Converting sound pressure levels (dB) to psd (g^2/Hz)

[westmoors]

I have data supplied as broadband sound pressure levels in dB against frequency. I need to convert this into a psd (g^2/Hz) versus frequency. Anyone know how to do this?

 

[GregLocock]

At first sight you need to know something else about the system, as you are trying to get acceleration from pressure data.

Cheers

Greg Locock

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[CESSNA1]

WESTMOORS: Db is a ratio or relative scale. There should be a reference DB to start with. It may be that the information you have is presented in "one-third octave" format. If so, this format consists basically of averages between two frequencies. I know of no way to convert back. If someone does, please let me know. I suggest you contact the source of the data and ask if they can give you the PSD or, preferably, the raw data so that you can work with it. Also ask for the "highest frequency of interest", The "filter frequency", The "sampling rate", and if the accelerometers were calibrated in "G peak or G rms". Then you can do the conversion

Regards
Dave
"An engineer is someone who can make a decision without having all the information"

 

[GregLocock]

So you are claiming that if I gave you a time history of sound pressure level you could work out a spectrum in g^2/Hz? Astonishing. What's the conversion factor for sound pressure to acceleration, my engineering reference books seem to have left it out?





Cheers

Greg Locock

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[CESSNA1]

GREGLOCOCK: Not sound pressure levels but an acceleration time history. The one third octave analysis is done on the acceleration time history. The PSD is done on the acceleration time history also. The time history has to be in G peak since when the PSD is performed it convets it to Grms^2/hz.

Regards
Dave

 

[GregLocock]

Now reread the original post.



Cheers

Greg Locock

Please see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[CESSNA1]

GREGLOCOCK: Sorry

 

[tomirvine]

There are some semi-empirical methods for converting a sound pressure level (dB) into a random vibration acceleration PSD (G^2/Hz).

These methods are for structures such as plates and cylinders.

The Franken method is used for cylinders.

In addition, Steinberg gives a simplified example in text.

This type of analysis is called "vibroacoustics."

Tom Irvine

http://www.vibrationdata.com/fatigue.htm

http://www.vibrationdata.com/vibroacoustics.htm

 

[djf123]

LMS has a module for Sysnoise called Random Vibro Acoustics. If you have an acoustic pressure power spectrum you can determine surface vibrations through a boundary element technique. It requires you to have a model of the system you are testing. You can think of this in simple terms of knowing the acoustic transfer vector (acoustic pressure / surface velocity) and knowing the acoustic field (your SPL spectrum) you can then determine the surface velocities and therefore, the surface acceleration. This could then be scaled into G^2/Hz.

Reference this link on their website:

http://www.lmsintl.com/?sitenavid=8CAB4136-408C-4C58-BC2A-2B5BCE6CA977

The SYSNOISE software has migrated into Virtual.Lab Acoustics.

 

[Rob45]

Vibroacoustics works just fine; I've been doing some in-cab floor vibration testing of late, and decided to also measure SPL.

Lo and behold, all the low-frequency floor vibration modes were equally well-represented by the acoustic data.

Will wonders never cease!