Download as pdf or txt
Download as pdf or txt
You are on page 1of 31

Department of Electronics and Communication

Engineering

Digital Signal Processing

Subject Code – ECN-312

Frequency conversion for filters


(IIR)
Elliptic Filter
Jacobian Elliptic Function

*Equiripple
response in both
the pass band and
stop-band.
* Both poles and
zeros.
Elliptic Filter

Passband Ripple:
Limitation: Nonlinear phase response!
Computation of frequency response
in terms of pole-zero
Computation of frequency response
in terms of pole-zero

*If poles and zeros are known, frequency response can be calculated!
Example: Suppose we know pole at point A and zero at point B.
We wish to calculate at specific value of frequency.
•When fourier-transform is evaluated at unit circle, it represents Z-transform!!
•We can rotate phase by ω on unit circle to get point L.
•Z-transform at point L is equivalent to F-transform in complex plane.
Geometric Interpretation
Application of Geometric
interpretation
Example case given in the adjacent plot:
What happens if
=0
=0
What happens if
=0
=∞
• Zero near unity circle makes magnitude smaller!
• Pole near unity circle makes magnitude larger!
• Placing zero near pole cancels effect of poles and zeros.
Application

* Peak of filter response is at ω=0, which is closest to pole on unity circle.


Pole-Zero approach for designing basic
filters
*To locate poles near unit circle corresponding to
the frequency to be emphasized.
* To locate zeros near unit circle corresponding to
the frequency to be deemphasized.
* All poles should be inside unity circle (for
stability).
* All complex zeros and poles must occur in
complex conjugate pairs for filter coefficients to
be real.
Pole-Zero approach for designing basic
filters

* b0 is selected so that
*
*N is selected to be higher or equal to M.
*Pole –zero placement for high-pass, low-pass,
band-pass, digital resonators and comb filters
can be done.
Pole-Zero approach for designing
basic filters
• Low-pass filters: Poles should be near to unity
circle for low frequencies, and zero at high
frequency.
• High-pass filters: Poles should be near to unity
circle for high frequencies, and zero at low
frequency.
Example of single pole filter:
Select low-pass and high-pass filter
response
Further:

Case-1: when
only pole is
there!
High -Pass

Case-2: Zero is
adding to low
pass behavior

Low -Pass
Example
Find b0 and P such that:
Converting from low-pass to high pass
filter
* Shifting in frequency domain allows conversion from
low-pass to high-pass and vis-à-vis.
Low-pass to high-pass: If:

Similarly:

Can be obtained by changing


signs of odd number samples!!
Digital Resonators
• Special two-pole band-pass filter with the pair of complex
conjugated poles near unity circle.
• Called resonator because large response near pole.
• Angular frequency decides resonant frequency.

Select b0 such that


Digital Resonators
Magnitude is high at or near ω= ω0. Select b0 such that

Lets recheck our theory by maximizing


Note: When r is unity,
following expression:  r = 0
b0
H ( ) = which is angular
(1 − re j0 e− j )(1 − re− j0 e− j ) frequency of pole.
Notch filters
*Notch filter contains one or
more deep notches are selected
frequencies.
*Useful in applications when
we want to suppress particular
frequency.
•Opposite to resonators, we put pair of complex-conjugate zeros at
desired frequencies.

• Simple FIR filter can be given as:


Notch filters
For

Limitation : Relatively large bandwidth!


A simple solution can be: Introducing a pair of poles!
Effect of adding pole

*Introduction of pole will introduce ripple in the pass-


band.
*ripple can be reduced by adding more poles and zeros.
* Limitation is that final result is achieved by hit and trial
method!
Comb filters
* Comb filter is extension of notch filter which
repeats notches at periodic intervals.

Zeros can be given as:

There is only one pole which


is cancelled by z=1.

Example case of M=10


Comb filters

In more general term, we can


create a comb filter by creating:

By replacing z by zL , we
introduce FIR filtering:
Comb filters
Now coming back to original
example, by replacing z by zL

For M=5. For M=10, L=3.


All-Pass Filters
Constant magnitude for all frequencies:

system function for an all-pass filter can be given as:


All-Pass Filters
If we calculate magnitude of system function, we
will find that

* If there is a pole at , there will a zero at .


Example of pole-zero pattern for all pass filters
Frequency Transformation in analog
Domain
To

To
Frequency Transformation in analog
Domain
To
Example

Zeros and Poles


of the system?
Frequency Transformations in Digital
Domain
Frequency Transformations in Digital
Domain

Example:
Frequency Transformations in Digital
Domain

You might also like