Forums

Re : Ripple and Filter

Started by Vipperla Ravi Chander N Rao October 7, 2003
Hi Mob,
When you look to design a filter, ideally you would want a brick wall
type of filter. But then to design a filter with such a characteristic
is physically not possible. So you try and design a filter whose curve
would fit within the desired characteristics. That is you allow for
certain ripple in the passband (which is nothing but the fluctuation in
the gain at different frequencies in the spectrum of interest) and also
a transition band.

say for eg:
|______________
|_____A________|__
| | |
| | |
| |B |
| | |
| | |
| |__|_________________
|______________|__|_____C___________
fp fs f
In the above case you specify that you can allow rp amount of deviation
in the gain in the passband and want 'As' amount of attenuation
relative to passpand in the stop band.

Using chebyshev, butterworth, elliptic etc approximations, you fit in a
curve that would lie entirely within the three boxes (named A,B and C)
in the diagram. Each type of approximation would give you a different
type of curve with different order of polynomial (and hence filter) in
each case.

Now coming to your question of choice of rp and rs, The smaller ones
you choose, the order of filter would increase accordingly. So it is
more of a tradeoff between your desired response and computational
complexity.

Hope whatever I have written makes some sense to you.

regards,
Ravi >Could anyone tell me how to choose ripple for passband and stopband in
>fliter design? Any definition please? To me it is not clear.
>Great thanks.
>Mob.

>example:
>rp = 0.01; % Passband ripple
>rs = 0.1; % Stopband ripple
V Ravi Chander
Student ID : 200211014
MTech(ICT)
DA-IICT, Gandhinagar
.
http://intranet.da-iict.org