| f is the frequency (usually in cycles per second or Hz). For digital
samples it's usually given as a fraction of the sampling frequency.
One-over-f (1/f) means that the POWER spectrum of the signal varies as the
reciprocal of the frequency. There's usually a lower limit to the 1/f
relationship or else the DC component of the signal is infinite. The most
common way to check your spectrum is to take an FFT of a reasonably long
section of your signal, multiply by the complex conjugate of your FFT (i.e.
FFT*FFT(cc)) to get your power spectrum, and plot the log of power versus
the log of frequency. If your plot is more-or-less a straight line with a
slope of -1, you have 1/f noise. If, instead of power, you plot the log of
magnitude of the FFT versus the log of the frequency, you should get a
slope of -1/2. If you need to know what any of the terms I have used mean,
write me personally.
At 4:46 AM -0500 7/1/98, pete moss wrote:
>i recently wrote a small program to generate one over f noise. i know
>that the values produced should have a spectrumclose to one over f. any
>ideas how to do this? what does one over f mean? what does f mean?
>thanks
-------------------------------------------------------------------------------
Erik Spjut (rhymes with cute) - Acting Director,The Center for Design Education
and/or Associate Professor of Engineering
Harvey Mudd College, Claremont, CA 91711 USA
Erik_Spjut@hmc.edu Ph & Voice mail (909) 607-3890 Fax (909) 621-8967
|