notch waveform

[ampm]

Dear colleagues,

I need to construct a notch waveform for my research project.
"Notch waveform" means the waveform of the following type: it contains all frequencies in the interval f1-f2, nothing in the interval f2-f3 and all frequencies from the f3-f4 interval. In other words "notch" is obtained via reverse Fourier transformation of the following profile in the frequency domain:

    f1 _________f2       f3________________f4
__|                       |____|                                     |_________

Currently I have no idea how to do this in LabView and any help would be highly appreciated.

Thanks in advance!

[BillMe]

If you have the Signal Processing package you can use the Waveform Conditioning "Filter.vi" to implement a bandstop (notch) filter. If you don't have the package, you can do it by the grunt method as I've done:
Do a FFT on your time domain data. Then examine the bins of the FFT data and overwrite the frequency amplitudes (from F2 cutoff to F3 cutoff as in your example) that you want to notch out with appropriate values (similar to your expected noise floor amplitude or with some rolloff on the band edges).

Message Edited by Bill@NGC on 07-26-2007 11:11 PM

[ampm]

Thank you!
Question: how one can do FFT and inverse FFT in LabView?
My task is not to make a bandstop filter and I don't have any time domain data.
What I actually need is to construct the time domain waveform by inverse Fourier transformation of the "notch" profile in the frequency domain. Currently I have no clue how to do this in LabView.

[johnsold]

The FFT and inverse FFT VIs are in Signal Processing >> Transforms. The output is in the form of an array. Determine which elements of the array represent the frequencies you want to notch out and set them to zero before doing the inverse FFT. The frequency of element [i] is i*df. The help files specify how to calculate df.

Lynn

[BillMe]

Just to reemphasize that the value to which you set the bandstop frequencies should equal your expected noise floor and Y value units. In other words, you want to push those bandstop frequencies well down into the noise. A zero value may not necessarily be appropriate.