# Why does the filtered signal amplitude increase exponentially?

I am new to signal processing. I am having a problem applying a filter to a test signal. I am working with MatLab.

My test signal is a chirp up to Nyquist frequency and my filter is in zeros and poles form. I am using zp2tf and filter functions but I have amplitudes that increase with frequency up to 3000dB approximately.

The problem is that this amplitudes surely are bad, then I ask for help to understand how can I correct this problem and why it appears.

Next, I paste my code for your review

close all
clear all
clc
fl = 0.05;                            % chirp low frequency
fh = 3.3125;                          % chirp high frequency
fs = 2*fh;                            % signal sampling frequency
dt = 1/fs;                            % delta of time
T = 20;                               % signal durartion
t = [0:dt:T]';                        % time vector
x = 10^(-8)*chirp(t,fl,T,fh);         % test signal
z = ';                             % zeros vector (only one)
p = [-3.56047E-01+2.83702E+00*1i,...  % poles vector
-3.56047E-01-2.83702E+00*1i,...
-6.28319E-02+0.00000E+00*1i,...
-8.06237E+00+3.33954E+00*1i,...
-8.06237E+00-3.33954E+00*1i,...
-8.06237E+00+3.33954E+00*1i,...
-8.06237E+00-3.33954E+00*1i,...
-3.33954E+00+8.06237E+00*1i,...
-3.33954E+00-8.06237E+00*1i,...
-3.33954E+00+8.06237E+00*1i,...
-3.33954E+00-8.06237E+00*1i]';
k = 2.88869E+07*3.57916E+09;          % gain constant
[b,a] = zp2tf(z,p,k);
y = filter(b,a,x);
hfig = figure;
plot(t,db(y));


Thank you for all your help and valuable time!

• It looks like the poles are poles of a continuous-time system. You can't use them directly in a discrete-time system, because the corresponding system will be unstable, and that's what you see in the amplitude. – Matt L. Jun 21 '20 at 7:39