Escolar Documentos
Profissional Documentos
Cultura Documentos
Mark Fowler
Note Set #28
C-T Systems: Laplace Transform Solving Differential Equations Reading Assignment: Section 6.4 of Kamen and Heck
1/21
Ch. 2 Convolution
C-T System Model Convolution Integral D-T System Model Convolution Sum
ytotal (t ) = y zs (t ) + y zi (t )
Ch. 2
Ch. 2 Ch. 5
Weve seen how to find this using: convolution w/ impulse response or using multiplication w/ frequency response
Weve seen how to find this using the characteristic equation, its roots, and the socalled characteristic modes
Here well see how to get ytotal(t) using LT get both parts with one tool!!!
3/21
First-order case: Lets see this for a 1st-order Diff. Eq. with a causal input
and a non-zero initial condition just before the causal input is applied. The 1st-order Diff. Eq. describes: a simple RC or RL circuit. The causal input means: we switch on some input at time t = 0. The initial condition means: just before we switch on the input the capacitor has a specified voltage on it (i.e., it holds some charge).
Input: Time-Varying Voltage (e.g., guitar, microphone, etc.) Output: Time-Varying Voltage
switch @ t = 0
R +
+ x(t) C
VIC
+ y(t)
Thus the cap is fully charged to VIC volts
4/21
Assume that for t<0 this has been switched on for a long time
This circuit is then described by this Diff. Eq.: Cap voltage just before x(t) turns on
dy (t ) 1 1 + y (t ) = x (t ) dt RC RC
x(t) = 0, t < 0
For this ex. well solve the general 1st-order Diff. Eq.:
dy (t ) + ay (t ) = bx(t ) dt
Now the key steps in using the LT are: take the LT of both sides of the Differential Equation use the LT properties where appropriate solve the resulting Algebraic Equation for Y(s) find the inverse LT of the resulting Y(s) Laplace Transform: Differential Equation turns into an Algebraic Equation Hard to solve Easy to solve
5/21
Use Linearity of LT Use Property for LT of Derivative accounting for the IC Solve algebraic equation for Y(s)
[sY ( s) y (0 )] + aY ( s) = bX ( s)
y (0 ) b Y ( s) = + X ( s) s+a s+a
Part of soln driven by IC Zero-Input Soln Part of soln driven by input Zero-State Soln
Note that 1/(s+a) plays a role in both parts Hey! s+a is the Characteristic Polynomial!!
Example: RC Circuit Now we apply these general ideas to solving for the output of the previous RC circuit with a unit step input. x (t ) = u (t )
dy (t ) 1 1 + y (t ) = x(t ) dt RC RC
y (0 ) 1 / RC Y (s) = + X (s) s + 1 / RC s + 1 / RC
This transfers the input X(s) to the output Y(s) Well see this later as The Transfer Function Now we need the LT of the input From the LT table we have:
x (t ) = u(t )
X ( s) =
1 s
1 / RC 1 y (0 ) Y ( s) = + s + 1 / RC ( s + 1 / RC ) s
Now we have just a function of s to which we apply the ILT
7/21
y (0 ) 1 / RC -1 -1 Y L { ( s )} = L + s + 1 / RC ( s + 1 / RC ) s y (0 ) 1 / RC -1 -1 y (t ) = L + L s + 1 / RC ( s + 1 / RC ) s
This part (zero-input soln) is easy Just look it up on the LT Table!!
Linearity of LT
This part (zero-state soln) is harder It is NOT on the LT Table!! So the part of the soln due to the IC (zeroinput soln) decays down from the IC voltage
8/21
y (0 ) ( t / RC ) u (t ) L-1 = y (0 )e s + 1 / RC
y (0 ) e t / RC u(t ) t
Now lets find the other part of the solution the zero-state soln the part that is driven by the input:
y (0 ) 1 / RC -1 -1 y (t ) = L + L s + 1 / RC ( s + 1 / RC ) s
We can factor this function of s as follows:
1 / RC 1 1 L = L-1 s s + 1 / RC ( s + 1 / RC ) s
-1 -1
Can do this with Partial Fraction Expansion, which is just a fool-proof way to factor
1 1 -1 = L L s s + 1 / RC
Linearity of LT
= u(t )
= e ( t / RC )u(t )
= 1 e ( t / RC ) u(t )
9/21
[1 e
( t / RC )
]u(t )
[1 e
1
t / RC
] u (t )
t
Now putting this zero-state response together with the zero-input response we found gives:
y (t ) = y (0 )e
( t / RC )
u(t ) + [1 e
( t / RC )
] u (t )
IC Part
Input Part
Notice that: The IC Part Decays Away but The Input Part Persists
10/21
Here is an example for RC = 0.5 sec and the initial VIC = 5 volts:
Zero-Input Response
Zero-State Response
Total Response
11/21
Second-order case
Circuits with two energy-storing devices (C & L, or 2 Cs or 2 Ls) are described by a second-order Differential Equation
dx(t ) d 2 y (t ) dy (t ) + a1 + a0 y (t ) = b1 + b0 x(t ) 2 dt dt dt
Assume Causal Input w/ ICs
y (0 ) & y (0 )
x(t ) = 0 t < 0
x (0 ) = 0
We solve the 2nd-order case using the same steps: Take LT of Diff. Equation:
2
Note: The role the Characteristic Equation plays here! It just pops up in the LT method! The same happened for a 1st-order Diff. Eq and it happens for all orders Like before to get the solution in the time domain find the Inverse LT of Y(s)
13/21
To get a feel for this lets look at the zero-input solution for a 2nd-order system:
which has either a 1st-order or 0th-order polynomial in the numerator and a 2nd-order polynomial in the denominator For such scenarios there are Two LT Pairs that are Helpful:
Ae nt sin n 1 2 t u(t ) where : A=
[(
)]
n 1 2
2 s 2 + 2n s + n
[(
) ]
( n )2 + 1 A= 2 2
n (1 )
n 1 2 = tan n
1
s + 2 2 s + 2n s + n
Ae nt sin n 1 2 t u (t )
This form gives yzi(0) = 0 as set by the IC
[(
)]
2 s 2 + 2n s + n
If y(0-) = 0
Ae nt sin n 1 2 t + u(t )
[(
) ]
s + 2 s 2 + 2n s + n
Otherwise
15/21
y (0 ) = 4
Pulled a 2 out from each term in Num. to get form just like in LT Pair.
Then
Yzi ( s ) =
s + (2 + a1 ) 2 s + (4 + a1 2 ) = 2 2 s 2 + a1s + a0 s + a1s + a0
a0 = 100 & a1 =4
Compare to LT:
s + 2 2 s + 2n s + n
=6 =2
And identify:
2 n = 100 n = 10
2n = 4 = 4 / 2n = 4 / 20 = 0.2
16/21
=6 =2 n = 10 = 0.2
Ae nt sin n 1 2 t + u(t ) where :
( n )2 A= 2( 2
n (1
1
) +1 = 2
[(
) ]
( n )2 A= 2( 2
n (1
1
n 1 2 = tan n
) +1
n 1 2 = tan n
Notice that the zero-input solution for this 2nd-order system oscillates 1st-order systems cant oscillate 2nd- and higher-order systems can oscillate but might not!!
17/21
Zoom In
4 of + pe Slo
y (0 ) = 2
y (0 ) = 4
18/21
Nth-Order Case
Diff. eq N d N 1 y (t ) dy (t ) dx M (t ) dx(t ) d y (t ) of the + a N 1 + ... + a1 + a0 y (t ) = bM + b1 + b0 x(t ) dt dt N dt N 1 dt dt M system
i = 0,1, 2, ..., M 1
Y ( s) =
where
IC ( s ) B( s ) + X ( s) A( s ) A( s )
A( s ) = s N + a N 1s N 1 + ... + a1s + a0 output-side polynomial M B ( s ) = bM s + ... + b1s + b0 input-side polynomial IC ( s ) = polynomial in s that depends on the ICs
]
19/21
N X ( s) DX ( s )
Y ( s) =
IC ( s ) B( s ) IC ( s ) B( s ) N X ( s ) X ( s) = + + A( s ) A( s ) A( s ) A( s ) DX ( s ) IC ( s ) E ( s ) F ( s ) + + A( s ) A( s ) DX ( s )
Zero-Input Response
IC ( s ) E ( s ) F ( s ) + + Y ( s) = A( s ) A( s ) DX ( s )
Decays in time domain if roots of system char. poly. A(s) have negative real parts Transient Steady-State Response Response
20/21
B( s) Y (s) = X (s) A( s )
H (s )
Zero-State Response
E ( s) F ( s) + Y ( s) = A( s ) DX ( s )
Transient Steady-State Response Response
21/21
Summary Comments: 1. From the differential equation one can easily write the H(s) by inspection! 2. The denominator of H(s) is the characteristic equation of the differential equation. 3.The roots of the denominator of H(s) determine the form of the solution recall partial fraction expansions
BIG PICTURE: The roots of the characteristic equation drive the nature of the system response we can now see that via the LT. We now see that there are three contributions to a systems response: 1. The part driven by the ICs a. This will decay away if the Ch. Eq. roots have negative real parts 2. A part driven by the input that will decay away if the Ch. Eq. roots have negative real parts Transient Response 3. A part driven by the input that will persist while the input persists Steady State Response
22/21