MATLAB - Troubles filtering very noisy signals using narrow pass band filter

[juancervi]

Hello,

I am trying to filter a very noisy signal but using a narrow pass band filter there is an undesireded behavior on ressults.
While narrower the filter is applied, more signal sections becomes unusable. I apply filters type Chebishev TI and TII, elliptic and Butterworth.
The figure attached shows an undesirable effect: in the first seconds the filtered signal remains without values. For narrower filters, the apparition of signal values are delayed even more.

Thanks in advance.

 

[Skogsgurra]

Narrow filters need time to "swing in". So, what you see is what you should expect to see.


Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[IRstuff]

What exactly are you trying to accomplish? Your narrowband filter, in the absolutely ideal case is going to pull out a sinewave, which is useful how?

TTFN (ta ta for now)
I can do absolutely anything. I'm an expert!

https://www.youtube.com/watch?v=BKorP55Aqvg

faq731-376 forum1529

 

[benta]

Got to agree with skogsgurra. The steeper the filter you design, the longer the delay through the filter. If you look at the transfer functions, it's pretty obvious.

Benta.

 

[electricpete]

That's similar to the way that machines are moved quickly through resonance to avoid high vibrations. Moving slowly thru resonance creates higher vibration than moving faster thru resonance.

The more lightly damped, the higher the steady state amplitude for given excitation. But the higher that steady state amplitude the longer it takes to put in all that energy to achieve that steady state amplitude.

=====================================
(2B)+(2B)' ?

 

[electricpete]

The more lightly damped, the higher the steady state amplitude for given excitation. But the higher that steady state amplitude the longer it takes to put in all that energy to achieve that steady state amplitude.

to finish the thought... the more lighly damped, the longer it will take to achieve the steady state amplitude.

=====================================
(2B)+(2B)' ?

 

[snarkysparky]

Pick the filter to match the spectrum of the desired signal. The noise looks broadband.
Check out Weiner filter.