Incorrect formula for octave frequencies in docs.

[Daverz]

This is a comment on the document

Octave Analysis Standards (Sound and Vibration Measurement Suite)

at

http://zone.ni.com/reference/en-XX/help/372416A-01/svtconcepts/ansi_iec_stands/

(also http://www.ni.com/white-paper/2975/en#toc3 and several other places.)

The formula "for 1/N octave filters when N is even" is incorrect. It should be

In other words, there's a missing factor of 2 before the i.  The mistake is repeated in the PDF manual for the Sound and Vibration Measurement Suite.

Also the comment under that

"i is an integer when i = 0, f0 = 1 kHz, which is the reference frequency"

only applies for 1/1 or 1/3 octave bands. For even N as defined above, 1kHz is the lower edge of the i=0 band, not the center.

I would suggest, however, that the formula

has less potential for confusion than the one above.  

Here's how ANSI defines the center frequencies for even N:

Their b is the reciprocal of yours, G=2, and f_r is 1000 Hz.  The -59 just gives a different numbering for the bands.

[MaggieAggie]

Hello Daverz, 

Thank you for bringing this to our attention. I have looked through NI documentation, and the equation without the factor of 2 is consistently used. I will look into this further. 

[Charlie_P]

Hello Daverz,

Thanks for bringing this typo to our attention.  I confirmed the code that executes the algorithm is correct so your results have not been compromised; it is solely a problem with the documentation.  Corrective Action Request (CAR) 388126 has been filed to address the inaccuracy in the affected documents.