Convolution Integral

Convolution Integral:
Lecture 5
y (t ) = x(t )* h(t ) = x( )h(t )d

Time-domain analysis:
Convolution
System output (i.e. zero-state response) is found by convolving input x(t)
with Systems impulse response h(t).
(Lathi 2.4)

y (t ) = x(t )* h(t )
LTI System
Peter Cheung Impulse Response
Department of Electrical & Electronic Engineering
Imperial College London
h(t)

URL: www.ee.imperial.ac.uk/pcheung/teaching/ee2_signals
E-mail: p.cheung@imperial.ac.uk

PYKC 24-Jan-11 E2.5 Signals & Linear Systems Lecture 5 Slide 1 PYKC 24-Jan-11 E2.5 Signals & Linear Systems Lecture 5 Slide 2

Convolution Table (1) Convolution Table (2)

Use table to find convolution results easily:

L2.4 p177

PYKC 24-Jan-11 E2.5 Signals & Linear Systems Lecture 5 Slide 3 PYKC 24-Jan-11 E2.5 Signals & Linear Systems Lecture 5 Slide 4
 Convolution Table (3) Example (1)

Find the loop current y(t) of the RLC circuits for input when all
the initial conditions are zero.

We have seen in slide 4.5 that the system equation is:

The impulse response h(t) was obtained in 4.6:

The input is:

Therefore the response is:

L2.4 p177 L2.4 p178

PYKC 24-Jan-11 E2.5 Signals & Linear Systems Lecture 5 Slide 5 PYKC 24-Jan-11 E2.5 Signals & Linear Systems Lecture 5 Slide 6

Example (2) When input is complex

Using distributive property of convolution:
What happens if input x(t) is not real, but is complex?

If x(t) = xr(t) + jxi(t), where xr(t) and xi(t) are the real and imaginary part of x(t),
then

Use convolution table pair #4:

That is, we can consider the convolution on the real and imaginary components
separately.

L2.4 p178 L2.4 p179

PYKC 24-Jan-11 E2.5 Signals & Linear Systems Lecture 5 Slide 7 PYKC 24-Jan-11 E2.5 Signals & Linear Systems Lecture 5 Slide 8
 Intuitive explanation of convolution Convolution using graphical method (1)

Assume the impulse response decays Determine graphically y(t) = x(t)*h(t) for x(t) = e-tu(t) and h(t) = e-2tu(t).
linearly from t=0 to zero at t=1.

Divide input x() into pulses.

The system response at t is then
determined by x() weighted by h(t- )
(i.e. x() h(t- )) for the shaded pulse,
PLUS the contribution from all the
previous pulses of x().

The summation of all these weighted Remember: variable of
inputs is the convolution integral.
integration is , not t

y (t ) = x(t )* h(t )

L2.4-2 p191 L2.4-1 p183

PYKC 24-Jan-11 E2.5 Signals & Linear Systems Lecture 5 Slide 9 PYKC 24-Jan-11 E2.5 Signals & Linear Systems Lecture 5 Slide 10

Convolution using graphical method (2) Interconnected Systems

Parallel connected system

x (t ) y(t ) = h1 (t )* x(t ) + h2 (t )* x(t )

Cascade systems &

Commutative property x (t ) y(t ) = [h1 (t )* h2 (t )]* x(t )

L2.4-3 p192

PYKC 24-Jan-11 E2.5 Signals & Linear Systems Lecture 5 Slide 11 PYKC 24-Jan-11 E2.5 Signals & Linear Systems Lecture 5 Slide 12
 Interconnected Systems Total Response

Integration:

Also true for differentiation:
Let us put everything together, using our RLC circuit as an example.

Let us assume x(t) = 10e3t u(t), y(0) = 0, y
(0) = 5.

Let ( x(t) is an impulse, and h(t) is the impulse response of
In earlier slides, we have shown that
the system)

Then g(t), the step response is:

L2.4-3 p193 L2.4-5 p197

PYKC 24-Jan-11 E2.5 Signals & Linear Systems Lecture 5 Slide 13 PYKC 24-Jan-11 E2.5 Signals & Linear Systems Lecture 5 Slide 14

Natural vs Forced Responses Additional Example

Note that characteristic modes also appears in

*
zero-state response (because it has an impact
on h(t)).

We can collect the e-t and e-2t terms together, and
call these the NATURAL response.

The remaining e-3t which is NOT a characteristic
mode is the FORCED response.

L2.4-5 p197

PYKC 24-Jan-11 E2.5 Signals & Linear Systems Lecture 5 Slide 15 PYKC 24-Jan-11 E2.5 Signals & Linear Systems Lecture 5 Slide 16