Nulling filter for removing an pulse

Status
Not open for further replies.

pattalol

Newbie level 3
Joined
Oct 17, 2014
Messages
4
Helped
0
Reputation
0
Reaction score
0
Trophy points
1
Visit site
Activity points
29
hey!

I'm having some problems removing a pulse from my signal.

I have the output y[n]=x[n]+0.25x[n-Ne], where Ne = 2205.

System function H(z)=(z^Ne + 0.25)/z^Ne

What i want to do is to remove the pulse 0.25*delta[n-Ne].

Im very new to DSP, so what i first did was to try cascading FIR filters, but it didn't work... so im kinda lost atm.

So can anyone give me some ideas of how I can remove the peak 0.25delta[n-Ne]???

As you can see, i want to remove the last pulse
 
Last edited by a moderator:

You'll remove the "echo" in the transfer function by a filter with the inverse transfer function.

The echo cancellation filter will be however an IIR filter.
 
Thanks for the idea Fvm!

Will read up on IIR filters, and try again
 

Calculated the inverse transfer function filter

Another question arose tho... why is this not a FIR filter???
You said it was an IIR filter, but i cant seem to understand how this is using feedback??
 

Since your primary and echo outputs are both impulses, a time domain filter based on a stable echo window can be gated off as done in Radar for early echoes rather than late echoes as in your case.

The system application needs more details.
 

Another question arose tho... why is this not a FIR filter???
You said it was an IIR filter, but i cant seem to understand how this is using feedback??

H(z)=1 + 0.25z^-Ne is the system transfer function, which is FIR

Hinv(z) = 1/(1 + 0.25z^-Ne) the inverted transfer function is an IIR filter
 

H(z)=1 + 0.25z^-Ne is the system transfer function, which is FIR

Hinv(z) = 1/(1 + 0.25z^-Ne) the inverted transfer function is an IIR filter

yeah i understand that now by proving that the output is only delta[n] when connecting an IIR filter with H(z)=1/(1+0.25z^-Ne), so ty!
If i wanted to find the impulse response for the filter, what should i do??

If Ne = 1 it would be easy... just use the table; (-0.25)^n * u[n]... if Ne was 2 i could use partial fractioning to find the answer...
But in my case Ne= 2205 - so how do i do it then?
 

If z-domain transfer functions aren't descriptive for you, you can model the system transfer function and compensating filter with hardware blocks (delay Ne, factor 0.25, summer respectively substractor).

- - - Updated - - -

 

Status
Not open for further replies.

Similar threads

Cookies are required to use this site. You must accept them to continue using the site. Learn more…