Filter Butterworth for EEG Signal with Mindwave

[rmfajar]

Hello master

I've problem to make filter for my EEG Signal using Mindwave. I want to extract this signal into Alfa (8-15Hz) and Beta (16-30Hz). Cause this signal in time domain, I use FFT to make it into freq domain. This FFT worked well but my filter didn't work.

Can anyone help me for this please. This is my diagram block

I try to make filter using mathscript node, but I make mistake in this syntax, here it is

Thanks anyway for your help master.

Sincerely,

Fajar

[GerdW]

Hi Fajar,

please revisit all your code and compare it with your requirements…

Why do you calc the FFT of a single sample? Does it make sense to you?

Why do you append phase and amplitude results of your FFT? What's the reason here?

Why do you bandpass the result of your FFT? What is the expected result?

Why do you save to a file named "xls" when it is just a simple text file?

[rmfajar]

Thanks for your response Gerdw 

I append phase and amplitude results of my FFT because I get it in tutorial to process raw signal into FFT. Is it wrong?

I want to filter the signal of 8 Hz-15 Hz and 16Hz-30Hz, then I changed from the time domain into the frequency domain, then I use a bandpass filter.
And the expected results in the form of the output signal of frequencies (8 Hz-15 Hz and 16Hz-30Hz).

I just want to see the raw signal in number so I save it into excel.

So, what I've to do to make it happen?

Thanks anyway

Sincerely,

Fajar

[Dennis_Knutson]

You haven't addressed the fundamental flaw of doing all of this processing on a single sample.

[johnsold]

Filtering the output of the FFT with time domain filters is meaningless. Use the time domain filters on the original time domain signal (once you have accumulated enough data to get useful results).

You can select certain bins in the output of the FFT and recombine those to create a filter effect, but it is not always a straightforward process.  For someone with limited understanding of how spectral analysis works (as you appear to be), I suggest that you start with simple time domain processing.

Also be aware that the filters you have specified will not have very much separation in the frequency bands you want. Those bands are close enough that very sophisticated filters may be required, depending on how much attentuation you need in the other band.

Lynn