It would also be nice to have a programmer doing all this work! My first priority is my (high) paying job, then building speakers (low paying job), and lastly developing software (no paying job). So its not going to happen soon.
This is what I originally though. But I also read a paper that seemed to come up with calculations that showed somehwhat closer to the constant directivity. The directivity was calculated to be slightly different at different frequencies. The most difficult issue is controling directivity down to the lower frequencies.
SoundEasy has also been evolving throughout the years, not a money maker, but I do respect the designer's continued effort, and listening to users. I've always supported through continuous upgrades regardless what features are added.
Fortunately its not as important down lower in frequency, so you can get away with less directivity control.
Hello Dr. Geddes,
If you are starting with a blank sheet, maybe you could consider using the same file formats (and import/export options) that Holmimpulse uses. This way it would be possible to switch between packages, with would be excellent for existing Holmimpulse users, but also excellent to debug your software.
Any idea how much work it would be to convert your current application into an executable with no additional features (except import from a Holmimpulse file)? That would already be fantastic for the DIY community to make detailed polar responses...
What you suggest is already the case. Maube if I explain the situation in some more detail it will be clearer.
What I have been doing is taking all the raw data on HOLM and using its Export all curves to export the angular impulse responses as a *.txt file. Then I would read these impulses into MathCAD and process them using the algorithms that I had been developing - over a decade at least. I would view the data in Mathacd - very poor graphics, or I used to export to Sigma Plot, which was pretty good, but very expensive. Signma plot ceased to work in Win7 and I was tired of having to upgrade it with each new OS. SO I wrote the graphics myself, and that is what is posted on my web site. It reads in the output from MathCAD.
But MathCAD has the same problem, even worse. Its very slow (albeit extremely flexible), but I had worked out the algorithms already so I didn't need flexibility. SO I have been writing the algorithms into an exe under dot-net. The polar map graphics will simply become a graphics page for viewing the output in the new program. Hence, what I am working on WILL read Holm data sets (and WinMLS or anything else that can export txt file impulse responses I guess).
I may, down the road, eliminate the need for HOLM down the road if it ceases to be reliable.
Hope this helps.
After reading the controlled directivity vs constant directivity post by Geddes. I became completely confused on the CD topic.
Like Davey, Im confused about it now because it seems those terms have been interchangeable in this and other threads. Maybe Davey's question was lost in the OT stuff.
Could someone define the distinct differences between "controlled Directivity" and "constant Directivity"?? For the life of me I can not see what differences are pointing to controlled directivity in the polar plots.
It would certainly be nice to be able to hear systems that have low frequency directivity control to feel the difference though.
In simplest terms, constant directivity could be interpreted as meaning the polar response is not a function of frequency. The directivity index could still vary with off axis angle. Controlled directivity would be less stringent. Controlled directivity would imply that the polar response varies in some prescribed manner both with frequency and angle. Thus a constant directivity system would be a more restrictive subset on controlled directivity. Dipoles and cardioid certainly fall into the category of constant directivity. Dipoles and cardioids are 1st oder gradient systems. If the requirement is for a narrower beam width then approaches other than 1st order gradient systems must be employed. These include horns, wave guides, 2nd order gradient system, nonlinear acoustics, and others. Horns and wave guides work well at higher frequency but are not so practical at lower frequency. 1st and 2nd order gradient systems work well at low frequency but, do to the 2nd order roll off, 2nd order systems require very large excursion and excessive equalization and again are not practical over a wide frequency range. Nonlinear acoustics is probably the most promising approach for a wide range solution.
I would have to disagree with Earl with regards to a dipole not being controlled directivity. (Not surprising.) It may not fill his requirement for beam width, but directivity is certainly well controlled when correcly executed.
By the way, ARTA can produce the type of sonograms (contour plots) {as well as surface plots (water falls) and polar plots} that have been presented here. Of course, the ATAR measurement system cost about $100.
I would have to disagree with Earl with regards to a dipole not being controlled directivity. (Not surprising.
) It may not fill his requirement for beam width, but directivity is certainly well controlled when correcly executed.By the way, ARTA can produce the type of sonograms (contour plots) {as well as surface plots (water falls) and polar plots} that have been presented here. Of course, the ATAR measurement system cost about $100.
Last edited:
To me, the point is that Constant Directivity is not restrictive enough, and so in this sense I believe that Controlled Directivity is the more stringent (quite the opposite of John). Omni and dipole are both constant directivity, but neither is very directive in the forward direction. Perhaps the best term would be High-Q Constant Directivity - its just a little wordy - or Controlled Constant Directivity (still ambiguous).
A dipole is not High-Q CD and hence I did not consider it "Controlled Directivity" if the goal was High-Q. There are no standard deffinitions to any of this stuff, which is why people use them in whatever fasion they want and John and I have completely opposite definitions. It's no wonder its confusing. When Constant Directivity was first used in the Pro world it implied High-Q. I had maintained that implied assumption, but then people started being more literal about it and started calling things like monopoles and dipoles CD, which is completely inconsistant with the initial usage of the term. So I started using CD to mean Controlled Directivity, meaning that the directivity was specified to be something required by the application and not simply the default directivity defined by the implimentation. But now even this is being warped. I suppose that I'll have to change terms again.
Should it be A? or B?
A dipole is not High-Q CD and hence I did not consider it "Controlled Directivity" if the goal was High-Q. There are no standard deffinitions to any of this stuff, which is why people use them in whatever fasion they want and John and I have completely opposite definitions. It's no wonder its confusing. When Constant Directivity was first used in the Pro world it implied High-Q. I had maintained that implied assumption, but then people started being more literal about it and started calling things like monopoles and dipoles CD, which is completely inconsistant with the initial usage of the term. So I started using CD to mean Controlled Directivity, meaning that the directivity was specified to be something required by the application and not simply the default directivity defined by the implimentation. But now even this is being warped. I suppose that I'll have to change terms again.
Should it be A? or B?
C.
Don't get mad at me Earl but perhaps a better term for you systems would be controlled beam width?. Even I would not argue with that. Your design objective clearly appears to be to control beam width above some frequency below which you consider it unimportant. I may not agree with what happens at low frequency in you systems but I would not argue on that point.
I think the point of contention that arises is centered around the term directivity. Used loosely, one may interpret it as a reference to the direction sound propagates. Used more technically, one may consider is as relating to the behavior of the directivity index as defined for acoustic sources.
Don't get mad at me Earl but perhaps a better term for you systems would be controlled beam width?. Even I would not argue with that. Your design objective clearly appears to be to control beam width above some frequency below which you consider it unimportant. I may not agree with what happens at low frequency in you systems but I would not argue on that point.
I think the point of contention that arises is centered around the term directivity. Used loosely, one may interpret it as a reference to the direction sound propagates. Used more technically, one may consider is as relating to the behavior of the directivity index as defined for acoustic sources.
By way of some further explainations:
The "order" of the radiation pattern comes from the Legendre polynomials that define the angular radiation form a sphere. There are an infitite number of them starting at zero. The zero order function is 1.0 meaning that the far field pressure is independent of theta - this is also called a monopole. The first order function is Cos(theta) also called a dipole, the second is .75 cos( 2 * theta) + .25, and they get evermore complex as the order goes up, and, most notably, narrower. Now ALL of these orders are CD by anyones deffinition simply because they have no frequency dependence of the theta functions (as long as there is only one term as we shall see).
For a source to have anything but omni or dipole directivity it must contain higher orders. The narrower the directivity, the higher the order that is required. Directional sources, like waveguides, contain many orders all summed together and this is were the frequency dependence comes in. Each order will have a different frequency dependence and hence the sum of these orders will be different at every frequency. Getting this sum to all add in precisley the manner that is desired is extremely complex and, in fact it is impossible to do exactly.
So if one takes Constant Directivity in its most restrictive sense then ONLY a monopole and a dipole can be Constant Directivity - a strange situation since the original deffinition would have excluded these two. So any directivity narrower than a dipole is guaranteed to not be CD in the strict sense. This is what John noted in his earlier comments. Getting Hi-Q Constant Directivity is extremely difficult because for all practical purposes it is impossible - there are better and worse implimentations, but there are no perfect ones. In the chapter on Waveguides in my book, I show why this is so and how to obtain a design which is as close as possible to the desired goal. This technique was used by some guy in Australia as his MS thesis where he used a complex computer algorithm to find the ideal contour of a waveguide that meets the design criteria. His results showed that the OS contour is very close to being optimal, but he used one constraint that I would not have used. He optimized for maximally flat axial response, which we know is not a design feature of the OS. Reduce that constraint and the OS (with mouth flare of course) does turn out to be optimal. But its still not perfect.
The "order" of the radiation pattern comes from the Legendre polynomials that define the angular radiation form a sphere. There are an infitite number of them starting at zero. The zero order function is 1.0 meaning that the far field pressure is independent of theta - this is also called a monopole. The first order function is Cos(theta) also called a dipole, the second is .75 cos( 2 * theta) + .25, and they get evermore complex as the order goes up, and, most notably, narrower. Now ALL of these orders are CD by anyones deffinition simply because they have no frequency dependence of the theta functions (as long as there is only one term as we shall see).
For a source to have anything but omni or dipole directivity it must contain higher orders. The narrower the directivity, the higher the order that is required. Directional sources, like waveguides, contain many orders all summed together and this is were the frequency dependence comes in. Each order will have a different frequency dependence and hence the sum of these orders will be different at every frequency. Getting this sum to all add in precisley the manner that is desired is extremely complex and, in fact it is impossible to do exactly.
So if one takes Constant Directivity in its most restrictive sense then ONLY a monopole and a dipole can be Constant Directivity - a strange situation since the original deffinition would have excluded these two. So any directivity narrower than a dipole is guaranteed to not be CD in the strict sense. This is what John noted in his earlier comments. Getting Hi-Q Constant Directivity is extremely difficult because for all practical purposes it is impossible - there are better and worse implimentations, but there are no perfect ones. In the chapter on Waveguides in my book, I show why this is so and how to obtain a design which is as close as possible to the desired goal. This technique was used by some guy in Australia as his MS thesis where he used a complex computer algorithm to find the ideal contour of a waveguide that meets the design criteria. His results showed that the OS contour is very close to being optimal, but he used one constraint that I would not have used. He optimized for maximally flat axial response, which we know is not a design feature of the OS. Reduce that constraint and the OS (with mouth flare of course) does turn out to be optimal. But its still not perfect.
John, I think the point is that my usage has always been consistant with the defacto standard used in the Pro industry. Why is it that we have to redefine all of our terms for use in DIY?