| T O P I C R E V I E W |
| Aike |
Posted - 01/30/2012 : 12:28:46 PM
I calculate power spectrum using FFT tool on OriginPro 8.1. I plot a graph: the x-axis indicates frequency (Hz) and the y-axis indicates Power Spectrum (???). I don't know the what is the unit of Power Spectrum. My data is about electron content. The unit of electron content is TECU. Can I refer the unit of Power Spectrum as TECU^2/Hz? |
| 2 L A T E S T R E P L I E S (Newest First) |
| Sam Fang |
Posted - 02/01/2012 : 9:19:38 PM
Origin's Power Spectrum supports three types of normalization. And its unit based on its expression:
MSA (Re^2+Im^2)/n^2 unit: TECU^2
SSA (Re^2+Im^2)/n unit: TECU^2
TISA dt*(Re^2+Im^2)/n unit: TECU^2/Hz
Note that Origin has shown units for FFT results if input data contain unit since Origin8.5.1 .
Sam
OriginLab Technical Services |
| Drbobshepherd |
Posted - 02/01/2012 : 10:56:19 AM
If your time-domain function is in units of TECU, and the time is in units of seconds, then your power spectral density (PSD) should be in units of TECU^2/Hz; however, if the FFT did not normalize the spectrum properly, your PSD units may be proportional to TECU^2/Hz.
If I were you, I would use Parseval's theorem to test the results. Compare the total integral of your PSD function to the total integral of the square of your time-domain function. If the ratio is close to unity, everything is OK; if not, then the ratio is your proportionality constant. |
