Lecture 10: Impulse Response of a Differential LTI System

4.3 Impulse Response of a Differential LTI System

Consider again the general form of a causal LTI differential system:

N
d k y (t ) M d k x(t )
ak dt k bk dt k
k 0 k 0

4.3.1 Method 1: Impulse Response Obtained by Linear Combination of Impulse

Responses of the Left-Hand Side of the Differential Equation

The step-by-step procedure to find the impulse response of a differential LTI system is as follows.
1. Replace the whole right-hand side of the differential equation by (t ) ,

2. Integrate from t 0 to t 0 to find a set of initial conditions at t 0 ,

3. Calculate the homogeneous response ha (t ) to the homogeneous equation with these initial
conditions,
4. Finally, differentiate the homogeneous response ha (t ) and use linear superposition to form
the overall response of the system

Step 1: Under the assumption that N M and the system is initially at rest, i.e.,
N 1
dy(0 ) d y( 0 )
y( 0 ) 0 we first replace the right-hand side of Equation by a
dt dt N 1
single unit impulse:
N
d k ha (t )
ak
dt k
(t ) .
k 0

Step 2: To solve this equation for ha (t ) , we first observe that the impulse can only be generated
by the highest-order derivative (i.e., k N ) of ha (t ) . Otherwise, we would have derivatives of the
impulse function in the right-hand side of . This means that the functions
dha (t ) d N 1ha (t )
ha (t ), , , are smooth or have finite discontinuities at worst. Such functions
dt dt N 1
integrated over an infinitesimally small interval simply vanish. This observation gives us the first
N-1 initial conditions for ha (t ) when we integrate those functions from t 0 to t 0 :

z
0
d k ha ( ) d k 1ha (0 )
d 0, k 1, , N 1
0
dt k dt k 1

Then, integrating both sides of from t 0 to t 0 we obtain

1
 z z
0 0
N
d k ha ( ) d N ha ( ) d N 1ha (0 )

k 0
a k
dt k
d a N
dt N
d a N
dt N 1
1
0 0

Which gives us our N th initial condition at t 0 :

d N 1ha (0 ) 1
.
dt N 1 aN

Step 3: Thus starting at time t 0 , we need to solve the homogeneous equation

N
d k ha (t )
ak dt k 0 ,
k 0

subject to the above initial conditions. Assume that the solution has the form of a complex
exponential Ae st for t 0 . Substituting this exponential into Equation , we get a polynomial in "s"
multiplying an exponential on the left-hand side:
N
Ae st ak s k 0
k 0

and this equation holds if and only if the characteristic polynomial p( s) vanishes at the complex
number s :
N
p( s): ak s k a N s N a N 1s N 1 a0 0
k 0

By the fundamental theorem of algebra, this equation has at most N distinct roots. Assume that
the N roots are distinct, and call them ls q . This means that there are N distinct functions
k
N
k 1
Ak e sk t that satisfy the homogeneous equation . Then the solution to can be written as a linear
combination of these complex exponentials:
N
ha (t ) Ak e sk t .
k 1

The complex coefficients lA q can be computed using the initial conditions:

k
N
k 1
N N
0 ha (0 ) Ak e sk 0 Ak

k 1 k 1

dha (0 ) N
0 Ak sk
dt k 1

d N 1ha (0 ) N
1 Ak sk N 1
dt N 1 k 1

This set of linear equations can be written as

2
 0 OL1
L 1 1 O
LA O
0P M
M P
MA P
1

M
M
0P
P M
M
s
s
1 s2
s2 2
sN
sN 2
P
M
P
M P
A P
2

M P
P M
M M
1

P
M
P
M P
.
P
3

M
1P
N QMN
s 1
N
s2 N sN N P
M
Q
NA P
Q N

The matrix in this equation is called a Vandermonde matrix and it can be shown to be nonsingular
(invertible). So a unique solution always exists for the Ak 's which gives us the unique solution
ha (t ) through Equation .

Step 4: Finally, by linearity of the differential system, the response of the left-hand side of to its
right-hand side is a linear combination of ha (t ) and its derivatives. To see this, note that we have
found the impulse response ha (t ) of the system
N
d k y1 (t )
ak dt k
x (t ) .
k 0

In a block diagram form:

x (t ) ( t ) y1 ha (t )
x( t) (t ) ha (t )

Now taking the first derivative (a causal LTI system with the "unit doublet" (t ) as impulse
response) of (t ) times a scalar as an input, we get

(t ) y1 (t )
k (t ) ha (t )

Which, by commutativity, is equivalent to

dha (t )
(t ) y1 (t ) k
dt
ha (t ) k (t )

Therefore the impulse response of the general causal LTI system described by Equation is given
by

d k ha (t )
M
h(t ) bk .
k 0 dt k

Example: Consider the first-order system initially at rest with time constant 0
dy(t ) d
0 y( t ) x ( t ) x ( t ) .
dt dt
Step 1: Set up the first problem of calculating the impulse response of the left-hand side of the
differential equation.

3
 dha (t )
0 ha (t ) (t )
dt
Step 2: Find the initial condition of the homogeneous equation at t 0 by integrating from
dha (t )
t 0 to t 0 . Note that the impulse will be in the term 0 , so ha (t ) will have a finite
dt
jump at most. Thus we have

z
0
dha ( )
0 d 0 ha (0 ) 1,
0
dt
1
hence ha (0 ) is our initial condition for the homogeneous equation
0
dha (t )
0 ha (t ) 0 .
dt
1
Step 3: The characteristic polynomial is p( s) 0 s 1 and it has one zero at s , which
0
t
means that the homogeneous response has the form ha (t ) Ae 0 for t 0 . The initial condition
allows us to determine the constant A:
1
ha (0 ) A ,
0
t
1
so that ha (t ) e 0 u(t ) .
0
Step 4: Finally, the impulse response of the differential system of Equation is
dha (t )
h(t ) ha (t )
dt
d 1 0 F 1
t
I t

dt 0 G
e u(t ) e 0 u(t )
H 0 JK
t t t
1
0 1 1
e u(t ) e 0 (t ) e 0 u(t )
02 0 0
F1 1 IJe
G

t
0 1
(t )
H K 0 0
2
u(t )
0

1
0 h(t )
1 1

0 02

4
4.3.2 Method 2: Impulse Response Obtained by Differentiation of the Step Response

dh(t )
We have seen that s(t ) . Thus we can obtain the impulse response of an LTI differential
dt
system by first calculating its step response, and then differentiating it.

Example: We will illustrate this technique by an example of a second-order system. Consider the
following causal LTI differential system initially at rest.
y(t ) 3 y (t ) 2 y(t ) x (t )
Let x (t ) u(t ) . The characteristic polynomial of this system is

p( s) s 2 3s 2 ( s 2)( s 1) ,
and its zeros (i.e., values of s for which p( s) 0 ) are s 2 and s 1 . Hence the
homogeneous solution has the form

yh (t ) A1e 2 t A2 e t .
We look for a particular solution of the form y p (t ) K for t 0 when x (t ) 1 . Substituting in
1
Equation , we find y p (t ) .
2
Adding the homogeneous and particular solutions, we obtain the overall step response for t 0:
1
s(t ) A1e 2 t A2 e t .
2
The initial conditions at t 0 are y ( 0 ) y( 0 ) 0 . Because the input signal has a finite jump
at t 0 , it will be included in (
y t ) only and y(t ), y(t ) will be continuous. Hence
y (0 ) y (0 ) 0, y(0 ) y(0 ) 0 and

1
y(0 ) A1 A2 0
2 .

y (0 ) 2 A1 A2 0
1
The solution to these two linear algebraic equations is A1 , A2 1, which means that the
2
step response of the system is:

F1 e 1I
H2 K
2 t
s(t ) et u(t ) .
2

s(t )

5
Finally the impulse response of the second-order system is obtained by differentiating the step
response. It is interesting to look around the origin to see whether or not there is a jump in the
derivative of the step response. The derivative is obviously 0 for t 0 . For t 0 ,

h(t )
d
dt
c h
s(t ) e t e 2 t ,

which evaluates to 0 at time t 0 . Hence there is no jump, and the impulse response is (check
that it solves Equation ):

c h
h(t ) e t e 2 t u(t )

h(t )