LabVIEW

cancel
Showing results for 
Search instead for 
Did you mean: 

frequency jump from extract single tone information VI

Solved!
Go to solution

I am analyzing a bunch of time series signals meausred at diffrent magnetic fields to pick up the main compoment frequency and plot them.

Method 1: I use the extract single tone information VI. It can easily detect the center freqency of the spectrum and the results looks smooth except there is always frequecy jump at some positions. See the following snapshot with red circle highlighted.

Method 2: I use curve fitting function(Lorentz function) on FFT data (from extract single tone information VI output) to get the center freqencies. The plot seems smooth all the way.  But it looks noisier than that the extract single tone information VI gives.

 

The question is, how these jump happens using the extract single tone information VI detected frequency? Why are the fitting results noisier than that the extract single tone information VI gives?

Thanks.

 

 

WT

0 Kudos
Message 1 of 9
(2,662 Views)

Hi WT,

 

do you mind to attach a VI with some default data in the graphs?

How are you creating all those plots?

Best regards,
GerdW


using LV2016/2019/2021 on Win10/11+cRIO, TestStand2016/2019
0 Kudos
Message 2 of 9
(2,652 Views)

HI GerdW,

 Thanks.

attached are part of the data set.

Use the FFT-fit-profile-auto_ver0.2a.vi to batch processing the data by select the folder where contains data and then click the button "AUTO start".

Method "Single Tone Info" using the extract single tone information VI to detect the main frequency.

 

Best regards,

WT

LV2013

0 Kudos
Message 3 of 9
(2,615 Views)

Hi Weitong,

 

when I run your VI with the first 3 data files I don't see any "jumps"…

Best regards,
GerdW


using LV2016/2019/2021 on Win10/11+cRIO, TestStand2016/2019
0 Kudos
Message 4 of 9
(2,608 Views)

Hi GerdW,

 

I may not present my question clearly. Please see the attached snaphsots.

 

Thanks.

 

WT

 

0 Kudos
Message 5 of 9
(2,601 Views)

Hi Weitong,

 

so you are comparing two very different methods to find a frequency in a spectrum - and wonder about small deviations of 0.05Hz around a main peak of ~16164Hz? (That is less then 3.1e-4%!)

Best regards,
GerdW


using LV2016/2019/2021 on Win10/11+cRIO, TestStand2016/2019
0 Kudos
Message 6 of 9
(2,596 Views)

GerdW

We are doing very sensitive measurement. This jump is a big deal !

 

Please see the full data results.

Thanks.

fequency detected by Extract single Tone Informationfequency detected by Extract single Tone Informationfrequency detected by  fittingfrequency detected by fitting

Best regards,

WT

0 Kudos
Message 7 of 9
(2,589 Views)
Solution
Accepted by topic author weitong

The Extract Single Tone Information.vi algorithm applies a Hanning window and then looks for the frequency bin that is largest in magnitude and uses that bin and the adjacent bins (total of three bins) to estimate the tone frequency. This means that other bins are ignored in the estimation. Depending on your noise you may get "unlucky" with the estimate. You might try using the Extract Multiple Tone Information.vi, as it uses a different algorithm that uses more bins. When I tried Extract Multiple Tone Information.vi on your data it yielded a smoother estimate.

I did not look carefully at the formulas, but the spectrum that is passed to the fitting VI is in units of dB. Extract Single Tone Information.vi does not convert to dB before estimation, so there may be slight differences because of that. 

-Jim

Message 8 of 9
(2,569 Views)

Hi DSPGuy

 

Thanks for your solutions. The Extract Multiple Tone Information.vi  does work!

 

Best regards,

WT

0 Kudos
Message 9 of 9
(2,552 Views)