Help defining wide/narrow directivity/disperison.

[MZKM]

So, if you guys don't know, I have a speaker selector Google Sheet that you can use and narrow down the list with multiple filters (both scores, price range, measured sensitivity, physical dimensions, enclosure type, etc.).

I am wanting to add dispersion categorization, but then of course we have to calculate the dispersion and then categorize it.

I saw that @hardisj posted that has made a Sheet of speakers he has measured, and he is showing the +/- degrees of directivity at 1kHz & 10kHz for both horizontal & vertical. That's fine, but wouldn't make it simple for filtering, especially for people who are less familiar with speaker measurements.

Calculating dispersion

My first thought was to normalize 90° H to the listening window (unlike my directivity graphs, which normalize to on-axis), and use the slope of the resulting curve.

My second thought, and what I currently am doing is because it is easier, is simply taking the slope [average] of the Sound Power DI curve.

As for the frequency range used, I currently have it at 500Hz-10kHz.

Here are some results:
Canon S-50: 1.8
Revel F208: 5.4
Buchardt S400: 6.1
KEF R3: 6.8

The one issue is the Magnepan LRS, which gets a 5.9, as even in the deep bass the DI is ~4. I could just override it with "dipole" though.

Defining dispersion

Using the calculations above, here is how I currently have them defined:
(-inf,2]: "Omni"
(2,4]:"Very Wide"
(4,5]:"Wide"
(5,6],"Average"
[6,7):"Narrow"
(7,inf): "Very Narrow"

Since most speakers will likely fall in the 5-7 range, maybe I should go by 0.5 increments.



Would love to hear thoughts.

 

[richard12511]

So, if you guys don't know, I have a speaker selector Google Sheet that you can use and narrow down the list with multiple filters (both scores, price range, measured sensitivity, physical dimensions, enclosure type, etc.).

I am wanting to add dispersion categorization, but then of course we have to calculate the dispersion and then categorize it.

I saw that @hardisj posted that has made a Sheet of speakers he has measured, and he is showing the +/- degrees of directivity at 1kHz & 10kHz for both horizontal & vertical. That's fine, but wouldn't make it simple for filtering, especially for people who are less familiar with speaker measurements.

Calculating dispersion

My first thought was to normalize 90° H to the listening window (unlike my directivity graphs, which normalize to on-axis), and use the slope of the resulting curve.

My second thought, and what I currently am doing is because it is easier, is simply taking the slope of the Sound Power DI curve.

As for the frequency range the slope is calculate on, I currently have it at 500Hz-10kHz.

Here are some results:
Canon S-50: 1.8
Revel F208: 5.4
Buchardt S400: 6.1
KEF R3: 6.8

The one issue is the Magnepan LRS, which gets a 5.9, as even in the deep bass the DI is ~4. I could just override it with "dipole" though.

Defining dispersion

Using the calculations above, here is how I currently have them defined:
(-inf,2]: "Omni"
(2,3]:"Very Wide"
(3,5]:"Wide"
(5,6],"Average"
[6,7):"Narrow"
(7,inf): "Very Narrow"

Since most speakers will likely fall in the 5-7 range, maybe I should go by 0.5 increments.



Would love to hear thoughts.

Is there a way to calculate separate slopes for horizontal and vertical? I tend to mostly look at horizontal width when assessing a speaker. What would the BMR be here? I would consider the BMR as very wide, but it's vertical is quite narrow.

500-10,000Hz seems logical. I wonder if psychoacoustics might suggest a smaller range? I have no idea.

 

[MZKM]

Is there a way to calculate separate slopes for horizontal and vertical? I tend to mostly look at horizontal width when assessing a speaker. What would the BMR be here? I would consider the BMR as very wide, but it's vertical is quite narrow.

500-10,000Hz seems logical. I wonder if psychoacoustics might suggest a smaller range? I have no idea.

Measurements:
https://www.audiosciencereview.com/...armonic-bmr-speaker-review.14781/#post-460044
For the way I’m calculating, it would be for sure Wide looking at the DI, it doesn’t even pass 5 until 3kHz.

As for separate horizontal and vertical, that depends how you calculate it. One complex way would be to find at what degree at each angle (positive and negative) that the SPL decreases a set amount (3dB, 6dB, etc.) from the listneing window SPL at that frequency, then average that together and then look at the data for a handful of speakers and decide what your thresholds for each categorization would be.

 

[thewas]

My second thought, and what I currently am doing is because it is easier, is simply taking the slope of the Sound Power DI curve.

But why taking the slope instead actually the (for example average) value of the DI which shows exactly how wide or narrow the dispersion is?
By taking the slope for example at a larger PA loudspeaker with constant high directivity starting quite early the slope from 500-10000 Hz would be low although its narrow dispersion loudspeaker.

 

[MZKM]

But why taking the slope instead actually the (for example average) value of the DI which shows exactly how wide or narrow the dispersion is?

The DI value is per frequency value.

 

[thewas]

The DI value is per frequency measurement.

??
As I wrote above you could take the average (or RMS etc) and the slope doesn't work well if you have a bigger constant directivity loudspeaker.

 

[MZKM]

??
As I wrote above you could take the average (or RMS etc) and the slope doesn't work well if you have a bigger constant directivity loudspeaker.

You know what, I’m an idiot. I currently am taking the average; I was taking the slope before and switched to the average and somehow forgot which I chose to keep.

 

[thewas]

You know what, I’m an idiot. I currently am taking the average; I was taking the slope before and switched to the average and somehow forgot which I chose to keep.

All fine then, we agree and you are of course no idiot, we are all just humans sitting to long in front of the monitor.

 

[vavan]

You know what, I’m

One of the most valuable members here, at least in my book