The Origin Forum
File Exchange
Try Origin for Free
The Origin Forum
Home | Profile | Register | Active Topics | Members | Search | FAQ | Send File to Tech support
Username:
Password:
Save Password
Forgot your Password? | Admin Options

 All Forums
 Origin Forum
 Origin Forum
 FFT smoothing
 New Topic  Reply to Topic
 Printer Friendly
Author Previous Topic Topic Next Topic Lock Topic Edit Topic Delete Topic New Topic Reply to Topic

schneide

Germany
45 Posts

Posted - 10/09/2012 :  12:10:37 PM  Show Profile  Edit Topic  Reply with Quote  View user's IP address  Delete Topic
origin ver. 7.5 / 8.5 /8.6sr3
operating system: win 7

i am using the smooth fft 7 points function. there are different results with each of the above origin versions. we started with origin 7.5 and like the results better than with the o8x versions (smoother beginning&end). so we kept using o7.5 for that purpose.
now our data numbers exceed 127 or 255 pts respectively and adding just this one extra data point alters the shape of the smoothed curve grossly. (it roughly corresponds to reducing the window from 7 to 3 or 4 points.

i find in the o8.6 help the window function used in o7.5 with parameters p,n,t unfortunately none of these are explained. how does it depend on the data point number?

can anybody explain the o7.5 window function?
how can i reproduce the o 7.5 behaviour for less than 128 data points with larger data numbers?


gi



Giselher

Kathy_Wang

China
159 Posts

Posted - 10/11/2012 :  05:27:46 AM  Show Profile  Edit Reply  Reply with Quote  View user's IP address  Delete Reply
Hi,

Regarding the o7.5 window function:

First, sorry there is a typo in the window function in 8.6 help file, the third equation should be 1-4p^2/(n^2)*(n-t)^2. It is used as a low pass parabolic filter.
In this equation, t is the index, p is the Points of Window (which can be set from the dialog) and n is the number of data points in your sample data.

We are not sure about what you actually meant by the second question, normally if you want to reproduce the FFT smoothing result of o7.5, you may run the following script to do the smoothing in 8.6:
smooth iy:=<active> method:=3 npts:=7 pad:=1 filter:=1 baseline:=0;


If this doesn't help to get what you want, it would be best if you could send us a sample of your data to us and explain the detailed steps you did in 7.5, so that we may try to reproduce. To send file to us, please press the Send Files to Tech Support button in the upper-right corner of the forum page and follow the instructions there.


Kathy
Originlab

Edited by - Kathy_Wang on 10/11/2012 05:28:25 AM
Go to Top of Page

greg

USA
1378 Posts

Posted - 10/11/2012 :  10:53:58 AM  Show Profile  Edit Reply  Reply with Quote  View user's IP address  Delete Reply
You can still use the smoothing method of 7.5 in 8.6, but that method does not support auto or manual recalculation and it still has the problem where changing the number of points can drastically alter the result.

To see an example, add a new column to your Book1, Sheet1 worksheet named Smooth.
Delete rows after 127
Run this script:
// BEGIN SCRIPT
curve.x$=Book1_E;
curve.data$=Book1_B;
curve.result$ = Book1_Smooth;
curve.smoothpts = 7;
curve.smoothleftpts = 3;
curve.smoothrightpts = 3;
curve.polydeg = 2;
curve.fftsmooth();

// END SCRIPT
You will get the same smoothing you got in 7.5. If you undo the deletion of the rows and run the script again, you will see the 'points' problem 7.5 had - one reason we changed the FFT library in 8.x versions.

I checked the 8.6 results and FFT Smoothing is behaving like twice that number in 7.5 and therefore you cannot duplicate the 7.5 results since half of 7 is 3.5 and the dialog only accepts integers. An average of 3 point and 4 point smoothing in 8.6 is about equal to 7 point smoothing in 7.5.
Go to Top of Page

schneide

Germany
45 Posts

Posted - 10/11/2012 :  11:18:23 AM  Show Profile  Edit Reply  Reply with Quote  View user's IP address  Delete Reply
Thanks Kathy, thanks Greg

I see now what is going on. It would be nice if smoothing would allow non integer entries. Then a much better fine tuning would be possible. (along the lines of the fft-filter function)



Giselher
Go to Top of Page
  Previous Topic Topic Next Topic Lock Topic Edit Topic Delete Topic New Topic Reply to Topic
 New Topic  Reply to Topic
 Printer Friendly
Jump To:
The Origin Forum © 2020 Originlab Corporation Go To Top Of Page
Snitz Forums 2000