Looking For Anechoic Data

[DDF]

Hi, all

I'm looking for loudspeaker data measured in an anechoic chamber in at least two different conditions: for example, IEC baffle and in box, or two different baffles.

Looking for 5 to 10" driver data, and with known measurement conditions (mic position, driver position). If anyone knows of a resource on-line, please post.

Thanks in advance.

Dave

[Jeff B.]

Re: Looking For Anechoic Data

Hi, all

I'm looking for loudspeaker data measured in an anechoic chamber in at least two different conditions: for example, IEC baffle and in box, or two different baffles.

Looking for 5 to 10" driver data, and with known measurement conditions (mic position, driver position). If anyone knows of a resource on-line, please post.

Thanks in advance.

Dave

Boy....I bet I know where this is leading :eek:

I was going inquire about something from the study you recently sent comparing the diffraction signatures. There had to be certain assumptions made about the data, and after giving it some thought, I questioned whether all of the required assumptions could be valid. For example, "Anechoic chambers" aren't really anechoic at all, at least not at all frequencies. Most are only rated down to a certain frequency limit, and below that room effects still begin to affect the lower frequencies some. Based on that, I am not sure how to determine the true reference level for the driver at 150 Hz compared to a mathematical model with the precision that was being applied. Am I making any sense? Let me know if my conclusions are off base here.

Jeff

[dlr]

Re: Looking For Anechoic Data

Boy....I bet I know where this is leading :eek:

I was going inquire about something from the study you recently sent comparing the diffraction signatures. There had to be certain assumptions made about the data, and after giving it some thought, I questioned whether all of the required assumptions could be valid. For example, "Anechoic chambers" aren't really anechoic at all, at least not at all frequencies. Most are only rated down to a certain frequency limit, and below that room effects still begin to affect the lower frequencies some. Based on that, I am not sure how to determine the true reference level for the driver at 150 Hz compared to a mathematical model with the precision that was being applied. Am I making any sense? Let me know if my conclusions are off base here.

Jeff

This looks like it could be an interesting thread. Here's hoping.

dlr

[DDF]

Re: Looking For Anechoic Data

Boy....I bet I know where this is leading :eek:

I was going inquire about something from the study you recently sent comparing the diffraction signatures. There had to be certain assumptions made about the data, and after giving it some thought, I questioned whether all of the required assumptions could be valid. For example, "Anechoic chambers" aren't really anechoic at all, at least not at all frequencies. Most are only rated down to a certain frequency limit, and below that room effects still begin to affect the lower frequencies some. Based on that, I am not sure how to determine the true reference level for the driver at 150 Hz compared to a mathematical model with the precision that was being applied. Am I making any sense? Let me know if my conclusions are off base here.

Jeff

Anechoic chambers are better than perhaps you've been led to believe. I worked in 3 anechoic chambers almost every day for 3 or 4 years and on and off for 10 years. I also researched them for purchase. Below about 50 Hz they can become challenged. The NRC chamber which Toole worked with for so many years was rumoured to be miscalibrated below 50 Hz, for example (I can't validate this but it was a rumour floating around the industry in Ottawa, and it does illustrate the challenges at the very lowest frequencies). However, 100Hz is a walk in the park for an even modest chamber. I used to perform free space design down to 100Hz all the time in a somewhat modest anechoic room. Our large chamber laughed at 100Hz. I have no reason to doubt Seas' chamber as evidenced by the lack of low frequency room effects in the data.

Don't fear what I'm analyzing. My intent is merely to understand the limits of the applicability of the tools at hand. This is in no way is a discredit to their obvious utility, its merely standard sound engineering practice. I also enjoy doing this. I've written my own audio tools and models and I always enjoyed calibrating them and analyzing their limitations.

For example, if there are other concerns with my analysis, I'd love to hear them so that they can be considered.

One concern I have is that a tweeter was used for the analysis, and its unclear that the SNR was high at the lowest frequencies. It certainly appeared to be, but you can never be too careful and this is one reason I'm looking for woofer anechoic data.

Dave

[Jeff B.]

Re: Looking For Anechoic Data

Anechoic chambers are better than perhaps you've been led to believe. I worked in 3 anechoic chambers almost every day for 3 or 4 years and on and off for 10 years. I also researched them for purchase. Below about 50 Hz they can become challenged. The NRC chamber which Toole worked with for so many years was rumoured to be miscalibrated below 50 Hz, for example (I can't validate this but it was a rumour floating around the industry in Ottawa, and it does illustrate the challenges at the very lowest frequencies). However, 100Hz is a walk in the park for an even modest chamber. I used to perform free space design down to 100Hz all the time in a somewhat modest anechoic room. Our large chamber laughed at 100Hz. I have no reason to doubt Seas' chamber as evidenced by the lack of low frequency room effects in the data.

Don't fear what I'm analyzing. My intent is merely to understand the limits of the applicability of the tools at hand. This is in no way is a discredit to their obvious utility, its merely standard sound engineering practice. I also enjoy doing this. I've written my own audio tools and models and I always enjoyed calibrating them and analyzing their limitations.

For example, if there are other concerns with my analysis, I'd love to hear them so that they can be considered.

One concern I have is that a tweeter was used for the analysis, and its unclear that the SNR was high at the lowest frequencies. It certainly appeared to be, but you can never be too careful and this is one reason I'm looking for woofer anechoic data.

Dave

I have no fears. I am just looking to understand the process you are implementing. If you have accurate anechoic response data on an IEC baffle, and in a loudspeaker enclosure with a much smaller baffle size, both under identical measurement conditions, and you take the difference between them it won't be a pure diffraction curve. The IEC baffle has a diffraction signature of its own - All you end up with is the difference between the two. The large IEC baffle should have a diffracton peak in the 100Hz region. So, with this being the case, how are you determining the true reference response of the the driver, in order to determine the exact amount of rise in the 100Hz to 500 Hz for the smaller baffle? The math used in the diffraction models simple shows the shift from 2pi to 4pi steradian space through this region. It would seem that if measurements don't match that exactly then we can't rule out that something outside the math, but in the data, is contributing to the response difference.

Now, if we can define that, and make the model more accurate, then I am all for it. See....I am just looking to understand this as well.

[dlr]

Re: Looking For Anechoic Data

Are no measurements made in an anechoic chamber that does not use an IEC baffle as a large baffle?

I would think that the best way to investigate diffraction models would be to use a smallish midrange unit with wide dispersion on a large circular baffle with a center mounted driver. This maximizes the impact and minimizes it's spread in the spectrum.

dlr

[DDF]

Re: Looking For Anechoic Data

I have no fears. I am just looking to understand the process you are implementing. If you have accurate anechoic response data on an IEC baffle, and in a loudspeaker enclosure with a much smaller baffle size, both under identical measurement conditions, and you take the difference between them it won't be a pure diffraction curve. The IEC baffle has a diffraction signature of its own - All you end up with is the difference between the two. The large IEC baffle should have a diffracton peak in the 100Hz region. So, with this being the case, how are you determining the true reference response of the the driver, in order to determine the exact amount of rise in the 100Hz to 500 Hz for the smaller baffle? The math used in the diffraction models simple shows the shift from 2pi to 4pi steradian space through this region. It would seem that if measurements don't match that exactly then we can't rule out that something outside the math, but in the data, is contributing to the response difference.

Now, if we can define that, and make the model more accurate, then I am all for it. See....I am just looking to understand this as well.

My analysis was taken with a tweeter: the sealed cavity ensured that there was no difference in low frequency loading between the two scenarios, IEC baffle and in-box. Taking the IEC baffle diffraction signature and removing its affect takes the response to a virtual point source with no baffle. Adding back in the box diffraction represents its response in the box. As shown by the accuracy of the models attached, the analysis is pretty sound.

The disruptions at higher frequencies could be surround reflections as I have a suspicion the tweeter in box was taken with a woofer also in box,

The analysis with the woofers would have to include box loading effects. The analysis should include a woofer in-box and in the same box but the front baffle being IEC. Chances of getting this are slim, but its worth seeing what's available.

The model accuracy also depends upon the angle off axis. For example, a paper in the AES found Vanderkooy's diffraction model was amongst the most accurate on axis, but not so off axis.

Without a significant amount of data, no analysis would be definitive and complete. I'm looking for a few data points to baseline the model for my own purposes. It's better than no baselining at all.

I've also seen some the common TL models out there handled stuffing differently than the latest state of the art that I was aware of. So, I've asked in the past for measurements for systems built against the TL models, just to baseline these models as well. This sort of information is very hard to come by. We tend to trust all our models without understanding their limitations. I think better sounding speakers can be made with less recursion if we understood the strengths and limitations of our models.

Dave

[fish-24_7]

Re: Looking For Anechoic Data

This is a very interesting topic. I will be following it closely. Like someone else said, I too hope it can stay on its current course.

Thanks Gentlemen!

[daryl]

Re: Looking For Anechoic Data

Hi DDF,

Are you looking to write your own diffraction algorithm?

I have been thinking how I could do this and figured it out earlier this year.

I came up with a cool way to implement the calculations which was completely in the opposite direction of where I thought I would find the answer.

I just don't know how to use a fast language like assembly or C.

I also would need to know how to write the code in multiple threads, my current processor is dual core.

My method also will require vast quantities of memory.

I have 8 gigabytes of ram but that may not be enough.

I have thought about using usb flash drives but they might wear out rather quickly with the number of times they would be re-written in each simulation.

Anyway my method would use a rediculous amount of calculation but you would be able to simulate horns, wave guides, baffles as well as internal baffling of enclosures.

The stumbling block to getting started is finding a way to quickly learn a fast programming language.

I have done some neat stuff using the visual basic in excel but I would not even be able to get started using excel for this idea.

I found an online visual C++ tutorial but it was rediculous.

I read the thing for hours and was no nearer being able to write a program than when I started.

It started off by having me use this drag and drop box in the editor (the visual part I guess).

It said do this, then this and then this and congratulations you've just created you first C++ program!

No mention why you were doing any of these things how they worked.

I was expecting to get a list of all the commands available, how each one works, the proper syntax, the range and instrucions relating to the operation of the compiler.

Then I would want to know how multi-threading is done and what would be really cool is being able to use the GPGPU capabilities of the newer GPU's which allow you to write code for GPU's not for graphics but for scientific use (a teraflop can be achieved on a $200 card).

[Jeff B.]

Re: Looking For Anechoic Data

Hi DDF,

Are you looking to write your own diffraction algorithm?

I have been thinking how I could do this and figured it out earlier this year.

I came up with a cool way to implement the calculations which was completely in the opposite direction of where I thought I would find the answer.

I just don't know how to use a fast language like assembly or C.

I also would need to know how to write the code in multiple threads, my current processor is dual core.

My method also will require vast quantities of memory.

I have 8 gigabytes of ram but that may not be enough.

I have thought about using usb flash drives but they might wear out rather quickly with the number of times they would be re-written in each simulation.

Anyway my method would use a rediculous amount of calculation but you would be able to simulate horns, wave guides, baffles as well as internal baffling of enclosures.

The stumbling block to getting started is finding a way to quickly learn a fast programming language.

I have done some neat stuff using the visual basic in excel but I would not even be able to get started using excel for this idea.

I found an online visual C++ tutorial but it was rediculous.

I read the thing for hours and was no nearer being able to write a program than when I started.

I started off by having me use this drag and drop box in the editor (the visual part I guess).

It said do this, then this and then this and congratulations youve just created you first C++ program!

No mention why you were doing any of these things how they worked.

I was expecting to get a list of all the commands available, how each one works, the proper syntax, the range and instrucions relating to the operation of the compiler.

Then I would want to know how multi-threading is done and what would be really cool is being able to use the GPGPU capabilities of the newer GPU's which allow you to write code for GPU's not for graphics but for scientific use (a teraflop can be achieved on a $200 card).

And yet, I have written a new one in Excel that does open or closed baffle diffraction calculating to any axial location in the forward hemisphere, with any edge radius, with piston directivity calcuated in, and it even includes boundary reinforcement from the walls and floor, and......it is only 380kB in size as a file. Why do you need some much ram and computing power to do this simple calculation? My diffraction array has 20,000 edge magnitude vs frequency points to sum and it's still a fairly small and very fast program - fast enough that you can click the spinners and see the response change in realtime.

Jeff

[brianpowers27]

Re: Looking For Anechoic Data

I also would need to know how to write the code in multiple threads, my current processor is dual core.

My method also will require vast quantities of memory.

I have 8 gigabytes of ram but that may not be enough.

I have thought about using usb flash drives but they might wear out rather quickly with the number of times they would be re-written in each simulation.

Anyway my method would use a rediculous amount of calculation but you would be able to simulate horns, wave guides, baffles as well as internal baffling of enclosures.

The stumbling block to getting started is finding a way to quickly learn a fast programming language.

I have done some neat stuff using the visual basic in excel but I would not even be able to get started using excel for this idea.

I found an online visual C++ tutorial but it was rediculous.

I read the thing for hours and was no nearer being able to write a program than when I started.

It started off by having me use this drag and drop box in the editor (the visual part I guess).

It said do this, then this and then this and congratulations you've just created you first C++ program!

No mention why you were doing any of these things how they worked.

I was expecting to get a list of all the commands available, how each one works, the proper syntax, the range and instrucions relating to the operation of the compiler.

Then I would want to know how multi-threading is done and what would be really cool is being able to use the GPGPU capabilities of the newer GPU's which allow you to write code for GPU's not for graphics but for scientific use (a teraflop can be achieved on a $200 card).

Daryl,


You may already be past the basics so I apologize if you are... Taking on a programming language is best done through guided study but unfortunately most online curriculums will leave you short on necessary skills. The first course I took towards my major helped immensely. It does not address any language specific constructs but instead focuses on general design logic.

Once you are past this point, you might consider downloading the express edition of Microsoft Visual Basic.net and work with what you know. I doubt that c++ is going to give you any large performance boost over Visual Basic.net.

BTW: Multi-threading is outside of my league.


"Programming Logic and Design"

http://www.amazon.com/Programming-Logic-Design-Joyce-Farrell/dp/0619160217

[daryl]

Re: Looking For Anechoic Data

And yet, I have written a new one in Excel that does open or closed baffle diffraction calculating to any axial location in the forward hemisphere, with any edge radius, with piston directivity calcuated in, and it even includes boundary reinforcement from the walls and floor, and......it is only 380kB in size as a file. Why do you need some much ram and computing power to do this simple calculation? My diffraction array has 20,000 edge magnitude vs frequency points to sum and it's still a fairly small and very fast program - fast enough that you can click the spinners and see the response change in realtime.

Jeff

Hi Jeff,

I see the BDS and Edge and as well as your own baffle simulators and I am curious as to how the calculations are done and what assumptions are made.

I also have heard talk about there being differences in the results of these and which might be correct.

I have not been exposed to the specifics of how anyone else would accomplish such simulations but I am a person who likes to come up with my own methods.

The reason for the vast recources required for my method has to do with how it is accomplished.

Without knowing how your method (or the BDS or Edge) is accomplished I think my method is probably much simpler but relies on brute force.

Also so long as you have enough memory and calculation speed my method will solve for drivers in horns/wave guides of any shape (you will enter the ENTIRE shape of the cabinet including the rear, the baffle might be shaped like a horn or anything and the rear might be a transmission line) as well as acount for nearby boundries or objects of any shape or location (might be behind the speaker cabinet).

The scope is vastly different.

My method is simply a three dimensional wave table where you create objects to place inside it and apply energy and pick measurement point.

I'm shure if you already thought so but the massive memory isn't for the code but to store the pressure, velocity and impedance data for each point in the wave table.

Thus the memory is a place holder to be used during the calculations and not used before or after.

For instance if the wave table had a resolution of 1000 x 1000 x 1000 you would have a billion points to store data during the calculations.

[daryl]

Re: Looking For Anechoic Data

Here it goes Jeff (and all),

Last October there was a thread on the Audio Circle site where a member was having difficulty with amplifier stability which was situation dependant.

I knew what the problem was and jumped in.

Then I set about writing an algorithm in Excel VB to simulate speaker wire as a transmission line (the thread was in it's 5th page by the time I finished, on the 6th page I added more features).

I did this by substituting a multi-section lowpass filter for the transmision line.

You can enter how many filter sections you wish to calculate (thousands if necessary, the finer you chop (more filter sections) R, L and C the greater the high frequency accuracy).

In effect you enter the characteristics for your speaker wire and the characteristics for the load it's terminated with then the algorithm assigns values to all of the resistors, inductors and capacitors in the circuit based upon the line and load specifications and then solves the entire circuit to find it's input impedance and transfer function.

I had been thinking for while that I could use this method also to model an acoustic wave table by creating a three dimensional lowpass filter circuit but the problem was that I don't seem to be savvy enough to solve such a complicated circuit with defined injection and measurement points.

If I could solve the circuit I wouldn't need all of the memory I was talking about.

Early this year it came to me that I could use the large circuit and instead of solving it in the frequency domain I could solve in the time domain by storing the instantaneous charge for all the componets in the circuit in memory then calculating what the change would be a moment later and resetting all of the values.

Then advance time moment by moment.

The finer you chop space and time the more high frequency accuracy you can achieve.

By recording the potential of any capacitor in the circuit as the calculations progress you will have the impulse response for the corresponding point in space which can be converted to frequency domain transfer function.

The number of calculations is just too great to use Excel.

Maybe you could elaborate as to how your method works I'm probably not the only one who would be interested in the details?

[JRT]

Re: Looking For Anechoic Data

Bjarne Stroustrup ("BS") webpage:

http://www2.research.att.com/~bs/C++.html

Programming: Principles and Practice Using C++

http://www2.research.att.com/~bs/programming.html

http://www.amazon.com/gp/product/0321543726

The C++ Programming Language: Special Edition

http://www2.research.att.com/~bs/3rd.html

http://www.amazon.com/gp/product/0201700735

I know very little to nothing about the subject, and wouldn't want to imply otherwise.

Bjarne Stroustrup is an expert on the subject, created C++, and enjoys a good reputation for his C++ textbooks.

http://en.wikipedia.org/wiki/Bjarne_Stroustrup

.

[daryl]

Re: Looking For Anechoic Data

Daryl,


You may already be past the basics so I apologize if you are... Taking on a programming language is best done through guided study but unfortunately most online curriculums will leave you short on necessary skills. The first course I took towards my major helped immensely. It does not address any language specific constructs but instead focuses on general design logic.

Once you are past this point, you might consider downloading the express edition of Microsoft Visual Basic.net and work with what you know. I doubt that c++ is going to give you any large performance boost over Visual Basic.net.

BTW: Multi-threading is outside of my league.


"Programming Logic and Design"
http://www.amazon.com/Programming-Lo.../dp/0619160217

I just want to know how to use the language all of the available commands, how they work and the proper syntax.

I can figure out how to structure my own program (I can't imagine it even being a challenge for anyone doing physics simulations).

Assembly language or Fortran might be more suitable, I'm not shure.

I need to be able to make the most effective use of the processor.

This visualbasic.net (sounds more like a website) is it a compiled language or interpeted?