[Kst] calculating the Amplitude Spectrum

[Barth Netterfield]

On January 31, 2014 06:55:23 PM Barth Netterfield wrote:

> You had "Interleaved Average" checked in the Power Spectrum Dialog.  This is
> useful if your input spectrum is random phase non-white gaussian noise and
> you want to determine the underlying power spectrum, but not correct in
> your case, where you want to determine the amplitude of a coherent line.
> 
> Un-checking this option gives a line of amplitude 0.999911705184 at
> 34.5011749626 Hz  (with the remainder of the power spread into nearby bins).

(also I unchecked the window option - otherwise the power is spread more).

> Sqrt(2) vs 1 is related to the definition of "one sided" or "two sided"
> power spectrum or some such thing like that.
> 
> cbn
> 
> On January 31, 2014 02:07:11 PM Chris Stoughton wrote:
> > I'm having trouble understanding the normalization of power spectra.
> > 
> > I generate a sine signal with amplitude 1 Volt, and use kst2 to plot the
> > amplitude spectrum.
> > I expect to get a peak of sqrt(2)*1 Volts in the frequency bin that
> > includes the sine wave.
> > 
> > Instead, it seems to have a value of 0.037, instead of 1.4.
> > 
> > I have attached the input data file PowerSpectrum.txt, the kst file
> > PowerSpectrum.kst, and a PDF of the plot as it appears when I run kst2
> > PowerSpectrum.kst/
> > 
> > Here is the version I am running.
> > 
> > Kst 2.0.x Revision 662928007df2521469b471a899017ed89a735d9b
> > 
> > 
> > Thanks for your help.

-- 
C. Barth Netterfield
416-845-0946