Bandpass Zero phase Filtering

[Haristhota]

I tried implementing zero phase filter for filtering noisy EEG signals.

It is behaving like a bandstop filter when tried bandpass mode.

but bandstop, HPF, LPF are working properly.

I tried cascading of ZPHPF & ZPLPF but it also didn't work.

What would be the problem with my VI?

[johnsold]

It appears that the typecast function in Zero Phase Filter.vi is sending the wrong information to Butterworth Coefficients.vi.  I placed U32 numeric indicators on the Type output of the Unbundle and on the output of the typecast. 

Type Unbundle  Typecast out

Lowpass   0            0

Highpass  1            3

Bandpass 2            3

Bandstop  4            3

That may not be the only problem however. When I bypassed the typecast, I still get NaN outputs for both bandpass and bandstop filters.

Why are you sampling at 500 kHz and then resampling to a much slower rate?  When I apply the same filter specifications to the raw data, I get the results shown below.

This was using Digital IIR Filter.vi from the Waveform Conditioning palette (LV2015). I am not sure whether the filter VIs have changed since LV 2011.

Why are you worried about a zero phase filter? The spectrum of the raw signal appears to be essentially random.

Lynn

[Henrik_Volkers]

Numeric filter don't like to have a 500kHz Samplerate and a filtersetting of 1Hz to 45Hz.

Think of a 1Hz sine sampled at 500kHz and a 4th order filter. the coefficients will scrap at the lower end of the numerical resolution ....  add noise and the filter is quickly lost 😄

Decimate (with filter) to 500Hz samplrate and try again 😉

Another way to compensate the filter phase of  finite samples is  to filter twice with reversed data.

data -> filter -> reverse data -> filter - reverse data

[Haristhota]

Hi Lynn & Henrik,

I got a problem in calcualation of percentatge of similarity between two EEG signals which are acquired by diffent type of electrodes  by using correlation technique. I tried avearaging the data 500Khz at 1000 samples also.

I believe that i am not getting the correct value of percentage.

I am auto correlating the referance signal and maximum of autocorrelation was found out.

similarly  i am crosscorrelating the referance signal with test signal and finding out the maximum.

from those two vaues i am tryoing to find out percentage of similarity.., is it correct aproach??

Please lookinto the attached VI.