Poll: Should We Get Into Testing Headphones or Speakers?

[scott wurcer]

A software called Monkey Forest generates snappier looking polar maps, but costs a lot more.

I took quick look at the manual, there was nothing in the examples that several open source plot libraries could not do these days.

 

[amirm]

BTW, did you ever get pricing from that place that has the Klippel system in the USA and will test speakers for a fee?

I just did. The baseline cost is $1000 per report if nothing extra is required than the directivity plot. Anything else is $125/hour on top of that.

 

[amirm]

But then I think of REW. The things REW can do for you would have been maybe a $200k+ proposition not all that long ago. Now a spare 10 year old laptop and a $100 calibrated microphone and there is so much you can do. I think it is only a matter of time until what Klippel does will be available in very inexpensive software. You might not ever get a super cheap mechanical apparatus for it to work. Heck, maybe in 10 years all this will be in REW.

The level of complexity and knowledge of math and acoustics here is hugely above what REW does. Earl Geddes says it took him 10 years to come up with the approximation of it in 2-d plane. There is a massive thread about it in the diyaudio site which died down eventually. You pay some $40,000 for the software alone, telling you how confident Klippel is that this is too hard for just about anyone to solve.

 

[dc655321]

The level of complexity and knowledge of math and acoustics here is hugely above what REW does. Earl Geddes says it took him 10 years to come up with the approximation of it in 2-d plane. There is a massive thread about it in the diyaudio site which died down eventually. You pay some $40,000 for the software alone, telling you how confident Klippel is that this is too hard for just about anyone to solve.

Are you referring to "The Waveguide" thread on diyaudio? If not, can you please provide a link to the one you're referencing?
I'm certainly being naive asking this, but what is it about the physics and signal processing of this problem that is "too hard for just about anyone to solve"?

 

[scott wurcer]

The level of complexity and knowledge of math and acoustics here is hugely above what REW does. Earl Geddes says it took him 10 years to come up with the approximation of it in 2-d plane.

It's too bad I followed it early on and Earl was willing to give his code as a library but not in a form that could be converted to another language. It was almost impossible to convince him that you did not have to use FORTRAN for any serious scientific programming (a little tongue in cheek here).

 

[amirm]

It's too bad I followed it early on and Earl was willing to give his code as a library but not in a form that could be converted to another language. It was almost impossible to convince him that you did not have to use FORTRAN for any serious scientific programming (a little tongue in cheek here).

I read, or shall I say suffered through every page of that thread! Earl is a tough guy to deal with. From day one he should have post the Fortran code for people to a) learn what he had done and b) port it to a different language. The notion of someone writing a GUI to just call his code made no sense to me. His code is not sufficient to do the real work and should have been a teaching aid. He had made some simplifying assumptions which should have had a more general solution.

At the end of the thread he did offer to release the code but by then, looked like everyone had lost interest.

The one guy (Dan?) that had sufficient background to get this done was not treated well by Earl which was a shame. He was on the right track to understand the whole thing and then he or others could have easily written the code.

 

[Blumlein 88]

I just did. The baseline cost is $1000 per report if nothing extra is required than the directivity plot. Anything else is $125/hour on top of that.

That doesn't sound too bad really. Like I said, if we could choose 5 benchmark speakers at a few different levels of cost/performance and have them tested it might be a way to start.

 

[Blumlein 88]

The level of complexity and knowledge of math and acoustics here is hugely above what REW does. Earl Geddes says it took him 10 years to come up with the approximation of it in 2-d plane. There is a massive thread about it in the diyaudio site which died down eventually. You pay some $40,000 for the software alone, telling you how confident Klippel is that this is too hard for just about anyone to solve.

I can't say I know what is required, but assumed it was somewhere in the ballpark of what you describe. Yet if things go just the right way I could see it being doable in a few years or a decade. And if they don't go the right way it might remain expensive for 20 years or more.

 

[Blumlein 88]

It's too bad I followed it early on and Earl was willing to give his code as a library but not in a form that could be converted to another language. It was almost impossible to convince him that you did not have to use FORTRAN for any serious scientific programming (a little tongue in cheek here).

Fortran. Learned it after Basic, and prior to COBOL. We could write the universe in Fortran, just might take us a few billion years. I'll take it over COBOL anyway.

 

[amirm]

COBOL was great for formatting control (and hence created the dreaded Y2K problem) but otherwise did suck. FORTRAN wasn't bad at all compared to it. I had to teach both at College. Felt bad teaching people yesterday's technology even then! (1982)

 

[DDF]

COBOL was great for formatting control (and hence created the dreaded Y2K problem) but otherwise did suck. FORTRAN wasn't bad at all compared to it. I had to teach both at College. Felt bad teaching people yesterday's technology even then! (1982)

I wrote my first speaker/xover CAD program in FORTRAN (my program was very similar to what was later released as CALSOD). As an engineering language for relatively low complexity math, it wasn't that bad.

 

[scott wurcer]

I read, or shall I say suffered through every page of that thread! Earl is a tough guy to deal with. From day one he should have post the Fortran code for people to a) learn what he had done and b) port it to a different language. The notion of someone writing a GUI to just call his code made no sense to me. His code is not sufficient to do the real work and should have been a teaching aid. He had made some simplifying assumptions which should have had a more general solution.

At the end of the thread he did offer to release the code but by then, looked like everyone had lost interest.

The one guy (Dan?) that had sufficient background to get this done was not treated well by Earl which was a shame. He was on the right track to understand the whole thing and then he or others could have easily written the code.

You echo exactly what I felt from the beginning.

 

[mitchco]

I hope this becomes a reality for you @amirm. In my limited testing of loudspeakers, I find there is quite a bit of variability between manufacturer specifications versus how the speaker really sounds/measures.

Don't know if this has been posted before, but the ANSI/CEA-2034-A Standard Method of Measurement for In-Home Loudspeakers is available as a free download at:
https://www.cta.tech/Research-Standards/Standards.aspx?search=2034

One needs to create an account, but if you are interested, it is worth the time to register and download the document as it is quite an interesting read. It also provides the basis for estimated in-room response calculation. I.e. providing a usefully accurate preview of how a given loudspeaker might perform in a typical domestic listening room. While the predicted curve is quite accurate, it cannot estimate the effects of room modes below transition/Schroeder. But we know how to take care of that

Note that Revision B is in final draft and will also be available for free download towards the end of the year: https://standards.cta.tech/apps/group_public/project/details.php?project_id=540

Enjoy!

 

[JohnPM]

the ANSI/CEA-2034-A Standard Method of Measurement for In-Home Loudspeakers is available as a free download at:
https://www.cta.tech/Research-Standards/Standards.aspx?search=2034

"Free" in a fairly loose sense, since in return you are required to provide a lot of personal information.

The information collected includes (but is not limited to) your name, job title, employer name, address, email address, phone number, credit card or account information, photos, and information about your interests in and use of various products. In these situations, if you choose to withhold any personal data requested by us, it may not be possible for you to gain access to certain Services and for us to respond to you.

 

[mitchco]

Hi John, having gone through the registration process, one can skip almost of that. A name, phone number and email address is all I provided, which all can be made up and using a throwaway email address, if concerned about privacy.

 

[NTK]

I know next to nothing about the mathematics of acoustics, but I am curious about what Dr. Geddes' software does. My question is mostly out of curiosity. Is it for simulating the acoustic performance of loudspeakers? If so, I wonder if modern commercial FEA program such as COMSOL can do what his software does. FEA is often the only practical tool when you have to deal with complex 3D geometries.
https://www.comsol.com/acoustics-module
(You will probably notice that I "borrowed" one of their loudspeaker SPL simulation plots for my avatar )

I am not advocating any of us getting COMSOL here. A license costs several tens of thousands of dollars. I used it in my previous job for structural, thermal, and fluid flow analyses, but have never touched the acoustics module (my previous company had no reason to get a license for the acoustics module anyway).

I am also curious if somebody skilled in acoustics (I am thinking PhD level knowledge, not me) can get this type of simulations to work using FOSS FEA tools such as FEniCS.

 

[NTK]

Apparently Samsung (i.e. Harman) uses COMSOL for their acoustics modeling too. There is a blog post in the COMSOL website. The video is quite interesting.
https://www.comsol.com/blogs/samsung-amps-up-loudspeaker-designs-with-simulation/

 

[scott wurcer]

I know next to nothing about the mathematics of acoustics, but I am curious about what Dr. Geddes' software does.

I'm to lazy to look it up but Earl used some transforms or systems of equations that were fairly obscure. There were closed or nearly closed form solutions/approximations (I don't remember the details), I assume when he did his original work there either was no FEA or it would have been so compute intensive as to be useless. It could also have been difficult to pose the problem in that form.

We had a PHD expert whose job it was to drive COMSOL for full 3D extraction of package and board parasitics on multi GHz circuits.

For the speaker problem it looked very promising simply using Python/SciPY/Numpy and the open source plot libs.

 

[NTK]

For the speaker problem it looked very promising simply using Python/SciPY/Numpy and the open source plot libs.

If plotting the measurement results is the only issue, it is a much much easier problem to solve

 

[amirm]

I know next to nothing about the mathematics of acoustics, but I am curious about what Dr. Geddes' software does.

It is the idea of measuring a speaker in a room but eliminating the effects of the room on what you measure. Reflections need to be extracted and dialed out of the measurements. And this is not easy.

It is NOT a simulation by the way. We are measuring a real speaker.