topic Re: Implementation of Fast Fourier Transform in different file formats in LabVIEW

Importing CSV data for FFT analysis

johnBURL — Wed, 11 May 2016 18:07:35 GMT

Hi guys,

Sorry if this is a dumb question, but I searched for anything relevant and I couldn't find a problem that was close enough to what I needed. Part of the problem is that I am not that experienced in Labview (I am more Matlab, my coworker usually works on Labview but he is in Europe right now).

I am attempting to extract data from a CSV file that contains waveform data captured from an oscilloscope. I was hoping to 1) extract the data from the file in Labview and then 2) perform an FFT analysis on it.

Here is what I have gathered I should do so far:

1) Use the "Read from Spreadsheet" function to import the CSV file. All I've worked out so far is set the delimiter and add a control for the file select. Not sure where to go from here. And would the output of this function result in an array or a plot?

2) Using the information from the Read from Spreadsheet function, I would then use an FFT function (presumably under the Waveform Function menu) which would then take the data and output it in some way (graph, etc).

I am combing through tutorials and other cases similar to mine, but any help would be appreciated. Thanks.

I included a CSV file in case it was needed.

Re: Importing CSV data for FFT analysis

B_Strange — Wed, 11 May 2016 19:25:32 GMT

Is this sort of what you're looking for?

Re: Importing CSV data for FFT analysis

johnBURL — Wed, 11 May 2016 19:33:45 GMT

That does look promising...so from how I interpret it, it looks like you are using a function that takes the CSV (or info inside the CSV) that posts it to an array, and then that array is fed into some soft of function which then results in a graph? Would you be able to explain further so that I could recreate it in my Labview? Thanks for the reply.

Re: Importing CSV data for FFT analysis

B_Strange — Wed, 11 May 2016 19:39:45 GMT

If you've got LabVIEW 2013 or newer, simply save the image I posted (say, to the desktop) and then drag it into the block diagram. The code will automatically be created.

Re: Importing CSV data for FFT analysis

johnBURL — Wed, 11 May 2016 20:07:21 GMT

D'oh, thanks. That does make it much easier.

I've enabled my Context Help button, and I see that you put the data from the CSV file into the box that is wired to the Spreadsheet String node of the "Spreadsheet String to Array" function. Would I be able to instead insert a control in there, so that I could select the CSV file? Or how would I select a different file? The reason why I am asking is that I think the Identifying info on the CSV file is messing up the array and possibly the FFT. I am really only interested in columns D and E on that file.

From there I see that the function outputs to an array, which also leads to the Index Array function. And then an FFT function. But I'm a bit lost on what exactly these are doing.

Most examples I've seen use the Read from Spreadsheet function, was there a reason why you decided not to?

Re: Importing CSV data for FFT analysis

B_Strange — Wed, 11 May 2016 20:36:43 GMT

I haven't messed with Read From Spreadsheet File.vi much before. After some quick testing, I can only get the first column of data out of it which is all zeros. It seems like you should be able to get columns D & E out without too much effort though.

In the code below, I'm pulling out column E (LabVIEW arrays are zero-indexed so a column index of 4 actually means the "fifth" column). I've reworked the code to include file I/O and also added some comments which will hopefully clear some of the hazy stuff up. If you're curious about what data is being passed through on any wires, right click and select either "Probe" or "Create > Indicator". They both do what amounts to basically the same thing (show the data on the wire) but the probe is a little less invasive.

Re: Importing CSV data for FFT analysis

Eric1977 — Wed, 11 May 2016 20:55:52 GMT

Just use the File I/O primitives Dave described above and feed the output to the spreadsheet string to array function. Also, there is nothing stopping you from manipulating that array once you create it. I'd use the Array Subset instead of Index Array to get the last 2 rows.

Re: Importing CSV data for FFT analysis

johnBURL — Wed, 11 May 2016 20:54:11 GMT

Thanks, Dave. Let me mess around a little bit with what you've made already, I don't want to bother you any more than I have to. I appreciate all of the help so far.

Re: Importing CSV data for FFT analysis

johnBURL — Wed, 11 May 2016 21:35:08 GMT

Thank you also, Eric. Just for curiousity's sake, I also ran your VI, and ended up getting quite a different graph than Dave's. I think the only difference was the Array Subset. I'm going to keep messing around with both, maybe clean up the CSV files (I am not sure if just column D/E are being calculated).

Re: Importing CSV data for FFT analysis

B_Strange — Thu, 12 May 2016 10:43:05 GMT

@johnBURL wrote:
quite a different graph

I don't have access to LabVIEW on this machine but I think with the "length" parameter of Array Subset unwired, you get all remaining columns (which in this case is two data columns and a column of zeros). Try wiring a 2 constant.

Re: Importing CSV data for FFT analysis

johnBURL — Thu, 12 May 2016 18:22:30 GMT

Not sure if that made a difference, really. If I can, I am going to upload my VI and the simpler dataset I was using for testing. I made a few changes but I think my main problem now, is that the FFT chart is using and X and Y axis labeled Time and Amplitude. The Amplitude label is fine, but I was hoping to see the frequency of the signal as the X-Axis.Not sure how to change that label or what info I need to pull from the CSV file.

To make it easier to show, I used a dataset that contained a 220khz signal being generated. So I would expect to see on the FFT a large spike for the 220khz and any smaller spikes for harmonics that may be present.

Thanks again, guys. I feel like with your help, I'm really close to accomplising my task.

Eric: I appreciate your help but I feel a bit lost with your function because the graph is not at all what I was expecting! Lol. Thanks again, brother.

Re: Importing CSV data for FFT analysis

RavensFan — Thu, 12 May 2016 18:31:53 GMT

Just double click on the word "time" and edit it. You can also change it in the properties dialog fo that graph.

Re: Importing CSV data for FFT analysis

johnBURL — Thu, 12 May 2016 18:50:14 GMT

Doesn't that just change the label, though? What I meant was, the FFT was defaulting "time" (or in this case, the record length, which is 10k) as the x-axis.The FFT should be an amplitude v frequency graph, time/record length shouldn't be a factor, I believe. Right now the FFT is showing me amplitude v time, so I see spikes of the amplitude but I cannot tell what frequency they are.

Re: Importing CSV data for FFT analysis

RavensFan — Thu, 12 May 2016 20:16:37 GMT

FFT gives you amplitude vs. frequency.

The axis on the graph says "time", but that is just a label.

But that FFT function puts out an array of frequency bins, you just need to scale those bins. The dF of that graph is the frequency of your incoming data divided by the number of data points.

Here is a subVI I put together to graph an FFT for me from the basic FFT function, so I didn't have to deal with the Express VI's.

Re: Implementation of Fast Fourier Transform in different file formats

sowji25 — Wed, 09 May 2018 04:58:13 GMT

Hi all,

I want to implement the Fast Fourier Transform to Technical data management streaming, in this file format one is time stamp data and other are sensors data, I create an array to select the sensor with respect to time data, when i run the VI the sensor data waveform spectrum shows correctly but sensor 2 data waveform spectrum is not showing, please the check other sensors data also, i want the frequency spectrum peaks at 2400 Hz and 3250 Hz. once check the code and please correct the VI in correct format

I also implement the fast Fourier transform to different file formats also but it not showing the correct waveform spectrum , please once check the code whether it is right or not