Calculating sound power response

[napilopez]

Long story short, I'm considering doing spinorama-type graph in my speaker measurements in addition to the graphs at various angles and listening window I currently do. Early reflections are a simple average, but I'm confused as to how to go about calculating the power response.

According to CTA 2034-A, which defines the spinorama:

"The sound power is the weighted rms average of all 70 measurements, with individual measurements weighted according to the portion of the spherical surface that they represent. Calculation of the sound power curve begins with a conversion from SPL to pressure, a scalar magnitude. The individual measures of sound pressure are then weighted according to the values shown in Appendix C and an energy average (rms) is calculated using the weighted values. The final average is converted to SPL. "

(CTA-2034A is a free document by the way, and a good resource should you ever want to make your own speaker measurements)

I'm lost at converting the SPL graphs to scalar magnitudes and how to perform weighted averages on graph data. Is there any present software that can do so? Plus I'd have to consider moving to 10-degree increments instead of 15, but that's another discussion.

If anyone can lead the way - or otherwise provide a useful way to generate a sound power curve, I'm all ears.

 

[RayDunzl]

I'm lost at converting the SPL graphs to scalar magnitudes

"Calculation of the sound power curve begins with a conversion from SPL to pressure, a scalar magnitude."

Convert SPL to pascals, maybe?

http://www.sengpielaudio.com/calculator-soundlevel.htm

 

[pozz]

Any luck with this @napilopez?

 

[Blumlein 88]

Along with Ray's link this one might help too.
http://www.sengpielaudio.com/calculator-soundpower.htm

 

[amirm]

It is the square root of the sum of squared SPLs*weighting coefficients.

 

[napilopez]

Thanks all for the advice!

My question is more about the processing of the data than the math though - I'm not sure what software I can use to perform those operations, for example. I'll admit I'm mainly familiar with REW and it doesn't seem I can multiply the graphs by their weighting coefficients there, for instance, let alone do so for 70 measurements in one go.

 

[pozz]

@JohnPM Do you have any suggestions?

 

[JohnPM]

Probably best exporting the responses, pulling them into Excel and doing the calculations there.

 

[jhaider]

If anyone can lead the way - or otherwise provide a useful way to generate a sound power curve, I'm all ears.

I can't do the math - I discussed it with Manny LaCarrubba when he started publishing spinoramas a little, but abandoned the effort it quickly became clear to me that I don't have the time or skill to generate spinoramas in Excel.

However, if you have a Parallels or a PC, you can import the impulse responses for each measurement into a crossover design program called Vituixcad and it will now give you CTA-2034A analysis, as well as horizontal and vertical polar maps.
https://kimmosaunisto.net/Software/Software.html

I believe I posted a spin of a speaker I designed in the big Neumann thread.

One hint to save yourself much anguish: Vituixcad is primarily a crossover design problem. You won't see any data until you go to the crossover tab and string a "wire" on the positive leg.

 

[napilopez]

Didn't realize you were on this forum @JohnPM, I'll try not to pester you too much =] and thanks for all your work.

Anyway, I havent made anything in Excel since high school. Guess it's time to learn!

 

[napilopez]

I can't do the math - I discussed it with Manny LaCarrubba when he started publishing spinoramas a little, but abandoned the effort it quickly became clear to me that I don't have the time or skill to generate spinoramas in Excel.

However, if you have a Parallels or a PC, you can import the impulse responses for each measurement into a crossover design program called Vituixcad and it will now give you CTA-2034A analysis, as well as horizontal and vertical polar maps.
https://kimmosaunisto.net/Software/Software.html

I believe I posted a spin of a speaker I designed in the big Neumann thread.

One hint to save yourself much anguish: Vituixcad is primarily a crossover design problem. You won't see any data until you go to the crossover tab and string a "wire" on the positive leg.

Posted a bit too soon, thanks @jhaider! Ive used vituixcad just a little for modelling baffle step for some Nearfield measurements, didn't know it could do anything related to spins. I'll give that a shot and report back.

 

[napilopez]

Sorry for the triple-post, but I think @jhaider solved all my problems. Thank you!

It's incredibly simple. VituixCAD actually automatically generates the full spins as soon as i import all the measurements, so long as they are labeled correctly. It even generates an estimated 'in-room graph,' which we don't normally see in spinorama's but is defined in CTA-2034A too. It almost always ends up looking very close the ER curve anyway, but it's nice to see.

Better yet, I can export the curves back to REW for consistency and more control over data presentation.

Here's a sample of what vituixcad generates using my Sonus Faber measurements. Note that the sound power and spdi curves aren't fully accurate because this is mostly front hemisphere data with one 180-degree curve. Also the data below 200hz is inaccurate because of gated measurements. Just sharing for demonstration purposes.

Should anyone else be considering making their own spins and come across this thread... a few notes and quirks:

• For some reason VirtuixCAD scales the Directivity Index and Early Reflections DI to 3dB increments instead of the usual 5 used for the other curves, making the DI look narrower than it is(see scaling on the right). Can't seem to find a way to change this. Luckily this isn't a problem when imported back to REW, but I spent several minutes trying to figure why the directivity seemed so much narrower until I noticed the different scaling.
• When importing to REW the average curves seem to be about half a dB off, but I'll survive.
• Vituixcad will also automatically generate early reflections and power response curves from 15-degree interval data, which means I can go back and try to make rough spins for some of my older measurements. It can even estimate 'missing' angles down to 5 degree intervals if you want.

The bad news is now I need to actually start doing all 70 spatial measurements. I was just doing enough to generate an early reflections curve before, which is mostly front hemisphere.

 

[pozz]

Congrats

Do you have the data to update your previous reviews with these graphs? If it's not too much work.

 

[napilopez]

Congrats

Do you have the data to update your previous reviews with these graphs? If it's not too much work.

Unfortunately not most of them, or at least not graphs I'm comfortable enough in their accuracy to share. I've been pretty inconsistent in the amount of data I've captured as I puzzled out my testing procedure and how much work I actually wanted to put into these measurements, and vertical data, in particular, has been a pain to capture. Most of the speakers I've measured have been returned.

But now I've made my workflow more efficient and figured out a way to do vertical measurements relatively easily, I should be able to generate quasi-anechoic spins for everything going forward. Hopefully Amir will get his hands on some of these models and be able to provide more/more accurate data.