Incorrect Sound Pressure Level from FFT in Matlab
in [Matlab]
|
Prev: Need your help for matrix calculations
Next: how to concisely choose the first element in a vector modified by a conditional statement?
From: noisemetric stojiljkovic on 24 May 2010 10:51 i'm no expert in the topics discussed above and don't want to be crucified for trying to contribute, but this is a forum so what the hell...i've come across similar posts before, and vaguely remember a note about r.m.s accuracy calculations based on the fft algorithms. could it be that the same formula is used to calculate rms from matlab which performs the calculation over double sided spectra, while labview and similar apps first convert to single sided spectra ? this would go in line with the 5-6 dB differences seen in the plots. now duck and cover... Milan
From: Rune Allnor on 24 May 2010 10:56
On 24 Mai, 16:51, "noisemetric stojiljkovic" <noisemet...(a)gmail.com> wrote: > i'm no expert in the topics discussed above and don't want to be crucified for trying to contribute, but this is a forum so what the hell...i've come across similar posts before, and vaguely remember a note about r.m.s accuracy calculations based on the fft algorithms. could it be that the same formula is used to calculate rms from matlab which performs the calculation over double sided spectra, while labview and similar apps first convert to single sided spectra ? this would go in line with the 5-6 dB differences seen in the plots. now duck and cover... The factor 2'ish issue (the DC and nyquist components must be treated separately, as they only appear once in the spectrum) is one of a large number os issues that must be handled correctly to get the correct numbers in the end. The problem as stated in this tread was how to relate the computed numbers to sound pressure. You can't do that without the calibration factors. Rune |