Question regarding y-axis units in FFT

[lennybogzy]

Hi all,

Havent turned to this forum in quite some time but hoping you might be able to explain to me my y-axis in the audio FFT I have below. How should I understand the units of dB/20u Pa? Ideally I want to convert this into dBa or something, I'm just not sure how to get a "feel" for this.

Could someone explain this to me like I'm 5?

Thanks, smart people.

FFT is https://www.physicsforums.com/showthread.php?t=703574 in case anyone cares

 

[davidm1732]

The units are acoustic decibels (dB). The reference value for the dB ratio shown is 20 micropascals..the standard reference for sound pressure level. When you ask the analyser display the units in the dB scale, you usually have to define the reference level it uses to perform the dB calculation.

So basically it's telling you what the sound pressure level is in dB, most likely using a linear scale.

I'm not a B&K Pulse user, but it shouldn't be difficult for you to select an A weighting option. You can possibly do this by applying a weighting to the front end, so the signal is weighted before you do any more processing on it, or you should be able to apply weighting to the already measured signal you're showing with a post processing feature.

I'm assuming you're familiar with the A weighting scale (vs a linear scale).

B&K has excellent tech support...see if you have on-line help available with the pulse system you're using, or call B&K directly.

Do you have a microphone calibrator? That's also an easy way to check if your measurement is reasonable (not to mention recommended practice before taking measurements).

 

[lennybogzy]

Hey, thanks a lot Dave. This is exactly the info I was looking for. Acoustic testing is brand new for me so I really appreciate this.

Unfortunately this software is new to everyone and all the data I have is in screenshot mode only due to time constraints.

A few questions, since you're clearly knowledgeable about this:

1. Anything above 0dB is "hearable" and anything below is not?

2. Assuming I'm interested in one or a few particular frequencies, I should be able to take those peaks and "plug in" the freq. and peak to get the gain and calculate dBA? (If I reference a curve such as http://www.sengpielaudio.com/WeightingsA-And-C.gif)

In this case 3.2KHz is the area of concern. It looks like the peak should actually be amplified if I convert into dBA. Correct?

Anyway, again, thanks a lot. Much appreciated!

 

[davidm1732]

Hi,

1. More or less... 20uPa is generally accepted as the threshold of hearing, so given that, 0 dB would be the lower limit for hearing...but it varies from person to person... age, exposure, etc... depends how picky you need to be with the application of that concept
2. I think that could work.. you have to be careful when interpreting very peaky type signals however... I don't normally convert between narrowband and 1/3 octave measurements, but you might see a difference using those two measurement options for the same source
3, Yes, it would, however it's only a few dB...so not a huge difference...again, how precise do you need to be... what's the application etc? Just looking at the curve, you see the real effect of the A weighting is at the lower frequencies... where you can see 10 dB differences or more.. so that's really significant. It's difficult for humans to identify a difference in perceived loudness between two narrowband sources that vary by less than 2 dB or so.
Plus, you may have other factors that muddy the waters... test to test repeatibility... how much variation do you see with that? Mic positioning... move the mic and see how your measured values change... directionality...is the source directional at all? As you move up in the frequency range, not all acoustic sources propagate sound evenly (like the waves from a pebble dropped into a pond)...