what's the diameter of your measurement microphone? and how far is it from the acoustic center? I'm sure there's a very acute cone described by those constraints.
or better yet, what's the distance between your ears? that may be your cone's diameter.
here's another possible cone you can use 😀
http://valiante.plugnpay.com/scstore/graphics/DunceCapLarge.jpg
or better yet, what's the distance between your ears? that may be your cone's diameter.
here's another possible cone you can use 😀
http://valiante.plugnpay.com/scstore/graphics/DunceCapLarge.jpg
Noah,
The "0 degree angle" you are talking about would only exist IF the microphone diaphragm had a radius of zero AND the loudspeaker diaphragm had a radius of zero.
Zero degrees is an axis, a direction; not a volumeless cone.
The truncated cone you're really measuring with your microphone (assuming it's time-gated to eliminate reflections) is outlined by the edges of the microphone diaphragm and the edges of the loudspeaker diaphragm. Neither have a radius of zero, so the "0 degree angle" in this case doesn't begin to exist except in the world of semantics.
And I don't think Earl is inclined to spend his time arguing semantics. He's not an English professor. If we want his participation here, let's not put words in his mouth with statements like "I was just incredulous that Dr. Geddes would keep asserting that n x zero does not equal zero". He never said that. What you are doing is called a "straw man argument" - you are falsely attributing to your opponent something easy to argue against that he never actually said.
What I am doing is calling you on it. I have nothing personal against you, but I do object to your behavior here.
Duke
The "0 degree angle" you are talking about would only exist IF the microphone diaphragm had a radius of zero AND the loudspeaker diaphragm had a radius of zero.
Zero degrees is an axis, a direction; not a volumeless cone.
The truncated cone you're really measuring with your microphone (assuming it's time-gated to eliminate reflections) is outlined by the edges of the microphone diaphragm and the edges of the loudspeaker diaphragm. Neither have a radius of zero, so the "0 degree angle" in this case doesn't begin to exist except in the world of semantics.
And I don't think Earl is inclined to spend his time arguing semantics. He's not an English professor. If we want his participation here, let's not put words in his mouth with statements like "I was just incredulous that Dr. Geddes would keep asserting that n x zero does not equal zero". He never said that. What you are doing is called a "straw man argument" - you are falsely attributing to your opponent something easy to argue against that he never actually said.
What I am doing is calling you on it. I have nothing personal against you, but I do object to your behavior here.
Duke
Filter based on power response (brown) and minphase approximation seems to work very well. On axis and with one strong reflection the pressure curve is flat from 1 to 10kHz within +-1dB. Same with 30 deg (blue curve). Now the problem rises with implementation. Passive version RC-LR, brute approximation without step roll of is probably no-go.
Attachments
Neither the mic nor driver diameter matters
Noah, what's important is to consider the math. The power response is the integration of all SPL vectors intersecting a sphere that encloses the driver. All of these vectors have an equal contribution to the power response. None are zero, including the single on-axis vector for a true point source.
My first thought was as y8s and audiokinesis mentioned, the mic surface area and driver origin define a cone, but in reality this isn't the case, either. This would be so for a true point source only, but drivers are not point sources, so the SPL of the on-axis response as measured by a microphone would be an integration of vectors from the entire surface area of the driver as they imping at all points of the mic surface area.
In any case, even considering for sake of argument that the driver is a true point source emitter and the mic a point source receptor, the key is that every vector has an equal contribution. But as Dr. Geddes pointed out, the off-axis, as he described it, has a significantly larger contribution because of the significantly large number of vectors that would make up a circle (not a cone) at the intersection with the aforementioned sphere for any given off-axis angle.
Dave
Noah, what's important is to consider the math. The power response is the integration of all SPL vectors intersecting a sphere that encloses the driver. All of these vectors have an equal contribution to the power response. None are zero, including the single on-axis vector for a true point source.
My first thought was as y8s and audiokinesis mentioned, the mic surface area and driver origin define a cone, but in reality this isn't the case, either. This would be so for a true point source only, but drivers are not point sources, so the SPL of the on-axis response as measured by a microphone would be an integration of vectors from the entire surface area of the driver as they imping at all points of the mic surface area.
In any case, even considering for sake of argument that the driver is a true point source emitter and the mic a point source receptor, the key is that every vector has an equal contribution. But as Dr. Geddes pointed out, the off-axis, as he described it, has a significantly larger contribution because of the significantly large number of vectors that would make up a circle (not a cone) at the intersection with the aforementioned sphere for any given off-axis angle.
Dave
Re: Neither the mic nor driver diameter matters
Dave
I mis-characterized Dr. Geddes description, he stated it is an annulus, not a circle, so it looks like he's using an averaging of measurements on the annulus with the center on the specific off-axis angle rather than just measuring on a circle at the specified off-axis angle. I like that methodology, but it's got to be extremely time-consuming. In either case, the on-axis contribution to power response is small, but not zero nevertheless.dlr said:
Dave
Thanks all for surpressing a rather rediculous argument that I had become uninterested in.
When we make a pressure measurement we are sampling the sound field just as we sample a voltage signal digitally. This has many impications, not the least of which is spatial aliasing in the data, but thats not the topic here. We have to consider what this sample measurment is intended to represent. If the source is omnidirectional then there is little to no error in assuming that the single axial point represents the entire spherical radiation, i.e. much greater than the diaphagm size. But when the source has directionality then one must carefully consider what this spatial pressure sample represents.
I used to assume axisymetry when calculating the power response, but I found this to be erroneous because virtually no source is truely axisymetric. Then I started to use horiontal and vertical planes which is an improvment, but since there are typically "holes" in the vertical plane they are inordinately represented in the power response calculation, again some errors. Now I interpolate the vertical response to the horizontal response and use those numbers for the intermediate values.
Its not too hard to get horizontal polars, a bit more difficult to get vertical ones, but a real pain to get any other angle.
The power response is extremely important to the subjective assesment of a loudspeaker and it needs to be measured and displayed. So these are not ACADEMIC questions and they need to be taken seriuosly.
When we make a pressure measurement we are sampling the sound field just as we sample a voltage signal digitally. This has many impications, not the least of which is spatial aliasing in the data, but thats not the topic here. We have to consider what this sample measurment is intended to represent. If the source is omnidirectional then there is little to no error in assuming that the single axial point represents the entire spherical radiation, i.e. much greater than the diaphagm size. But when the source has directionality then one must carefully consider what this spatial pressure sample represents.
I used to assume axisymetry when calculating the power response, but I found this to be erroneous because virtually no source is truely axisymetric. Then I started to use horiontal and vertical planes which is an improvment, but since there are typically "holes" in the vertical plane they are inordinately represented in the power response calculation, again some errors. Now I interpolate the vertical response to the horizontal response and use those numbers for the intermediate values.
Its not too hard to get horizontal polars, a bit more difficult to get vertical ones, but a real pain to get any other angle.
The power response is extremely important to the subjective assesment of a loudspeaker and it needs to be measured and displayed. So these are not ACADEMIC questions and they need to be taken seriuosly.
jzagaja said:
I really don't understand what you are doing, but the data seems questionable to me. How could you have "one strong reflection" and still have a flat frequency response? This is not possible. There should be peaks and nulls at least 10-20 dB deep. A reflection that is down by 6 db will still cause dips in the response of some 20 dB.
Dear Dr Geddes,
Reflection is 19dB down at 47cm. Peak to peak it is 10dB (grey curve) but if you run psychoacoustic filter (black curve, Farina et al) then reflection with eq filter makes +-1.5dB. With freq. windowing it's even less.
So you are saying my correction filter is completely senseless?
Reflection is 19dB down at 47cm. Peak to peak it is 10dB (grey curve) but if you run psychoacoustic filter (black curve, Farina et al) then reflection with eq filter makes +-1.5dB. With freq. windowing it's even less.
So you are saying my correction filter is completely senseless?
Attachments
I'm saying that I don't follow what you are saying so I can't comment.
If you are showing that you can smooth out a single measurement to be flat - OK. Thats not a big deal.
I keep saying that I would need to see the polars of a completed system to make any comments on anything. Your stand alone curves with labels that are unintelligable to me are not of much interest.
If you are showing that you can smooth out a single measurement to be flat - OK. Thats not a big deal.
I keep saying that I would need to see the polars of a completed system to make any comments on anything. Your stand alone curves with labels that are unintelligable to me are not of much interest.
"...let's not put words in his mouth with statements like "I was just incredulous that Dr. Geddes would keep asserting that n x zero does not equal zero..."
My mistake for making that interpretation, for which I apologize.
I was never talking about microphones or measurements, just the geometry of a cone.
Sorry for the OT.
My mistake for making that interpretation, for which I apologize.
I was never talking about microphones or measurements, just the geometry of a cone.
Sorry for the OT.
gedlee said:
I like this idea - a lot. Has it been done? Or how could it be done?
Of course we have to remember that if the zero axis is pointed straight at us, it will have a strong influence on the sound. Or any other given axis, for that matter. But that the way we usually think about it, so I'd love to so a more thorough representation of the "Power Field" or whatever it might be called.
panomaniac said:
I have done some things on my own, but I was not quite satisfied with the results. The "zero axis" has some real problems with it because almost never is one actually on this axis, and the way that we measure speakers this axis is highly over-weighted (and over-rated). In my own designs I almost universally ignore it.
Things can happen along this "unique" axis that are highly localized - to-wit my ESP designs. There is always a hole on axis from the waveguide, but this hole is limited to axial points only. In terms of its effect on the power response or the response along a real listeners axis these holes are insignificant. But if I were to EQ the axial response to flat (as is common), this would have a very pronounced negative effect on the power response and the response along the real listeners axis.
Things in the real world of loudspeaker design are not quite as simple as we would like them to be.
gedlee said:
What? Damn! Back to square one. Or should I say back to "Zero"? 😉
Getting a good power graph must be hard. But I look forward to seeing work in this area.
gedlee said:
I am complete sympathy with Dr. Geddes on this point. With a high-resolution measurement system (not a RTA analyzer, thank you very much) you quickly discover that radiation near the +/-5 degree axis isn't very good - and heroic measures to "smooth it out" have a heavy penalty everywhere else. The narrowness of this "pencil beam" depends very much on the driver properties.
Noah, it takes a big man to eat his slice of humble pie without choking on it. When my turn comes (as I'm sure it will), I hope to be as graceful as you were. You have my respect.
Back to audio: The waveguide I use has an on-axis dip centered on 11 kHz that completely disappears at 5 degrees off axis (and maybe sooner than that). Of much greater interest are a couple of off-axis energy ridges that are not apparent in the on-axis curve. These are audible, and these I equalize.
In fact, in a family of curves from 0 degrees out to 90 degrees taken at 7.5 degree intervals, the on-axis curve is the one that stands out as the most dissimilar from all the others. So the on-axis curve would be the worst one to base your equalization on (even assuming all curves are created equal, which from a power response standpoint they are not).
This issue is especially of interest to anyone using DSP on a speaker with an axi-symmetric horn or waveguide or coaxial driver - the implication is that you should place your measurement microphone off-axis a bit if you're signal-processing based on high resolution gated measurements.
Duke
Back to audio: The waveguide I use has an on-axis dip centered on 11 kHz that completely disappears at 5 degrees off axis (and maybe sooner than that). Of much greater interest are a couple of off-axis energy ridges that are not apparent in the on-axis curve. These are audible, and these I equalize.
In fact, in a family of curves from 0 degrees out to 90 degrees taken at 7.5 degree intervals, the on-axis curve is the one that stands out as the most dissimilar from all the others. So the on-axis curve would be the worst one to base your equalization on (even assuming all curves are created equal, which from a power response standpoint they are not).
This issue is especially of interest to anyone using DSP on a speaker with an axi-symmetric horn or waveguide or coaxial driver - the implication is that you should place your measurement microphone off-axis a bit if you're signal-processing based on high resolution gated measurements.
Duke
SO - given all these "of course" and "I agree", how does one EQ or do crossover designs from sources made from pistons - which don't have constant directivity? The piston source does not exhibit constant power, but does have constant axial pressure (but axial pressure only!!). These two features are not compatible with the previous discussion.
It was exactly this point that drove me to horns (then waveguides, then foam) so many years ago. There is no solution for piston source loudspeakers.
It was exactly this point that drove me to horns (then waveguides, then foam) so many years ago. There is no solution for piston source loudspeakers.
Earl,
I probably missed more than I picked up, but do remember how thorough you were when it came to getting accurate and comprehensive data. I figure there musta been a reason why you went to so much trouble.
Duke
p.s. - I did some Christmas shopping on your website, but no hurry on the shipping.
I probably missed more than I picked up, but do remember how thorough you were when it came to getting accurate and comprehensive data. I figure there musta been a reason why you went to so much trouble.
Duke
p.s. - I did some Christmas shopping on your website, but no hurry on the shipping.
"Of much greater interest are a couple of off-axis energy ridges that are not apparent in the on-axis curve. These are audible, and these I equalize"
Duke - this was the line I picked up on. Most people don't even look at the off-axis data let alone look for "energy ridges" to EQ, when the axial response shows nothing. I remember patiently explaining to you how I could tell from the data that this was happening. Don't give away ALL of my secrets!!
Duke - this was the line I picked up on. Most people don't even look at the off-axis data let alone look for "energy ridges" to EQ, when the axial response shows nothing. I remember patiently explaining to you how I could tell from the data that this was happening. Don't give away ALL of my secrets!!
