections, based on data that we have when there are not. In theory this can be done with a fixed mic. Why isn't that being considered? There are many ways to do this, but I have only given it a small bit of effort and basically got nowhere.
For example Bill Waslo did something several years ago that is intriguing. He used statistical forecasting to predict what the impulse response would be if there weren't reflections, based on data that we have when there are not. In theory this can be done with a fixed mic. Why isn't that being considered? There are many ways to do this, but I have only given it a small bit of effort and basically got nowhere.
So in this case you're talking about cases when there is a very deep nul in the polar response?
That's a pretty specific case with only a dipole or cardioid.
It's only the area (angles) around this deep nul that need so much precision.
I guess one other situation would be a line array.
But that's a totally animal all together.
Hence the reason we do multiple measurements in a radiating pattern around the DUT, and at differing heights. from the floor and ceiling. Only stating the obvious extrapolation of the reasoning. The multiple points if taken with relative consistency will allow a subtraction of the rooms effects.
The concept is a reasonable one when you think through the physics. Or better said the acoustics.
Earl is there any chance that you took a pic of your test setup?
I have done a lot of polars the old fashioned way with a mic boom and reasonable measurement. I use a gun bore LASER sighting tool to get a good idea where I am aiming the microphone.
Right now I use hi-tech tape.