Solution to systems - SISO systems

 Solution using impulse response approach

A continuous system represented by a continuous time linear
differential equation as

𝑑𝑛 𝑦(𝑡) 𝑑𝑛−1 𝑦(𝑡) 𝑑𝑦(𝑡)

+ 𝑎1 + … . . +𝑎𝑛−1 + 𝑎𝑛 𝑦(𝑡) =
𝑑𝑡 𝑛 𝑑𝑡 𝑛−1 𝑑𝑡
𝑑𝑚 𝑥(𝑡) 𝑑𝑚−1 𝑥(𝑡) 𝑑𝑥(𝑡)
𝑏0 + 𝑏1 + ⋯ . . +𝑏𝑚−1 + 𝑏𝑚 𝑥(𝑡)
𝑑𝑡 𝑚 𝑑𝑡 𝑚−1 𝑑𝑡

(Dn+a1Dn-1+ a2Dn-2+….+an-1D+an)y(t) =
(boDm+b1Dm-1+ b2Dm-2+….+bm-1 D+bm)x(t) ------(1)

Where

𝑑
D -----> is the differential operator
𝑑𝑡

ai’s, bi’s are the coefficients -----values of these depends on the

system parameter values, in various combinations.

y(t) is the output signal ; x(t) is the input signal

Generally in eqn. 1 (practically) m≤n ------> two reasons

1) M-n order differentiator which is unstable if m>n

2) Derivative of noise (if any) will swamps the desired system o/p

Equation (1) can be written as ,

Q(D)y(t)=P(D)x(t)

Response of a linear system –

 variation of the output signal y(t) w.r.t. time,

 Generally the variations are shown from the time instant at which
the experiment is going to start. i.e t = 0, to time instant at which the
experiment is going to end.

 For practical purpose, if the time instant at which the experiment

is going to end is long time, then for the analysis purpose it will be
considered as t = infinity.
 A system for t ≥ 0 is the result of two independent causes

1) Initial conditions of the system (or system state) at t= 0

2) The input x(t) for t ≥ 0

 Output Response of the system has two parts

 ZIR -----> Zero Input Response

Response of the system, to the initial conditions only. i.e

due to the energy stored in the system at t= 0 in terms of the
governing variables.

i.e. without external input, That is if x(t) = 0

 ZSR------> Zero State response

Response of the system due to external input only. i.e x(t) for all t ≥ 0

i.e. without initial conditions, or assuming all the energy stored in the
system is zero at t = 0.

To find the Zero Input Response of the system:

With x(t) = 0 , equation (1) becomes, Q(D) yo(t) = 0 -----> (2)

𝑑𝑛 𝑦(𝑡) 𝑑𝑛−1 𝑦(𝑡) 𝑑𝑦(𝑡)

i.e. + 𝑎1 + … . . +𝑎𝑛−1 + 𝑎𝑛 𝑦(𝑡) = 0
𝑑𝑡 𝑛 𝑑𝑡 𝑛−1 𝑑𝑡

Eqn. (2) shows that a linear combination of yo(t) and its successive
derivative is zero, not at some value of t, but for all values of time t,
of the response.

Such result is possible if and only if yo(t) and all n successive

derivatives are of the same form. Otherwise their sum can never be
zero, for all values of t.

eλt , where e and λ are constant has this property. Hence the non
trivial solution is

yo(t)=C eλt

D yo(t) = C λ eλt ;

D2 yo(t) = C λ2 eλt ; …. ;

Dn yo(t) = C λn eλt; all will have the same form.

Substituting these in equation (2),

C(λ n+a1 λ n-1+ a2 λ n-2+….+an-1 λ +an) eλt=0 ------------------> (3)

A non trivial solution valid for all value of t is λ n+a1 λ n-1+ a2 λ n-

2+….+a
n-1 λ +an = 0 ------> (4)

That is Q(λ) = 0; or (λ- λ1) (λ- λ2)…. (λ- λn)=0,

Where λi i = 1,2,3…n are the roots of the polynomial Q(λ)

For each values of λ, equation (3) will be satisfied, so that, total

solution to equation (2) is

yo(t)=C1eλ1t+ C2eλ2t+…..Cneλnt ------------> (4)

Here,

 Q(λ) = λ n+a1 λ n-1+ a2 λ n-2+….+an-1 λ +an = 0 is the characteristic

polynomial of the system

 λi i = 1,2,3…n are the characteristic roots of the system. Also

called eigen values, characteristic values

 eλ1t, eλ2t,…..eλnt are the characteristic modes of the system

 solution to the equation is nothing but the linear combination of

these characteristic modes

 Constants C1,C2,….Cn can be evaluated by the initial conditions of

the system.

Note : solution eqn.4) assumes n characteristic roots λi i =

1,2,3…n are distinct

If there are repeated roots [some roots occurring more

than once] the form of solution is modified.

Eg: (D – λ)2 y(t) = 0 : the solution is

yo(t) = C1eλ1t+ C2 t eλ1t

If (D – λ)r y(t) = 0 : the solution is

yo(t) = C1eλ1t+ C2 t eλ1t+ C3 t2 eλ1t+….+ Cr tr-1 eλ1t

The characteristic modes are, eλ1t, t eλ1t, t2 eλ1t, …. tr-1 eλ1t

Note : complex roots :
λ1, λ2 will combinedly take complex characteristic modes .

they occur in complex conjugate pairs

If , λ1 =a+jb λ2 = λ1*= a-jb

yo(t) = C1 e(a+jb)t + C2 e(a-jb)t

where, C1, C2 must be conjugates, say C1 = (C/2) ejθ, C2 = (C/2) e-jθ

yo(t) = C eatcos(bt+θ)

Examples :

Matlab code: ex02solndsolve1.m OR ex02solndsolve1.m

Example : using MATLAB

Use “dsolve” and “diff” functions

Finding Zero –Input Response:

Ex: Given the differential equation, (D2+4D+3)y(t) = (3D+5)x(t) and with

initial conditions : y(0) = -3 ; dy(0)/dt = -7

>> yzir=dsolve('D2y+4*Dy+3*y=0','y(0)=-3','Dy(0)=-7','t')

O/P -----> yzir =5*exp(-3*t)-8*exp(-t)

Write an inline function to plot the above response

 To find the Impulse Response of the system:

 Method to find the Impulse Response of the system:

h(t)
Consider the differential equation,

Q(D)y(t)=P(D)x(t) ------------> (1)

 Impulse response h(t) is the system response to an impulse input δ(t)

applied at t=0 with all initial conditions zero at t = 0 -.

 The impulse input δ(t) appears momentarily at t = 0, and then vanishes

for ever. When the impulse is struck on system, “it generates energy
storage”.

 i.e. it creates non-zero initial conditions instantaneously within the

system at t = 0+.

 Although, impulse input δ(t) vanishes as t > 0, so that there is no input

for t>0, there will be response generated by these newly created initial
conditions.

 That means the impulse response has characteristic modes valid for
t>0

 At a single moment t = 0, there can be at most an impulse of some

magnitude, Hence,

h(t) = Aoδ(t) + Linear combination of characteristic mode terms

for t ≥ 0 ------------> (2)

Setting x(t) = δ(t) and y(t) = h(t) in the eqn. Q(D)y(t)=P(D)x(t) ,

And with m = n , we get, (Dn+a1Dn-1+ a2Dn-2+….+an-1D+an)h(t) =

(boDn+b1Dm-1+ b2Dm-2+….+bm-1 D+bm) δ (t)

Substituting equation 2) in this and comparing the coefficients of

similar impulsive terms on both sides we get,

Ao= bo , (comparing the Dn coefficients)

Hence,

h(t) = boδ(t) + Linear combination of characteristic modes

h(t) = boδ(t) + [C1eλ1t+ C2eλ2t+…..Cneλnt ]

if, m < n, then bo = 0, hence, boδ(t) term exists only if m = n.

the unknown coefficients of n characteristic modes can be determined

by using technique of impulse matching.

Method -1: Jump discontinuities in h(t), ḣ(t) , ḧ(t) e.t.c. at t = 0 result

in impulse terms …. Example….

Method -2 : we know that, impulse response is

h(t) = boδ(t) + Linear combination of characteristic modes

= bo δ(t) + { P(D) yn(t) } u(t)

= bo δ(t) + { P(D) [C1eλ1t+ C2eλ2t+…..Cneλnt] }

Where

Constants C1,C2,….Cn of yo(t) can be evaluated by the initial

conditions selected as

Yn(0) = yn’(0) =….. = ynn-2(0) = 0 and ynn-1(0) = 1

…. Example….

Example : using MATLAB -----Finding Impulse Response:

For the same above example, (D2+4D+3)y(t) = (3D+5)x(t) , assuming initial

conditions zero,

>>yn=dsolve('D2y+4*Dy+3*y=0','y(0)=0','Dy(0)=1','t');

>>Dyn=diff(yn);

>>ht=3*Dyn+5*yn

O/P -----> ht =2*exp(-3*t)+exp(-t)

Write an inline function to plot the above response

 Convolution concept
To find the Zero State Response of the system:
That is the response of the system to an external arbitrary
input, x(t), making the initial conditions zero.

 Here we use superposition principle.

 Let us define a pulse p(t) of unit height and width Δτ, starting at t = 0

Pulse p(t) ------->

1

0.8

0.6

0.4

0.2

-0.2
0 ΔT
Time in secs ------->

 Let the Input x(t)

2.5

2
x(t) ------->

1.5

0.5

-0.5 0 0.5 1 1.5 2

Time in secs ------->
 A signal x(t), can be viewed as :
 Train of Rectangular pulses of width Δτ and of varying heights, which
can be viewed as several impulses.

2.5

2
x(t) ------->

1.5

0.5

-0.5 0 0.5 1 1.5 2

Time in secs ------->

 The pulse starting at t = nΔτ has a height x(nΔτ), can be expressed as

x(nΔτ)p(t-nΔτ) OR

 Now, Signal x(t) can be mathematically replaced by a weighted sum

of all such pulses (or treated as impulses) spaced Δτ (Δτ--->0) i.e.

𝑥(𝑡) = lim ∑ 𝑥(𝑛∆𝜏)𝑝(𝑡 − 𝑛∆𝜏)

∆𝜏→0
𝜏
𝑥(𝑛∆𝜏)
i.e. 𝑥(𝑡) = lim ∑𝜏 𝑝(𝑡 − 𝑛∆𝜏)∆𝜏
∆𝜏→0 ∆𝜏
 𝑥(𝑛∆𝜏) 𝑥(𝑛∆𝜏)
 𝑝(𝑡 − 𝑛∆𝜏) represents a pulse p(t-nΔτ) with height .
∆𝜏 ∆𝜏
 As Δτ ----> 0 , the height of this strip ( pulse ) -----> ∞ ,
but area remains x(nΔτ)

 Approximation improves as Δτ--->0

 It is equivalent to say that the strip approaches impulse

of height x(nΔτ) i.e. x(nΔτ) δ(t-nΔτ) as Δτ ----> 0

 Hence the arbitrary signal x(t) can be viewed as a summation of

several impulses of strength x(nΔτ) i.e.
𝒙(𝒕) = 𝐥𝐢𝐦 ∑𝝉 𝒙(𝒏∆𝝉)𝜹(𝒕 − 𝒏∆𝝉)∆𝝉
∆𝝉→𝟎

 If we treat these pulses as impulses of various magnitude, applied at

t = nΔτ, starting from n = 0, we get several impulse responses for all
t ≥ 0.
The sequence can be viewed as shown :
System

INPUT -------------> OUTPUT

δ(t) ------------> h(t)
δ(t- n Δτ) -------------> h(t- n Δτ)
{x(n Δτ)Δτ} δ(t - nΔτ) --------------> Δy(t)= {x(nΔτ) Δτ} h(t-nΔτ)
 𝒙(𝒕) = 𝐥𝐢𝐦 ∑𝝉 𝒙(𝒏∆𝝉)𝜹(𝒕 − 𝒏∆𝝉)∆𝝉 --------------->
∆𝝉→𝟎

𝒚(𝒕) = 𝐥𝐢𝐦 ∑ 𝒙(𝒏∆𝝉)𝒉(𝒕 − 𝒏∆𝝉) ∆𝝉

∆𝝉→𝟎
𝝉

 As Δτ ----> 0, if we do the summation of these impulse responses, we

get a “continuous summation of these impulse responses”
called “CONVOLUTION INTEGRAL” given by
∞ ∞
𝒚(𝒕) = ∫−∞ 𝒙(𝝉)𝒉(𝒕 − 𝝉)𝒅𝝉 = ∫−∞ 𝒉(𝝉)𝒙(𝒕 − 𝝉)𝒅𝝉

 To get the ZSR, we have to convolve the arbitrary input signal x(t)
with the impulse response of the system h(t). i.e. ys(t) = x(t) * h(t)

0.35

0.3

0.25

0.2

0.15

0.1

0.05

0
-2 0 2 4 6 8 10 12

MATLAB code : ex02convolutionconcept.m

 Relevance of the impulse response of the system.
 If we know the system response to a unit impulse, we can determine
the system response to an arbitrary input signal x(t) by adding the
system response to each impulse component of the input signal x(t).

 Applicable to Linear time invariant, non-causal OR causal signals

/ systems.

 ZSR and causality

Causality restriction on both signals and systems further simplifies the

limits of integration.

Causal system’s unit impulse response h(t) is a causal signal.

 CONVOLUTION INTEGRAL for a causal system and signal is given by

𝒕 𝒕
𝒚(𝒕) = ∫𝟎 𝒙(𝝉)𝒉(𝒕 − 𝝉)𝒅𝝉 = ∫𝟎 𝒉(𝒕 − 𝝉)𝒙(𝝉)𝒅𝝉

where,

t ----> time instant at which result of convolution is of interest.

Τ ------> is the Integrating variable. Integrating from 0 to t

If x(t) is causal, x(τ) = 0 for τ < 0 ;

Similarly, if h(t) is causal, h(t- τ) = 0, for (t- τ) < 0 ;

i.e 0, for t<τ OR

i.e 0, for τ>t

x(τ) = 0 0 t h(t- τ) = 0 τ

Hence the product of x(τ) h(t- τ) = 0 everywhere, except over a non

shaded interval 0 to t
𝒕
𝒚(𝒕) = ∫ 𝒙(𝝉)𝒉(𝒕 − 𝝉)𝒅𝝉
𝟎

Example : ?
 Properties
Commutativity x(t) *y(t) = y(t) * x(t)
Associativity {x(t) *y(t)}*z(t) = x(t) *{y(t)*z(t)}
Distributivity { x(t) + y(t) } *z(t) = x(t) *z(t) + y(t) * z(t)
If y(t) = x(t) *h(t) , then
Differentiation Property y’(t) = x’(t) *h(t) = x(t) * h’(t)
Area Property Area of y = (area of x ) X (area of h)
Scaling property y(at) = |a|x(at) *h(at)
Shift property x(t) * h(t-T) = x(t-T) * h(t) = y(t-T)
And x(t-T1)*h(t-T2) = y(t – T1 – T2)
Convolution with impulse x(t) * δ(t) = x(t)
Width Property If the durations (widths) of x1(t) and x2(t)
are finite, given bt T1 and T2, respectively, then the duration (width)
of x1(t) * x2(t) is T1 + T2.

Also,

Total solution (response of the system ) is the sum of ZIR and ZSR,
computed separately

i.e y(t) = yo(t) + ys(t) = ZIR + ZSR

 Graphical convolution
Explain with MATLAB : ex02graphicalconvolution1.m
x=inline('(t>=-1)','t')
h=inline('exp(-0.346*t).*(t>=-2)','t')
What is the mathematical expression for y(t) ????
 ex02graphicalconvolution2.m
x=inline('(t>=-1)&(t<=1)','t')
h=inline('(-2*(t-1)).*((t>=0)&(t<=1))','t')
What is the mathematical expression for y(t) ????
 Classical solution to differential equations
Total solution, (response of the system) can be viewed as another set
of two parts

1) Natural response ---> Homogeneous solution (or Complementary

solution), which contains linear combination of Characteristic modes.

2) Forced component of response -----> Particular solution, which

contains Non – Characteristic modes, generally response is of the same
form as forcing response (input response)

y( t) = yn(t) + yf(t)

Given the differential equation,

Q(D){yn(t) + yf(t)} = P(D)x(t)

i.e. Q(D)yn(t) + Q(D)yf(t) = P(D)x(t)

Since, yn(t) has only Linear combination of characteristic modes,

and has the same form as ZIR.

The first term in L.H.S of diff. eqn. Q(D)yn(t) = 0,

But, the arbitrary constants (coefficients of characteristic modes ) are

different, from that of ZIR and is evaluated from the auxiliary
conditions.

Note: Auxiliary conditions are defined /evaluated at t = 0 +

Now , The second term in L.H.S of diff. eqn. Q(D)yf(t) = P(D)x(t)

 To find the forced response of the system: yf(t)

The method of undetermined coefficient with following tables

 The method can be used for “class of inputs”, which yields only a
finite number of independent derivatives.
 i.e inputs having eεt tr fall into this category
 E.g. repeated differentiation of eεt is of the same form eεt and
repeated differentiation of tr yields only r independent
variations
 Forced response yf(t), to such input can be expressed as a linear
combination of the input variations and its independent derivatives.
 E.g. If x(t) = a t2 + b ; Its two independent derivatives are :
  2at
 2a
Then the forced response is given by yf(t) = β2t2 + β1t + β0
The undetermined coefficients β2 β1 β0 are evaluated by
substituting this expression for yf(t) in Q(D)yf(t) = P(D)x(t) and
equating the coefficient of similar terms.

Example:

For other class of inputs:

INPUT FORM OUTPUT FORM

1) eεt ε ≠λi(i=1,2,3….n)(characteristic roots) βeεt

2) eεt ε =λi(i=1,2,3….n)(characteristic roots) βteεt

3) K (a constant) β (a constant)

4) Cos(ωt+θ) β Cos(ωt+φ)

5) (tr+αr-1 tr-1+…..+α1 t+ αo) eεt (βr tr+ β r-1 tr-1+…..+ β1 t+ βo) eεt

Case 1) Exponential inputs: x(t) = eεt

Forced response yf(t), to such input can be expressed in the form βeεt

Hence, Q(D)yf(t) = P(D)x(t)

i.e. Q(D) βeεt = P(D) eεt

D βeεt = ε eεt ;

D2 βeεt = ε2 eεt; …. ;
 Dr βeεt = εr eεt; and then,

Q(D) eεt = Q(ε) eεt and P(D) eεt = P(ε) eεt

Substituting these in above equation,

β Q(ε) eεt = P(ε) eεt OR

𝐏(𝛆))
β= = H(ε) Hence,
𝐐(𝛆)

Forced response yf(t) = H(ε) eεt for t ≥ 0 and

Total response y(t) = yn(t) + yf(t)

N

K e
 jt
i.e. y(t) = j  H ( )et
j 1

Case 2) Constant inputs: x(t) = C

It is a special case of exponential eεt with ε = 0, hence,

Forced response yf(t) = C H(ε) eεt with ε = 0 for t ≥ 0
i.e. yf(t) = C H(0)

Case 3) Exponential inputs: x(t) = ejωt

It is a special case of exponential eεt with ε = jω, hence,

Forced response yf(t) = H(jω) ejωt for t ≥ 0

Case 4) Sinusoidal inputs: x(t) = cosωt

We know that, cosωt = ½{ ejωt + e-jωt }

Hence, Forced response,
yf(t) = ½{ H(jω) ejωt + H(-jω) e-jωt } for t ≥ 0
The two terms in the RHS, which are conjugates gets cancel each
other, resulting yf(t) = Re{ H(jω) ejωt }

But, H(jω) = |H(jω)|e jH ( j )

Hence, yf(t) = Re{ |H(jω)| e jH ( j )  jt }
= |H(jω)| cos [ωt+ H ( j ) ]
Hence, for generalised input, x(t) = cos(ωt+θ) ,
yf(t) = |H(jω)| cos [ωt+ θ + H ( j ) ]
 Example:

Finding Total Response: (using Classical Method)

For the same above example, (D2+4D+3)y(t) = (3D+5)x(t) and with initial
conditions : y(0) = -3 ; dy(0)/dt = -7 and with input x(t) = 2t+3,

Calculate, (3D+5)x(t) = 3D(2t+3)+5(2t+3) = 10t+21

>> y=dsolve('D2y+4*Dy+3*y=10*t+21','y(0)=-3','Dy(0)=-
7','t')

O/P -----> y =23/9+10/3*t+143/18*exp(-3*t)-27/2*exp(-t)

Write an inline function to plot the above response

Comment on the Nature, stability, and find the time constant of the system.