Anda di halaman 1dari 33

# FREQUENCY RESPONSES

## Signals and System Analysis

Transfer function G(s)

s =jω

## Frequency (strictly angular frequency) ω

Magnitude |G(jω)|

Phase ∠G(jω)

## Signals and System Analysis

Significance of the frequency response
It indicates how much the system responds to a
sinusoidal input at different frequencies

## Frequency response G(jω)

Mmagnitude |G(jω)|

Phase ∠G(jω)

## Signals and System Analysis

There are several ways of plotting frequency
responses but the two most common are:

Direct plotting
|G|-ω ∠G-ω

Bode diagrams
|G| - ω in log scale
∠G- ω in log scale

Real(G) -Imag(G)

## Signals and System Analysis

Any transfer function can be plotted

or

## Signals and System Analysis

The Bode diagram consists of two graphs:

## 1. |G(jω)| against ω on a log scale

The gain in dB, that is 20log10 |G(jω)|, is plotted on a linear
scale

## The Nyquist plot is a locus with ω as a parameter showing

|G(jω)| and ∠G(jω as a curve as ω varies.

## Signals and System Analysis

Bode plots of the following transfer functions

K – a gain

1/sT – an integrator

system.

## Signals and System Analysis

K – a gain

G(s)=K G(jω)=K

G ( jω ) = K
0
∠G ( jω ) = tan −1
=0
K

## Signals and System Analysis

- K – a negative gain

G(s)=-K G(jω)=-K

G ( jω ) = K
0
∠G ( jω ) = tan−1
= 180o
−K

## Signals and System Analysis

P lot Frequency Res pons e for K gain
2

1.5

1
|G|

0.5

0
0 10 20 30 40 50 60 70 80 90 100

1
K=1;
0.5 omega=0:0.1:100;
G(1:1001)=K;
phas e

-0.5
G_mag=abs(G);
G_phase=atan(imag(G)./
-1
0 10 20 30 40 50 60 70 80 90 100 real(G));
om ega subplot(2,1,1);
plot(omega, G_mag);
ylabel('|G|');
title('Plot Frequency
Response for K gain');

subplot(2,1,2);
plot(omega, G_phase);
ylabel('phase');

xlabel('omega');
Signals and System Analysis
P lot Frequency Response for K gain
1

20*log10(|G|) (db)
0.5

-0.5

-1
-1 0 1 2
10 10 10 10

0.5

phas e
0

K=1; -0.5
omega=0:0.1:100;
G(1:1001)=K; -1
-1 0 1 2
G_mag=abs(G); 10 10 10 10
G_phase=atan(imag(G)./ real(G)); om ega
subplot(2,1,1);
semilogx(omega,
20*log10(G_mag));
ylabel('20*log10(|G|) (db)');
grid;
title('Plot Frequency Response
for K gain');

subplot(2,1,2);
semilogx(omega, G_phase);
ylabel('phase');
grid;
xlabel('omega');

## Signals and System Analysis

1/sT – an integrator

G(jω) = 1/jωT

|G(jω)| = 1/ωT

∠G(jω = –90º.

In dB the gain

## On a graph with frequency on a log scale this will be a

straight line with a negative slope.

## Signals and System Analysis

P lot Frequency Response for integrator
10

|G|
0
0 10 20 30 40 50 60 70 80 90 100

phas e 0
omega=0.1:0.1:100;
G=1 ./(j .*omega); -2
G_mag=abs(G);
G_phase=atan(imag(G)./0 10 20 30 40 50 60 70 80 90 100
om ega
(real(G)+0.0000001));
subplot(2,1,1);
plot(omega, G_mag);
ylabel('|G|');
title('Plot Frequency
Response for integrator');
subplot(2,1,2);
plot(omega, G_phase);
ylabel('phase');
axis([0 100 -pi pi]);
xlabel('omega');
Signals and System Analysis
P lot Frequenc y Res pons e for integrator
20

20*log10(|G|) (db)
0

-20

-40
-1 0 1 2
T=1; 10 10 10 10
omega=0.1:0.1:100;
G=1 ./(j .*omega);
2
G_mag=abs(G);
G_phase=atan(imag(G)./

phas e
0
(real(G)+0.0000001));
subplot(2,1,1);
-2
semilogx(omega,
20*log10(G_mag)); -1 0 1 2
ylabel('20*log10(|G|) 10 10 10 10
om ega
(db)');
grid;
title('Plot Frequency
Response for integrator'); Slope is –6dB/octave
semilogx(omega, G_phase);
axis([0 100 -pi pi]); ∠G(jω = –90º = -π/2 =-1.57 rad
ylabel('phase');
grid;
xlabel('omega');
Signals and System Analysis
An octave is a frequency change of 2.

## Geometric mean of two frequencies ωa

and ωb is the ω aω b .

## Signals and System Analysis

1/(1+sT) – a time constant (first order system)

G(jω) = 1/(1+jωT)

= 1 for ω small

## Signals and System Analysis

P lot Frequenc y Res pons e for 1s t-order s ys tem
1

0.5

|G|
0
0 10 20 30 40 50 60 70 80 90 100

T=1; -1
omega=0.1:0.1:100;
phas e

## G=1 ./(1+T*j .*omega);

-2
G_mag=abs(G);
G_phase=atan(imag(G)./
(real(G))); -3
subplot(2,1,1); 0 10 20 30 40 50 60 70 80 90 100
plot(omega, G_mag); om ega
ylabel('|G|');
title('Plot Frequency Response
for 1st-order system');

subplot(2,1,2);
plot(omega, G_phase);
ylabel('phase');
axis([0 100 -pi 0]);
xlabel('omega');
Signals and System Analysis
P lot Frequency Response for 1st order sys tem
0

20*log10(|G|) (db)
-10
T=1; -20
omega=0.1:0.1:100;
G=1 ./(1+T*j .*omega); -30
G_mag=abs(G); -40
G_phase=atan(imag(G)./ 10
-1
10
0
10
1
10
2

real(G));
0
subplot(2,1,1);
semilogx(omega,
-1
20*log10(G_mag));

phas e
ylabel('20*log10(|G|)
-2
(db)');
grid;
-3
title('Plot Frequency -1 0 1 2
10 10 10 10
Response for 1st order
om ega
system');

subplot(2,1,2);
semilogx(omega, 0dB line to ωT = 1,
G_phase);
axis([0 100 -pi 0]); `Integrator’ line has slope –6dB/octave
ylabel('phase');
grid;
The phase shift from 0 to –90º
xlabel('omega'); –45º at ωT = 1.

breakpoint

## ωT 0.1 0.125 0.25 0.5 1.0 2.0 4.0 8.0 10

dB -0.05 -0.1 -0.3 -1 -3 -7 -12.3 -18.1 -20.05
Approx 0 0 0 0 0 -6 -12 -18 -20.0
dB
ϕ -6° -7° -14° -26° -45° -64° -76° -83° -84°

## Signals and System Analysis

In DB scale
20log10 (|G(jω)|)

• If |G(jω)|=1

• If 2 ×

• If 10 ×

## Signals and System Analysis

ω02/(s2+2ζsω0+ω02) – a complex pole pair or second
order system.

unity (0dB).

## At high frequencies G(jω)> -ω02/ω2; this is two integrators

with 0dB gain at ω = ω0.

## The asymptotic approximation is therefore two straight

lines, 0dB for ω < ω0 and –12dB/oct for ω > ω0.

## The shape of both the gain and phase curves near to ω0

change significantly with the value of ζ.

## Signals and System Analysis

P lot Frequency Res pons e for 2s t-order sy stem
6

|G|
2

omega0=1; 0
zeta=0.1; 0 1 2 3 4 5 6 7 8 9 10
omega=0.1:0.1:10;
G=omega0^2 ./(- omega 0
.^2 +j* 2*zeta*omega0
.*omega +omega0^2); -1
G_mag=abs(G);
G_phase=atan2(imag(G), phas e
-2
(real(G+0.000001)));
subplot(2,1,1);
plot(omega, G_mag); -3
ylabel('|G|'); 0 1 2 3 4 5 6 7 8 9 10
title('Plot Frequency om ega
Response for 2st-order
system');

subplot(2,1,2);
plot(omega, G_phase);
ylabel('phase');
axis([0 10 -pi 0]);
xlabel('omega');

## Signals and System Analysis

P lot Frequenc y Response for 2st order s ys tem
20

20*log10(|G|) (db)
0

-20
omega0=1;
zeta=0.1; -40
omega=0.1:0.1:10; 10
-1
10
0
10
1

## G=omega0^2 ./(- omega .^2

+j* 2*zeta*omega0 .*omega 0
+omega0^2);
G_mag=abs(G);
-1
G_phase=atan2(imag(G),
phas e
(real(G+0.000001)));
subplot(2,1,1); -2
semilogx(omega,
20*log10(G_mag)); -3
ylabel('20*log10(|G|) -1 0 1
(db)'); 10 10 10
grid; om ega
title('Plot Frequency
Response for 2st order
system');
0dB for ω < 1
subplot(2,1,2);
semilogx(omega, G_phase);
–12dB/oct for ω > 1.
axis([0 10 -pi 0]);
ylabel('phase');
grid;
xlabel('omega');

## Signals and System Analysis

Plot Frequency Response for 2st order system
20
ζ=0.05,
20*log10(|G|) (db) 0
ζ=0.1
-20
ζ=0.25
-40 ζ=0.707
-60
-1 0 1
ζ=1
10 10 10

-1
phase

-2

-3
-1 0 1
10 10 10
omega

## Phase symmetrical about the –90º phase shift at ω = 1 for any ζ.

Signals and System Analysis
%Nyquist and Bode Plot
Nyquist (or polar) plot
clear all;

Example: w=logspace(-1,1,500);
s=j*w;
g=1 ./(s.^2+0.6 .*s+1);
1 Re=real(g);
G( s) = 2 Im=imag(g);
s + 0.6s + 1
figure(1);
plot(Re, Im,'*');
axis([-2 2 -2 2]);
1 grid;
G ( jω ) = title ('Nyquist Plot');
− ω 2 + 0.6 jω + 1
figure(2);
subplot(2,1,1);
semilogx(w,20*log10(abs(g)));
ylabel('20log10(|G|)');
title ('Bode Plot');
grid;
subplot(2,1,2);
semilogx(w,atan2(Im,Re));
grid;

## Signals and System Analysis

Nyquist P lot
2

1.5

0.5

-0.5

-1

-1.5

-2
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2 B ode Plot
10

20log10(|G|)
-10

-20

-30

-40
-1 0 1
10 10 10

-1

-2

-3

-4
-1 0 1
10 10 10

FILTERS

## For many signal processing situations we need to design

filters to remove or amplify specific frequency
components.

## Classify filters according to four basic frequency

response properties:

1. Low pass
2. High pass
3. Band pass
4. Band reject

## Signals and System Analysis

A filter’s prpoerty can be described by the frequency
response

G (ω )
Or by the loss function

H (ω )

where

1
H (ω ) =
G (ω )

## Signals and System Analysis

A simple low-pass filter

I(s) R

1
Vi(s) Vo(s)
sC

V0 (s) 1 / sC 1
The transfer function: G ( s ) = = or G (s) =
V i ( s ) R + 1 / sC 1 + sT
1 1
T he frequency response: G (ω ) = , G (ω ) =
1 + jω T 1 + ω 2T 2

|G ( ω ) |

ω
Signals and System Analysis
There are many types of filters, including

Butterworth Filter

Chebyshev Filter

## Signals and System Analysis

Butterworth Filter

1
G (ω ) = 1/ 2
  ω  2n

1 +   
  ω1  

ω1 - cut-off frequency

## Signals and System Analysis

Chebyshev Filter

1
G (ω ) = 1/ 2
 2n

2 2 ω 
1 + ε Cn   
  ω1  

ω1 - cut-off frequency

## Parameter ε links t the pass-band ripple δ

δ = 1 − (1 + ε 2 ) −1/ 2
Signals and System Analysis
|G(ω )|