Levenberg Marquardt, phase-frequency characteristics analysis

[Kizia]

Hi, i`ve got a problem with function on wavegraph (atan). Is there any way to get rid of this middle line in between of the waveform? Or is it a problem with used data from accelerometer 

[GerdW]

Hi Kizia,

---

@Kizia wrote:

Is there any way to get rid of this middle line in between of the waveform?

---

Yes, there is.

Insert a NaN where you want to "split" the interpolation line…

[altenbach]

---

@GerdW wrote:

---

@Kizia wrote:

Is there any way to get rid of this middle line in between of the waveform?

---

Yes, there is.

Insert a NaN where you want to "split" the interpolation line…

---

While this will cosmetically solve the display problem, you mention Levenberg-Marquardt in the subject, so I assume that you maybe want to apply a non-linear fit to it. For that you need to eliminate these points and maybe even tweak the model.

Your data looks does not look like an atan function at all. What is the actual model you are trying to fit? What does the data represent? What parameters are you trying to get out of this? To me it almost looks like the raw data (before scaling) are signed integers and you interpret them as unsigned, causing wrapping. You need to fits that! Once you do, it will look like part of an atan function!

Can you attach your VI containing typical data?

[Bob_Schor]

You mentioned "this middle line in between of the waveform".

Did you mean the vertical yellow line whose "foot" is where the red arrow is pointing?  I think that it might be a cursor placed at the "peak" of the curve, where the green arrow is pointing (it is clearer in your larger picture -- it has symbol that I think is a cursor, and shows the horizontal (yellow) and vertical (also yellow) lines so you can see how the cursor position "fits" with the rest of the plot.

Bob Schor

[altenbach]

As I said, your problem is with incorrect interpretation of the measurement data.

Study this! You want the interpretation on the right! (This demo assumes 8 bit data, but can be adapted for any other number of bits)

[softball]

Hi

LabVIEW ( not base development system ) includes the VI :

Unwrap Phase VI

It does the same as Altenbachs solution, eliminating the phase jump.

Regards

[altenbach]

---

@softball wrote:

Unwrap Phase VI

---

Agreed, but that is correcting for an earlier mistake. It is significantly better to solve it upstream.

Hard to tell what's better without seeing the entire code. 😄