how to do fast fourier transform of a 50Hz motor

[Drizvi]

no sir the problem is not solved... i mistakenly clicked on the "accept as solution button" where as i wanted to click on REPLY.....

i've sent you the "actual VI".... please see to it and help me through,,

i'll be glad if u guide me through... thanks

[altenbach]

Please make the indicator named "data" contain typical data. (run the VI, stop, right-click the "data" indicator...make current values default). Save the VI and attach it once more.

Did you read any of the links I posted above? Your data is nyquist centered and he right side is only the mirror image of the left side. You also need to set the correct df based on dt of the original data.

The fourier transform you currenty doing does NOT give you a power spectrum, but a simple complex array containing the FFT. You are only graphing the real part, throwing away the imginary part. You are dropping valueble information left and right. First you are dropping all the waveform attributes, only retaining the naked data array, then you are only keeping the real part anfter the transform.

Whenever you see a red dot near a connector, you are wiring data that the target does no expect!

There are tools that accept waveforms, have a look at e.g. FFT power spectrum or similar.

(Also, when posting images, please crop to the interesting part. We don't need to see the frame and task bar, for example.)

[Drizvi]

i'm including typical data for the two different times i ran the program.

[Drizvi]

yes i did sir.... what do u mean when u say " You also need to set the correct df based on dt of the original data."

The fourier transform you currenty doing does NOT give you a power spectrum, but a simple complex array containing the FFT. You are only graphing the real part, throwing away the imginary part. You are dropping valueble information left and right. First you are dropping all the waveform attributes, only retaining the naked data array, then you are only keeping the real part anfter the transform.

sir i told u that i'm not very learned in the labview software... so can you please modify this thing that it is corrected??... or tell me step by step how to do it??...

thanks

[altenbach]

I don't think any of your problems have to do with programming or LabVIEW. You need to get a better theoretical understanding of these transforms.

Here's a simple example of a few ways. Note that your raw data lenght is not an integer multiple of your fundamental frequency, thus you get significant spectral leakage. The waveform tools actually include some windowing to minimize the problem. Still, I would recommend to slice the original data to an integer multiple of the main 50Hz wavelenght.

See how far you get. This is just a rough draft. You need to decide what you actually want.

[Drizvi]

I DID WHAT U SAID SIR.. i passed my incoming time signal through the fft power spectrum and psd and displayed the output on a graph.

The results were amazingly similar to what i wanted. i am sampling at a rate of 3200 and number of samples per channel are 1024.

the required result was supposed to be somewhat like

and what i'm getting is this

but still the results are not very fine as is required... what can i do to cater this???

i'm also attaching both the VI 's so that you can have a good look.

thanks

[Drizvi]

and also if u could please tell me that 

1) what is the significance of sampling rate and number of samples per channel??

2) how do the two variables effect our results

3) why For a Fast Fourier transform there must be a binary progression number of data points (2ⁿ)

thanks!

S M Danish Rizvi

[altenbach]

---

Drizvi wrote:

1) what is the significance of sampling rate and number of samples per channel??

2) how do the two variables effect our results

---

Why ask the question if you could simply simulate some data at different rates and different number of points and see for yourself? You'll automatically learn something. try it! 😄

---

@Drizvi wrote:

3) why For a Fast Fourier transform there must be a binary progression number of data points (2ⁿ)

---

This is an incorrect statement. True, some of the early computer implementations (radix-2 Cooley Tukey) had this limitations, but any size transform can be done very efficiently these days. Have a look at e.g. the wikipedia FFT entry.

[Drizvi]

my time signal data is being recorded by this 'data' reader and through this i am able to export the volt vs. time data to excell. i want to export the similar sort of data (power vs. frequency)  to excell after the signal has passed thrught the fft power spectrum psd. but i am unable to connect it with the data link!... can u help me here?

i'm also attaching the VI here!

[Matt_McLaughlin]

Hi Drizvi,

The 'data 2' indicator is not working because you are passing it a different data type than is expected. The output of the PS/PSD vi is a cluster of three elements rather than a waveform. If you want to display this output cluster on the front panel as an indicator, just delete the 'data 2' indicator and then right click the output from the PS/PSD vi and select Create>>Indicator. This will create an indicator of the proper data type.