| T O P I C R E V I E W |
| amartya_iitd |
Posted - 10/15/2018 : 01:28:19 AM
Origin Ver. and Service Release (Select Help-->About Origin):
Operating System:9.0.0 SR2
Hello: I am doing the FFT analysis of a time domain signal having 1401 evenly spaced data points, however, Origin gives me the FFT analysis with only 701 data points, which means it is under-sampling! I have tried both the FFT Gadget and signal analysis and have tried changing the sampling interval, but to no avail! Can somebody please guide me as to where I am going wrong?
Thanks, Amartya |
| 5 L A T E S T R E P L I E S (Newest First) |
| Drbobshepherd |
Posted - 10/16/2018 : 3:13:04 PM
You have 1401 time-domain samples, which means your tranform can only have at most 1401 freq-domain points. Depending on how you perform the FT, half your results may be negative freqs which Origin considers redundant and discards. Or maybe half are imaginary, and you are only looking at the real components. 1401 real time data points cannot transform to 1401 real and 1401 imaginary spectral data points; this violates information theory. You could try padding the time data with 1400 zeroes to get your desired spectral resolution.
DrBob |
| easwar |
Posted - 10/16/2018 : 11:48:36 AM
Hi Amartya,
In very old versions of Origin, we used to pad the data with zeroes or truncate data, to make the data size be a power of 2.
With the FFTW library (http://fftw.org/) that we incorporated into Origin many versions ago, such padding/truncation is no longer necessary, so we do not do that any more.
You can pad your data with zeroes to extend it to whatever size/power of 2 that you desire, prior to computing the FFT results. Keep in mind, you may then want to apply suitable window and window correction settings. The default window is Rectangle.
Hope this helps.
Easwar
OriginLab |
| amartya_iitd |
Posted - 10/16/2018 : 07:00:45 AM
Thanks for your comments, however, my time domain data has 1401 data points, and the FFT is showing only 701 data points. Isn't it supposed to show at least 2048 points, 2 raised to the power n, depending on the original time domain data point number?
Please help. Thanks! |
| yuki_wu |
Posted - 10/16/2018 : 01:37:13 AM
Hi,
Select Options tab and select Two-sided in Spectrum Type drop-down.
Regards,
Yuki
OriginLab
|
| Castiel |
Posted - 10/15/2018 : 9:39:19 PM
quote:
Originally posted by amartya_iitd
Origin Ver. and Service Release (Select Help-->About Origin):
Operating System:9.0.0 SR2
Hello: I am doing the FFT analysis of a time domain signal having 1401 evenly spaced data points, however, Origin gives me the FFT analysis with only 701 data points, which means it is under-sampling! I have tried both the FFT Gadget and signal analysis and have tried changing the sampling interval, but to no avail! Can somebody please guide me as to where I am going wrong?
Thanks, Amartya
I suppose you were expecting a two-sided spectrum, an option of FFT X-Function.
#####
#### _\_ ________
##=-[.].]| \
#( _\ | |------|
# __| | ||||||||
\ _/ | ||||||||
.--'--'-. | | ____ |
/ __ `|__|[o__o]|
_(____nm_______ /____\____
|
|
|
