t5 PDF
t5 PDF
This tutorial is a continuation of tutorial 4 and is of interest to any student studying control
systems and in particular the EC module D227 – Control System Engineering.
• Examine the various ways of writing the standard second order transfer function.
• Define the roots and poles of the standard second order transfer function.
• Explain and calculate the response of a standard 2nd order system to a step change in
the input.
• Explain and define design parameters of the second order step response.
• Explain and calculate the response of a standard 2nd order system to a ramp
(velocity) change in the input.
• Explain the response of a standard 2nd order system to a sinusoidal change in the
input.
• Explain the relationship of the complex pole position with the damping ration and
natural frequency.
• Explain the root locus diagram for the complex poles with respect to damping ratio.
If you are not familiar with instrumentation used in control engineering, you should
complete the tutorials on Instrumentation Systems.
In order to complete the theoretical part of this tutorial, you must be familiar with basic
mechanical and electrical science.
You must also be familiar with the use of transfer functions and the Laplace Transform
(see maths tutorials).
© D.J.DUNN 1
1. STANDARD 2ND ORDER SYSTEM
The standard second order transfer function can be expressed in many ways. Here are the most useful.
θo k
In terms of the time constant T (s ) = 2 2
θi T s + 2 T δ s +1
θo ω 2n k
In terms of the natural frequency ωn (s ) = 2
θi s + 2 δ ω n s + ω 2n
θo ω 2n k
In terms of the poles (polynomial) (s ) =
θi (s - p1 )(s − p 2 )
θo ω 2n k ω 2r + σω 2r
In terms of two parameters σ and ωr (s ) = =
θi (s + σ) 2 + ω 2r (s + σ) 2 + ω 2r
Remember that δ is the damping ratio (ξ is often used in the exam and other texts)
If you are happy with this you should skip the following proofs.
Since the solution of the quadratic is the same for both, it follows that x and y are the same and must be
the roots of the denominator.
− b m b 2 − 4ac
Solve the quadratic equation y= a = 1 b = -2δωn c = ωn2 hence:
2a
2 δ ωn m (2 δ ω n )2 − 4ω 2n
y=x= = δ ωn m ( δ ω n )2 − ω 2n
2
y = x = δω n m ω n δ 2 − 1 = δω n m jω n 1 - δ 2
θo ω 2n k
(s ) =
θi ⎧s + δω + jω 1 - δ 2 ⎫⎧s + δω − jω 1 - δ 2 ⎫
⎨ n n ⎬⎨ n n ⎬
⎩ ⎭⎩ ⎭
© D.J.DUNN 2
σ AND ωr FORM
This is simply the substitution σ = δω n and ω r = ω n 1 - δ 2 in the last equation.
ωr is the damped resonant frequency at which the output value peaks. σ (sigma) has no name.
θ ω 2n k
The transfer function is o (s ) = and this further simplifies to
θi {s + σ + jω r }{s + σ - jω r }
θo ω 2n k
(s ) = (Try multiplying out to see it is the same)
θi (s + σ) 2 + ω 2r
This form is often manipulated further to make it a recognisable form for Inverse Laplace transforms.
Using the definitions σ = δωn and ωr = ωn√(1-δ2)
We can arrive at ωr2 = ωn2(1-δ2) ωr2= ωn2- δ2ωn2 ωr2 = ωn2 - σ2 ωn2= ωr2 + σ2ωr2
θo ω 2r + σω 2r
(s ) =
θi (s + σ) 2 + ω 2r
POLES
Poles are covered more fully in a later tutorial. They are the values of s that make the denominator zero.
The denominator is a quadratic equation in s. Note that in general, the solution of s that satisfies the
equation (s2 + 2δωns + ωn2) = k are the roots of the equation. The poles are the roots when k = 0. To find
the poles we must solve (s2 + 2δωns + ωn2) = 0. Using the quadratic equation the solution is:
− b m b 2 − 4ac
s= a = 1 b = 2δωn c = ωn2 hence:
2a
− 2 δ ωn m (2 δ ω n )2 − 4ω 2n
s= = − δ ωn m ( δ ω n )2 − ω 2n
2
s = −δω n m ω n δ 2 − 1 s = −δω n m jω n 1 - δ 2
© D.J.DUNN 3
1.2 RESPONSE TO A STEP INPUT
The standard 2nd order transfer function represents many real systems. The derivation of this equation for
some real systems is covered in tutorial 1. Finding the output time response for this type of system is
much more difficult than anything covered so far. The simplest case is when δ = 0. Let’s use a worked
example to examine this case.
Find the time response of a standard second order system to a step input when δ = 0 (undamped).
SOLUTION
In this example δ is made zero in order to simplify the solution but it should be remembered that this
is a special case. Starting with the standard transfer function we have
θo 1
= 2 2
θ i T s + 2 δ Ts + 1
θ
Put δ = 0 and make θo the subject. θ o = 2 2i
T s +1
H/s H
For a step input θi = H and θi (s) = H/s θ o (s) = 2 2 =
T s +1 sT s + 1
2 2
( )
2
H/T
θ o (s) =
Rearrange into a form recognised in the table.
(
s s 2 + 1/T 2 )
If we substitute 1/T = ω the following I recognised in the table.
Hω 2
θ o (s) = 2
(
s s + ω2 )
From the table of transforms we find that the inverse transform which converts the output into a
function of time gives the result
θo (t)= H(1 – cos ωt )
This means that the output is a negative cosine curve as shown. The amplitude is H and the frequency
of oscillation is ω = 1/T rad/s.
Figure 1
Remember that this is a special case and the 2nd order equation will be examined in greater detail
later. If the system was a mass on a spring, the mass would oscillate up and down. If the system was
an electrical L – C circuit, the current would oscillate back and forth.
© D.J.DUNN 4
Now consider the time response of a standard second order system with damping to a unit step input. We
θ ω 2n
will use the standard form o (s ) =
θi (s + σ) 2 + ω 2r
1 ω 2r + σω 2r
For a unit step θi(s) = 1/s hence θ o (s ) =
s (s + σ) 2 + ω 2r
σ
From the table of Laplace transforms, the inverse is θ o (t) = 1 − e − σt (cos ω r t +
sin ω r t)
ωr
This may be plotted for different values of δ to produce a normalised graph that may be used for solving
specific problems. The settling level is 1.0. ωn is the natural angular frequency of the system. The natural
frequency is fn = ωn /2π. This is the frequency at which the system would oscillate if there was no
damping.
Figure 2
A damped system with a natural frequency of 4 rad/s has a damping ratio δ = 0.4. Find the time taken
from the start of a step change to when the output overshoots by 20% for the first time.
SOLUTION
We are looking for a value of θo/H = 120% or 1.2 on the vertical scale and where this intersects with
the graph for δ = 0.4.
This gives us a value of ωn t = 2.8 on the horizontal scale. The time taken is hence t = 2.8/4 = 0.7 s.
Note that unfortunately the normalised response graph is never given in the EC Exam and other
methods are needed to solve problems like this.
© D.J.DUNN 5
A fuller analysis of the step response is given in a later tutorial. The following is useful to know at this
stage. It was shown earlier through use of the polynomial form that the step response is :
σ
θ o (t) = 1 − e − σt (cos ω r t + sin ω r t) where σ = δω n and ω r = ω n 1 - δ 2
ωr
σ
Since it is a unit step the amplitude is θ A (t) = −e −σt (cos ω r t + sin ω r t) . The diagram compares the
ωr
response and amplitude plot for the unit step.
Figure 3
⎧ σ ⎫
− e −σ(t ) ⎨cos ω r (t ) + sin ω r (t) ⎬
θ1 ⎩ ωr ⎭
=
θ2 − σ(t + Tp ) ⎧ σ ⎫
−e ⎨cos ω r (t + Tp ) + sin ω r (t + Tp ) ⎬
⎩ ωr ⎭
⎧ σ ⎫
⎨cos ω r (t ) + sin ω r (t) ⎬
⎩ ωr ⎭
It may be shown that =1
⎧ σ ⎫
⎨cos ω r (t + Tp ) + sin ω r (t + Tp ) ⎬
⎩ ωr ⎭
θ1 − e − σt θ1 σ(t + Tp ) σT
So the amplitude ratio of two successive oscillations is = −σ(t + Tp )
= e −σt e = e p Hence
θ2 − e θ2
⎛θ ⎞
ln⎜⎜ 1 ⎟⎟ = σTp This expression is usually called the logarithmic decrement. Substituting for σ and Tp
⎝ θ2 ⎠
⎛ θ ⎞ 2π δ
ln⎜⎜ 1 ⎟⎟ = This is also called the amplitude reduction factor.
θ
⎝ ⎠ 2 1- δ 2
© D.J.DUNN 6
WORKED EXAMPLE No.3
A recording of a damped oscillation is made and it is determined that the periodic time is 0.2 s and
that the ratio of two successive amplitudes is 20 mm and 5 mm. Calculate the natural frequency and
the damping ratio.
SOLUTION
⎛ θ ⎞ 2π δ ⎛ 20 ⎞
ln⎜⎜ 1 ⎟⎟ = = ln⎜ ⎟ = 1.386
⎝ θ2 ⎠ 1- δ2 ⎝ 5 ⎠
(2 π δ )2 = 1.386 2 = 1.9218
2
1- δ
( )
(2 π δ )2 = 1.9218 1 - δ 2 = 1.9218 - 1.9218 δ 2
39.478 δ = 1.9218 - 1.9218 δ 2
2
41.4δ2 = 1.9218
δ = 0.215
fr = 1/Tp = 1/0.2 = 5 Hz ωr = 2π fr = 10π rad/s
ωr = ωn 1 - δ 2
ωr 10π
ωn = = = 32.17 rad/s
1- δ 2 0.9765
© D.J.DUNN 7
1.3 RAMP INPUT
θo 1
The standard transfer function is (s) = 2 2
θi T s + 2 δ Ts + 1
c
The input is a ramp so θi(s) = c/s2 Substitute and θ o (s) =
s (T s 2 2 2
+ 2 δ Ts + 1)
The inverse transform using the computer software Mathcad™ is:
The plot of this function shows that the output is oscillatory and the level of oscillation depends on the
damping ratio δ. When δ =1, the result is similar to that of the first order system.
Figure 4
© D.J.DUNN 8
1.4 SINUSOIDAL INPUT
θo K
The standard transfer function is (s) = 2 2 K is the gain and δ is the damping ratio.
θi T s + 2 δ Ts + 1
The input is sinusoidal so θi(t) = A sin(ωt) A is the amplitude
ω
The Laplace transform is θ i (s) = 2
ω + s2
K Aω
Substitute and θ o (s) =
(T s + 2 δ Ts + 1) ω + s 2
2 2 2
The inverse transform using the computer software Mathcad™ is a very long complicated expression so it
is not reproduced here but plotting the results with ω <1/T yields the following.
Figure5
During the transient stage, the output grows but settles down to constant amplitude out of phase with the
input. If δ = 1 (critical damping) it reaches the steady state condition quicker.
If the plot is repeated with a higher frequency such that ω=1/T we reach a condition called resonance and
the plot is as shown.
Figure 6
The lightly damped response grows rapidly to a much bigger amplitude than the input and the phase shift
is ¼ cycle. If the damping is very small, the output amplitude would be enormous. If we raise the
frequency further so that ω > 1/T the result is very different as shown.
Figure 7
The output is much reduced and the phase shift is almost a half cycle.
Clearly to understand this fully we need to examine the problem from a different angle and this is covered
in the next tutorial.
© D.J.DUNN 9
SELF ASSESSMENT EXERCISE No.1
1. The voltage Vi in the L – C - R circuit shown is suddenly changed from 0 to 10 V. Calculate the
natural frequency of oscillation and the critical value of R that will just stop oscillation. You should
use the following information from tutorial 1.
V 1
The transfer function for the system is G(s) = o (s) = 2 2
Vi T s + 2 δ Ts + 1
R L
The damping ratio δ = The critical damping resistance is Rc = 4 and T = √CL
L C
4
C
C=20mF L = 5 µH R=0.020 Ω
Figure 8
(Answers δ = 0.632, Rc = 0.032Ω and fn = 503.3 Hz)
2. Using the answers from Q1, determine the time taken for the voltage to reach 10V for the first time.
(ωn =3163 rad/s produces t = 0.47 ms)
3. A single acting pneumatic cylinder has to push a mass M kg as shown. The cylinder is returned by a
spring with stiffness K N/m. There is damping of kd N s/m. The air pressure is p N/m2 and the piston
area is A m2. The position of the mass is x metres.
x A/M
(s) = 2
Show that the transfer function for this system is
p (
s + 2 δ ωn s + ωn 2 )
ωn is the natural frequency of the system (K/M)½
δ is the damping ratio. δ = kd/(4MK)½
Figure 9
Given M = 50 kg, kd = 80 N s/m, K = 2000 N/m and A = 0.2 m2, determine the natural frequency
and the damping ratio. (6.325 rad/s or 1.007 Hz and δ = 0.126)
Determine the natural gain and the time constant. (1 x 10-4 and 0.158 s)
A step change is made in p. Using the normalised response graph, find the time when the output
reaches the new value for the first time. (0.277s approx)
© D.J.DUNN 10
4. A standard second order system is subjected to a unit step disturbance and a recording is made of the
time – output response. It is found that ratio of two successive amplitudes is 3.6 and the periodic time
is 0.641 s. Determine the damping ratio and the natural frequency of the system.
(0.2 and 10 rad/s
As with first order system, the D.C. gain is the magnitude of the transfer function when it contains no
dynamic terms so we simply put s = 0.
θo K
(s) = 2 2
θi T s + 2 δ Ts + 1
When the last figure on the bottom line is plus 1 the system has negative feed back and K is the D.C. gain
of the system.
A system has a transfer function G(s) = 1/ (4s2 + 8s + 3). Find the D.C.gain.
SOLUTION
G(s) = G(s) = 1/ (4s2 + 8s + 3) put s = 0 and the gain is 1/3 = 0.333 which in fact is attenuation.
1. Find the d.c. gain of the following closed loop transfer functions.
1000
G(s) = (5)
2s 2 + 50s + 200
4
G(s) = (10)
0.05s + 0.2s + 0.4
2
500(s + 2)
G(s) = 2 (1000)
s + 10s + 1
© D.J.DUNN 11
1.6 STEADY STATE ERROR
The steady state error produced by a system may be found by applying the FINAL VALUE THEOREM
(which is not explained here). This states that the final value of a time function G(t) as t→0 is given by
the value of s G(s) as s→ 0.
The steady state error of a system may be found by applying the theorem to the transfer function
representing the error.
Consider a system with a closed loop transfer function G(s)cl. The error is:-
θ e = θi − θ o θ o = G clθ i θ e = θ i − G clθ i = θ i [1 − G cl ]
Steady State Error = θ e (t) = sθ e (s) = s θ i (s)[1 − G cl ] as s→0
Consider a system with an open loop transfer function G(s)ol and unit feedback. The closed loop transfer
function is:
θ G ol
G(s) cl = o =
θ i 1 + G ol
sθ (s)
Steady State Error = θ e (t) = sθ e (s) = i as s→0
1 + G ol
Find the steady state error when a unit step input is applied to a system with an open loop transfer
1
function of G(s) = with unit feedback.
(s + 4)
SOLUTION
sθ (s) 1 1
θ e (t) (steady state) = i For a step input, θ i (s) = so θ e (t) =
1+ G s 1+ G
1 1
θ e (t) = and as s → 0 this becomes = = 0.8
⎡ 1 ⎤ 1 + 1/4
1+ ⎢ ⎥
⎣s + 4 ⎦
© D.J.DUNN 12
WORKED EXAMPLE No.8
Find the steady state error when a unit step input is applied to a system with a closed loop transfer
3
function of G(s) = 2
(s + 2s + 1)
SOLUTION
θ e (t) (steady state) = s θ i (s)[1 - G cl ] For a step input, θ i (s) = so θ e (t) = 1 − 2
1 3
as s → 0
s (s + 2s + 1)
3
This becomes = 1 − = -2
0 + 0 +1
This is logical since with a gain of the steady state output is 3 and the error is -2.
Find the steady state error when a unit ramp input is applied to a system with an open loop transfer
3
function of G(s) =
s(s + 1)(s + 2 )
SOLUTION
⎡ 1 ⎤ 1 1 ⎡ 1 ⎤
θ e (t) (steady state) = s θ i (s) ⎢ ⎥ For a unit ramp input, θ i (s) = 2 so θ e (t) = s x 2 ⎢ ⎥
⎣1 + G ol ⎦ s s ⎣1 + G ol ⎦
⎡ ⎤ ⎡ ⎤
1 ⎢ 1 ⎥ ⎢ 1 ⎥ 1
θ e (t) = ⎢ ⎥=⎢ ⎥ as s → 0 θ e (t) = = 0.5
s⎢ 3 ⎥ ⎢ 3 ⎥ 0+2
1+ s+
⎢⎣ s(s + 1)(s + 2) ⎥⎦ ⎢⎣ (s + 1)(s + 2) ⎥⎦
SOLUTION
⎡ 1 ⎤ 1 0.2
θ e (t) (steady state) = s θ i (s) ⎢ ⎥ θ i (s) = + 2
⎣1 + G ol ⎦ s s
⎛ 1 0.2 ⎞ ⎡ 1 ⎤ ⎛ 0.2 ⎞ ⎡ 1 ⎤
θ e (t) = s ⎜ + 2 ⎟ ⎢ ⎥ = ⎜1 + ⎟⎢ ⎥
⎝ s s ⎠ ⎣1 + G ol ⎦ ⎝ s ⎠ ⎣1 + G ol ⎦
⎡ ⎤ ⎡ ⎤ ⎡ ⎤
⎛ 0.2 ⎞ ⎢ 1 ⎥ ⎢ s + 0.2 ⎥ ⎢ s + 0.2 ⎥
θ e (t) = ⎜1 + ⎟⎢ ⎥=⎢ ⎥=⎢ ⎥ as s → 0 we may write
⎝ s ⎠⎢
1+
2 ⎥ ⎢ s+
2 ⎥ ⎢ s+
2 ⎥
⎢⎣ s(s + 1)(s + 2 ) ⎥⎦ ⎢⎣ (s + 1)(s + 2 ) ⎥⎦ ⎢⎣ (s + 1)(s + 2 ) ⎥⎦
⎡ 0.2 ⎤
θ e (t) = ⎢ ⎥ = 0.2
⎣ 1 ⎦
© D.J.DUNN 13
SELF ASSESSMENT EXERCISE No.3
1. Find the steady state error when a unit step is applied to a system with the following closed loop
transfer functions.
4
G(s) = (-9)
0.05s 2
+ 0.2s + 0.4
500
G(s) = (-4)
s2 + 10s + 100
2. Find the steady state error when a unit step is applied to a system with the following open loop
transfer functions with unit feedback.
1
G(s) = (0)
s(s + 1)(s + 4 )
5
G(s) = (4/9)
(s + 1)(s + 4)
3. Find the steady state error when a unit ramp is applied to a system with the following open loop
transfer functions with unit feedback.
5
G(s) = (∞)
(s + 1)(s + 4)
1
G(s) = (4)
s(s + 1)(s + 4 )
4. Find the steady state error when a system with the following open loop transfer function and unit
feedback has an input θi(t) = 1+ 0.5t
5
G(s) = (0.3)
s(s + 1)(s + 3)
© D.J.DUNN 14
2. A DETAILED LOOK AT THE POLES OF THE SECOND ORDER SYSTEM
This section introduces you to the concept of root locus plots that are covered more fully in later tutorials.
This part only applies to a standard second order system.
If we plot the poles for a range of values of δ we get the diagram shown. This is the root locus plot with
respect to δ and it takes the form of a circle of radius ωn.
Figure 10
Consider the radius radius = √(σ2 + ωr2) = √[(δ ωn)2 + ωn2 (1-δ2)] = ωn
It follows that the position of the poles depends upon the value of δ and ωn. Note that when dealing with
higher order systems in later tutorials, the complex pole can be represented in terms damping ratio δ and
natural frequency ωn.
The smaller the value of δ, the more oscillatory the response to a step input and the closer the resonant
frequency is to the natural frequency. This can be summed up as follows. Consider the time response to a
unit step input. The solution was shown earlier as
σ
θ o (t) = 1 − e − at (cos ω r t + sin ω r t)
ωr
© D.J.DUNN 15
Note that positive values of damping are stable and negative values are unstable. The diagram shows the
unit step response of the system for various pole positions. This shows that greatest stability is when the
poles are on the negative real axis and maximum instability is on the positive real axis. The response is
more oscillatory when the poles have a large imaginary (j) term. This is an important point that is used in
later tutorials where the choice of pole positions is considered in general.
Figure 11
A standard second order system has a natural frequency of 120 rad/s and a damping ratio of 0.4.
Determine the position of the poles on the s plane.
SOLUTION
A standard second order system has a natural frequency of 50 rad/s and a damping ratio of 0.3. Determine
the position of the poles on the s plane.
© D.J.DUNN 16