Comparing acoustic data (units question)

[castwerks]

I wanted to compare noise levels for a couple different data sets I have seen presented in papers. Each one of the papers seems to plot their noise level spectrum levels using different units:

dB re 1 microPa^2/Hz @ 1m
dB re 1 microPa/sqrt(Hz) @ 1m
dB re 1 microPa/sqrt(Hz) RMS @ 1m
dB re 1 microPa/Hz

How do I convert values presented in PSD (microPa^2/Hz) or spectrum level (microPa/Hz) into source level units (microPa/sqrt(Hz)) so I can make an 'apples to apples comparison?

Thanks

 

[GregLocock]

There's a fair old muddle

I tried to write a quick answer yesterday and it was worse than useless.

I think 2 and 3 are probably identical.

4 may or may not be comparable in any sense with the others.

case 1 is in power like terms, and will increase as 10*log(p1/p0), whereas the others are in volt like terms, and will increase as 20*log(p1/p0)

My basic approach would be to turn them back into Pa^2, and then try and sort the frequency resolution effect out.



Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[MikeyP]

Not really an answer but another question...

Why on earth would someone use 1 uPa instead of 20 uPa as a reference (unless it was in some sort of low pressure environment?

M

--
Dr Michael F Platten

 

[castwerks]

Mike
1 microPa is typically referenced for acoustic measurements in water.

Greg
Can you explain what you mean by 'frequency resolution effect'. I thought that the /hz and /sqrt(hz) notation was used to normalize the data so band width information isn't needed.

 

[SomptingGuy]

20 uPa is typically though of as the threshold of human hearing. If human hearing's not involved, any other reference seems valid, so long as it is quoted with the dB measuement.

- Steve

 

[GregLocock]

That's why I get nervous when people start quoting psd type data. Every calibrator I've used has been a pure sine wave, and most problems I've worked on have been pure tones. I don't want the height of the peak jumping around a lot just because I change frequency resolution. Fair enough, to calibrate we often use the power in a wider band than 1 bin. I'm not saying my usage is right, it is just why I am not going to give you a cookbook answer, I'd have to go through it carefully from first principles.

The first principls of course is that sigma v^2 over a given frequency range should be the same, whatever units you are using.



Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?