From dB to dBA

[giuliosensi]

Hello there, this is my first, maybe sutpid, question.
How can I convert dB to dBA ? In particular I'm modelling a sound source with a Noise Rating NR-90 (dB) curve.
How can I get NR-90 (dBA) curve ?



Thanx !


Giulio

[Eric.Desart]

How can I get NR-90 (dBA) curve ?

You speak about NR curves. I assume you are European or work with ISO related standards or whatever?
Still, what is a NR-90 dB(A) curve?

What exactly do you have? Frequency related values? And what exactly do you want to obtain? Please give an example with the numbers you have.
Do you know what a dB(A) weighting is? Do you know the correction values?
Do you know the NR curves (mostly commonly used for ventilation purposes and the likes).
Do you know how to add dB values?

Some ideas here already: viewtopic.php?f=6&t=3300

And with these formulas you can weigh your curves as narrow banded as you like:
http://en.wikipedia.org/wiki/A-weighting

[Eric.Desart]

Accidentally I needed an old excel file of mine which is uploaded in the forum here (hadn't seen him myself for years..)

USNorms3.zip
download/file.php?id=1150

In this file you find all dB(A), dB(B) and dB(C) corrections for the center frequencies AND cutoff frequencies, for 1/1, 1/3, 1/6, 1/12 and 1/24 octave.
If you do not respond to my questions, it's rather hard to help you. But you know that of course.

[giuliosensi]

Thanx Eric and forgive me for my delay in answering you.
NOISE RATE curves are developed by ISO European equivalent to US Noise Criterium.
I know what dBA weighting is and everithing about theory, but in pratcice I had some troubles.

I'm making prediction for isolating a control room . I'm starting from an already made acoustic project .

In this project sound source is modeled following a NR 90 curve ( it's an hometheater).


Hz 125 250 500 1000 2000 4000 GLOBAL

NR 90 dB 100,30 95,70 92,50 90 87,80 86,20 117

NR 90 dBA 84,20 87,10 89,30 90 89 87,40 96


I would like to model my source as a NR 105 curve , I don't know how to find dBA values and I would like to find a general rules.

Thanx for the xls, I'm taking a look at it , maybe I will have to ask you some questions about it.

Thanx

[worthman]

Sound is made up of different frequency components, which create an overall level. However, the different ratios of frequency content affect the sound quite significantly. For convenience, signals are generally described in octave bands (each octave is a doubling of frequency) or ratios of such as third octaves.

The Noise rating or NR curve specifies a the maximum value at a particular octave band with respect to a set of reference curves - i.e. you could have a sound level which meets say NR35 at each octave band from 63Hz to 8000 Hz, but exceeds it by 5 dB at one octave band. This sound would then be rated as NR 40.

The reference curves are found in several standars including BS EN ISO 8233:1999. You can create them yourself using the following:

Octave band level = a + b*NR level

a and b for each octave band are as follows:
a={35.4, 22, 12, 4.2, 0, -3.5, -6.1, -8}
b={0.79, 0.87, 0.93, 0.98, 1, 1.015, 1.025, 1.03}

The octave band centre frequencies are 63, 125, 250, 500, 1000, 2000, 4000 and 8000 Hz.

By rearranging the NR equation, you can obtain an NR value for a certain octave band sound level. You then take the maximum of all the octave bands to get the overall NR level.

dB(Z) - originally dB(lin) or sometimes just dB, is the overall level of a sound and is the combination of all frequencies with a linear frequency weighting. Care should be taken when using dB, as the decibel is a general term and can apply to octave band levels as well as overall broadband levels. This can cover all frequencies or the more usual restricted bands of 63 - 8kHz, but you would get a lower level when using the restricted bands. This is why sometimes with a sound level meter, the level shown is not quite the same as when it is calculated from the octave bands, although this difference is usually small in the real world (0.1 dB ish), unless there was a very significant low frequency sub 50 HZ component.

dB(A) is a weighted broadband level which approximates the ear's sensitivity to different frequencies. The weightings are as below:
{-26.2, -16.1, -8.6, -3.2, 0, 1.2, 1, -1.1} (from 63 to 8k)

There are also dB(B) and dB(C). B is never really used and C is sometimes used for aircraft noise.

To calculate a dB(A), weihgt each octave band level accordingly and then logarithmically add each band together.

LA=10*log10(sum(10^((Ln-Wn)/10))) where n=each octave band, L = level and W = weighting.

This weighting does extend to 32 Hz, but is rarely used in normal applications.

As a result, you can get lots of different octave band combinations which will achieve the same dB(A) level, but would have significantly different frequency contents. Therefore, there is no definitive relationship between dB(A) and NR. This can only be derived from the calculations above.

However, there is an approximation for mechanical services (air conditioning) noise whereby NR + 6 ~ dB(A).

Hope this helps.

[krasmuzik]

You have that wrong - dB(D) is sometimes used for aircraft noise. dB(C) is the recommended weighting for measuring theater and studio levels as it matches ear response at higher mastering and playback levels. dB(A) is used in hearing protection as it matches ear response at lower levels - despite also being applied to higher levels. dB(B) is used for moderate levels. Most SPL meters just do A/C curves, they also do fast/slow exponential time weighting - which is another complication in computing dB. Standard ISO descripters are LWtau (Wtau are subscript letters for frequency weighting and time weighting) i.e. LAS or LCF etc.

[Eric.Desart]

@ worthman,

Welcome to the forum. Glad to see a pro joining us.

@ krasmusic,
Can't you say the same things more gentle?
How the heck do you know if C isn't used in the UK for aircraft noise?

You say that D is sometimes used instead. Where, by who, with what?
I'm not going to check exact standards now, but the D-weighting disappeared ages ago from the standards. The only reason you find this D still in lots of graphs is that people like to copy info from one another because it looks learned (and hardly bother what they copy or where it comes from or stands for).
90% has no idea idea it disappeared ages (decades) ago and isn't representative for modern planes and related traffic anymore (which is why it disappeared from the standards VERY long ago, I have to search real antique data just to find it).

Also dB(B) disappeared from modern standards (which I pitty, since it can be a useful weighting), which is why modern SLMs, in accordance with current standards mostly only show dB(A), and dB(C) + the linear ones..
And C is maybe the standard for you and in your surroundings for theater and studio levels, but where does ISO tells that? It's better to use C than A, that's evident (not only for the level relationship, but because of the high low frequent content for studio purposes).
ISO 717-1 gives a weighting for traffic and Music.
In Belgium and the Netherlands we give specific weightings for different types of music. And a C-weighting can also be useful at low levels for music isolation single number ratings.

What I mean is: please doing science, be a bit more more careful with your absolutes.

@ worthman,

"The reference curves are found in several standards including BS EN ISO 8233:1999."

Just a question: Is your Standard code correct or is that BS 8233:1999
As far as I know the NR curves were only shown in ISO/R 1996:1971 (withdrawn 1982)
A lot of countries toke these curves over in own National norms (back then the ISO number wasn't preserved). But the NR curves got a second live outside the ISO standards.
Your Standard code tells that this is a British, European and ISO standard, which can't be true: ISO 8233 has no relation with sound.
Hence I guess it's a typical British Standard BS 8233:1999, without the EN ISO.

[worthman]

Eric,

Thanks for the welcome.

Apologies on the BS standard. You are correct - it is just BS8233. I use so many I forget which are BS only.

I couldn't find a copy of the original ISO 1996 curves, and I've seen the NR coefficients copied incorrectly in some text books, hence the reference to BS8233.

Krasmusic -

With regards to the use of dB(C), in the UK, it generally gets used as part of our Noise at Work assessments, where there is a dB(C) peak limit before hearing protection is required, coupled with an 8 hour LAeq measure. Other than that, I only tend to use dB(A) on a day to day basis - even our airport noise contours are in dB(A). There were some studies undertaken a while ago which show that dB(A) is the best correlation to disturbance and nuisance, although I am aware of the low frequency arguments that come up from time to time - although I suspect this is now moving outside of the original topic for this discussion. I am aware of dB(B) and dB(D), but in the UK these are not generally used, although I don't have absolute knowledge of all applications.

[giuliosensi]

I got to thank you worthman for your very precise answer, I got things much clearer now !

Big thanx

Giulio

[hansdubarry]

Hi all,
has anyone ever plot NR curves in excel. I ve tried it myself but am stuck with. the problem is that the x-axis is 'non-lineared' scaled. I trying to set the values 62.5, 125,250,500,1000, etc on the horizontal axis given the values in the y-plane. the problem in excel is that it would always try to show values that are equally spaced on the horizontal space, unlike the Octave Band Center Frequencies.
Can any one help?

Thanks
Hans

[entdh]

Hi Hans
I use Excel 2003 and have had a little success plotting NR curves using the data from BS8233 as posted above
the file is attached (Ignore the data on sheet 2 and 3 this is a work in progress)
you can enter your own spectrum of values and it will plot the NR curve, I should be able to calculate the NR directly but need a little more time yet.modified version now does this
you need to use a scatter diagram to get a logarithmic scale on the x axis, but it is still difficult to scale correctly
I believe Excel 2007 is much improved??

regards
Dave H

[bert stoltenborg]

Eric is always making fun of me when I use Excel 3 for this stuff, but at least it works in a logical way

[entdh]

Eric can make Excel sing and dance.....

I have been trying to show the decisive band for NR by using LOOKUP function but I have given up for now

regards
Dave H

[entdh]

It is a poor excel spreadsheet but at least it now works
(I used VLOOKUP instead)

Dave H

[joelpj]

Here's another Excel spreadsheet for dB to dB(A) conversion
http://forum.studiotips.com/viewtopic.php?f=1&t=3806#p51411

[Laucoustic]

Hello,

In the post of Worthman of 14 augustus 2009 the coëfficiënts for composing the complete NR table are given. This is very useful for me. However, dies somebody know the coefficients a and be more detailed (more decimals)?

Octave band level = a + b*NR level

a and b for each octave band are as follows:
a={35.4, 22, 12, 4.2, 0, -3.5, -6.1, -8}
b={0.79, 0.87, 0.93, 0.98, 1, 1.015, 1.025, 1.03}