LOUDSPEAKER DATA – Pat Brown (2006)

/ Fedele De Marco

LOUDSPEAKER DATA – Pat Brown (2006)

“๐‘ƒ๐‘Ž๐‘ก…๐‘ ๐‘๐‘’๐‘Ž๐‘˜๐‘–๐‘›๐‘” ๐‘œ๐‘“ ๐‘™๐‘œ๐‘ข๐‘‘๐‘ ๐‘๐‘’๐‘Ž๐‘˜๐‘’๐‘Ÿ ๐‘‘๐‘Ž๐‘ก๐‘Ž ๐‘œ๐‘กโ„Ž๐‘’๐‘Ÿ ๐‘กโ„Ž๐‘Ž๐‘› ๐‘กโ„Ž๐‘’ ๐‘ก๐‘–๐‘š๐‘’ ๐‘–๐‘ก ๐‘ก๐‘Ž๐‘˜๐‘’๐‘ , ๐‘คโ„Ž๐‘Ž๐‘ก ๐‘ก๐‘Ÿ๐‘Ž๐‘‘๐‘’ ๐‘œ๐‘“๐‘“๐‘  ๐‘Ž๐‘Ÿ๐‘’ ๐‘กโ„Ž๐‘’๐‘Ÿ๐‘’ ๐‘ค๐‘–๐‘กโ„Ž โ„Ž๐‘–๐‘”โ„Ž๐‘’๐‘Ÿ ๐‘Ÿ๐‘’๐‘ ๐‘œ๐‘™๐‘ข๐‘ก๐‘–๐‘œ๐‘› ๐‘š๐‘’๐‘Ž๐‘ ๐‘ข๐‘Ÿ๐‘’๐‘š๐‘’๐‘›๐‘ก๐‘ ?”

๐—ฃ๐—ฎ๐˜ ๐—•๐—ฟ๐—ผ๐˜„๐—ป: Higher frequency resolution produces more detail in the magnitude (and phase) response of the loudspeaker for each measurement angle.

Higher angular resolution produces more detail in the radiation balloon.

So, it’s all about detail, and higher detail better resolves the

response of the loudspeaker. The only details that we are interested in are those that

a. can be attributed to the loudspeaker

b. will be consistent from unit-to-unit

As detail is increased you are going to see artifacts of the measurement system (mic, loudspeaker rotator, measurement environment) as well as loudspeaker characteristics that will vary from unit-to-unit.

“๐ป๐‘œ๐‘ค ๐‘‘๐‘œ๐‘’๐‘  ๐‘œ๐‘›๐‘’ ๐‘‘๐‘’๐‘ก๐‘’๐‘Ÿ๐‘š๐‘–๐‘›๐‘’ ๐‘กโ„Ž๐‘Ž๐‘ก ๐‘Ž ๐‘™๐‘‘๐‘ ๐‘๐‘˜๐‘Ÿ ๐‘๐‘Ž๐‘› ๐‘๐‘’ ๐‘Ž๐‘‘๐‘’๐‘ž๐‘ข๐‘Ž๐‘ก๐‘’๐‘™๐‘ฆ ๐‘โ„Ž๐‘Ž๐‘Ÿ๐‘Ž๐‘๐‘ก๐‘’๐‘Ÿ๐‘–๐‘ง๐‘’๐‘‘ ๐‘Ž๐‘ก ๐‘™๐‘œ๐‘ค๐‘’๐‘Ÿ ๐‘Ÿ๐‘’๐‘ ๐‘œ๐‘™๐‘ข๐‘ก๐‘–๐‘œ๐‘› ๐‘ค๐‘–๐‘กโ„Ž๐‘œ๐‘ข๐‘ก ๐‘ก๐‘Ž๐‘˜๐‘–๐‘›๐‘” โ„Ž๐‘–๐‘”โ„Ž๐‘’๐‘Ÿ ๐‘Ÿ๐‘’๐‘ ๐‘œ๐‘™๐‘ข๐‘ก๐‘–๐‘œ๐‘› ๐‘š๐‘’๐‘Ž๐‘ ๐‘ข๐‘Ÿ๐‘’๐‘š๐‘’๐‘›๐‘ก๐‘ ?”

๐—ฃ๐—ฎ๐˜ ๐—•๐—ฟ๐—ผ๐˜„๐—ป: You proceed on experience gained by measuring similar devices.

“๐ด๐‘ก ๐‘คโ„Ž๐‘Ž๐‘ก ๐‘‘๐‘’๐‘”๐‘Ÿ๐‘’๐‘’ ๐‘œ๐‘“ ๐‘Ÿ๐‘’๐‘ ๐‘œ๐‘™๐‘ข๐‘ก๐‘–๐‘œ๐‘› ๐‘‘๐‘œ๐‘’๐‘  ๐‘Ž๐‘›๐‘ฆ โ„Ž๐‘–๐‘”โ„Ž๐‘’๐‘Ÿ ๐‘Ÿ๐‘’๐‘ ๐‘œ๐‘™๐‘ข๐‘ก๐‘–๐‘œ๐‘› ๐‘๐‘’๐‘๐‘œ๐‘š๐‘’ ๐‘ข๐‘›๐‘—๐‘ข๐‘ ๐‘ก๐‘–๐‘“๐‘–๐‘’๐‘‘?”

๐—ฃ๐—ฎ๐˜ ๐—•๐—ฟ๐—ผ๐˜„๐—ป:

1. When it does not enhance the use of the data for *estimating* the performance of real-world sound systems. No matter the resolution, the response cannot be predicted exactly.

2. When the extra detail provided is exquisitely sensitive to variables outside of the control of the system designer (that are always present).

For example, one can resolve with great accuracy the lobing that occurs at xover in a 2-way loudspeaker, but ultimately the final response is temperature-dependent and can only be predicted for the ideal case. No matter what you predict, something different will happen (like the weather). A designer should be content knowing that the lobing will occur, without getting overly concerned with exactly predicting the pattern on the audience (which is one of the reasons cited for higher res data).

Don’t get me wrong. I’m not against higher resolutions. I measure them on a regular basis. But higher than 5-deg angular is a “special case” and should not be required for all loudspeaker types. 1/1-octave frequency resolution is high enough for most system design tasks, especially if you are doing acoustical predictions as part of your design. 1/3-oct is defensible for direct field predictions.

“๐‘„๐‘ข๐‘’๐‘ ๐‘ก๐‘œ ๐‘’ฬ€ ๐‘–๐‘™ ๐‘›๐‘œ๐‘๐‘๐‘–๐‘œ๐‘™๐‘œ ๐‘‘๐‘’๐‘™๐‘™๐‘Ž ๐‘ž๐‘ข๐‘’๐‘ ๐‘ก๐‘–๐‘œ๐‘›๐‘’. ๐ฟ๐‘Ž ๐‘ฃ๐‘’๐‘Ÿ๐‘–๐‘ก๐‘Žฬ€ ๐‘’ฬ€ ๐‘โ„Ž๐‘’ ๐‘๐‘œ๐‘โ„Ž๐‘–๐‘ ๐‘ ๐‘–๐‘š๐‘–

๐‘๐‘Ÿ๐‘œ๐‘”๐‘’๐‘ก๐‘ก๐‘–๐‘ ๐‘ก๐‘– ๐‘‘๐‘– ๐‘ ๐‘–๐‘ ๐‘ก๐‘’๐‘š๐‘– ๐‘Ž๐‘ข๐‘‘๐‘–๐‘œ ๐‘–๐‘› ๐‘ก๐‘ข๐‘ก๐‘ก๐‘œ ๐‘–๐‘™ ๐‘š๐‘œ๐‘›๐‘‘๐‘œ ๐‘ ๐‘“๐‘Ÿ๐‘ข๐‘ก๐‘ก๐‘Ž๐‘›๐‘œ ๐‘Ž๐‘๐‘๐‘–๐‘’๐‘›๐‘œ ๐‘™๐‘’ ๐‘๐‘Ž๐‘๐‘Ž๐‘๐‘–๐‘ก๐‘Žฬ€ ๐‘‘๐‘’๐‘– ๐‘‘๐‘Ž๐‘ก๐‘– ๐‘๐‘œ๐‘› ๐‘ข๐‘›๐‘Ž ๐‘Ÿ๐‘–๐‘ ๐‘œ๐‘™๐‘ข๐‘ง๐‘–๐‘œ๐‘›๐‘’ ๐‘‘๐‘– 1 ๐‘œ๐‘ก๐‘ก๐‘Ž๐‘ฃ๐‘Ž ๐‘’ 10 ๐‘”๐‘Ÿ๐‘Ž๐‘‘๐‘–.”

๐—ฃ๐—ฎ๐˜ ๐—•๐—ฟ๐—ผ๐˜„๐—ป:

Again, I am not against high resolution data, but the above statement is

the current reality in the overall predictive process. A complete sound

system design should include:

1. direct field modeling (coverage and level)

2. reflected-field predictions

3. reverberant field predictions

4. noise predictions

5. speech intelligibility estimates

Each of these must be done for EACH 1/N-octave band, where N is

1,3,6,9,… etc. This whole discussion is about the value of N.

N = 1 produces 10 octave bands that must be *individually* considered

for the above criteria.

N = 3 produces about 30 bands.

Current assessment of all but the direct field is limited to N = 1 for a

variety of reasons.

The direct field is the only field of those listed that can be

practically examined at higher-than-1/N-octave resolution. The next

logical choice is 1/3-octave, which is the current most widely-used

frequency resolution. I shudder at the thought of N = 6.

Angular resolution is a different subject, and it has already been

pointed out that some devices require higher-than-5-deg angular

resolution to be fully characterized. Since each halving of the angular

resolution quadruples the number of test positions around the

loudspeaker, one needs to have a good reason for going higher than

5-deg. And as the angular resolution is increased to better define sharp

radiation lobes, the factors that produce changes to those lobes (i.e.

thermal gradients, mfg tolerances, component tolerances, etc.) become a

greater factor.

Increased resolution is always an easy sell, as is a bigger power

rating. But it is important to see the whole picture with regard to

both. Higher than 1/3-oct, 5-deg (for loudspeakers) is a special case

and should be handled as such.”

Lascia un commento