McMaster - Intro To Linear Control Systems

EE 3CL4, L 1 1 / 21 Tim Davidson Why are you here? What is a control system? What tools will we use?
Administrative details Parting shot
EE3CL4: Introduction to Linear Control Systems

Lecture 1 Tim Davidson
McMaster University
Winter 2011
EE 3CL4, L 1 2 / 21 Tim Davidson Why are you here? What is a control system? What tools will we use? Administrative details Parting shot
Outline
1 Why are you here? 2 What is a control system? 3 What tools will we use? 4 Administrative details 5 Parting shot
Why are you here?

You might be interested in:
Designing guidance systems for: the next Mars Rover, a space craft or a UAV
Why are you here?

Designing industrial or biomedical robots
Why are you here?

Designing humanoid robots
Why are you here?

You might be interested in: Designing control systems for hybrid or electric cars
Designing disk drives
Developing technologies for enhanced power
distribution (Smart Grid)
Why are you here?

Designing an automated insulin delivery system
Why are you here?

Designing and analyzing nancial systems
What is a feedback control system?
What is a feedback control system?

Example: You driving a car
What tools will we use?

Newtonian mechanics, linear and rotational
(Phys 1D03)
Basic electromagnetism (Phys 1E03, EE 2CJ4) Electric circuit analysis (EE 2CI5, EE 2CJ4, EE 2EI5) Step response of rst and second order systems
(Math 2P04/2Z03, EE 2CI5, EE 2CJ4)

Laplace transforms
(Math 2P04/2Z03, EE 2CJ4, EE 3TP4)

Transfer functions (EE 2CJ4, EE 3TP4) Bode diagrams (EE 2CJ4) Structured problem solving methods
(EE 2CI5, EE 2CJ4, EE 3TP4, . . . )
Contact details
Tim Davidson
ITBA310 Ext. 27352 davidson@mcmaster.ca, EE3CL4 in subject line

Course web site (soon to be fully updated for 2011)
http://www.ece.mcmaster.ca/davidson/EE3CL4
A formal course outline appears on the web site.
Here we will focus on some key points
Class details
Lectures Tuesday, Thursday, Friday, 12:30pm, TSH/B128 Tentative topic schedule will appear on web site Tutorials (starting next week) T01: Tuesday, 8:30am, T13/125 T02: Monday, 11:30am, BSB/106 Labs (tentatively starting 24 January) Four labs One every other week, ITB/154 Signicant pre-lab work will be required
Marking scheme
Laboratory reports: 20% Midterm test: 25%
Tentatively scheduled for week starting Monday 28 Feb (rst week after midterm break), 7:00pm 8:30pm
Final examination: 55% Students must personally complete all laboratories and
all laboratory reports in order to be eligible for a nal grade

Formally deferred tests & exams may be conducted
orally
Remarking requests will require documentation On tests & exams, expect to see problems that you
have not seen before
Some suggestions
Be active in lectures Participate in tutorials Take advantage of the labs Do half of the assigned problems under examination
conditions
In exams, explain your methodology
Parting shot
EE 3CL4, L 2 1 / 19 Tim Davidson What is control engineering Examples Design process Disk drive example

McMaster University
Winter 2011
Outline
1 What is control engineering
2 Examples
3 Design process
4 Disk drive example
What is control engineering

First we have a system that we want to understand Typically, that means a mathematical model Then we use that understanding to design a secondary
system that controls the behaviour of the rst

Must obtain appropriate control, even if model
inaccurate, or subject to noise/disturbances
Mathematical model Must balance accuracy against insight generated This course: models will be linear Hence, tools available for insight: superposition, transfer function, Laplace
Model, open-loop, closed-loop
What about disturbances? Mismatch (temperature, age)
Multivariable control
Something for fourth year!
Watts yball governor
CD player speed control: Open-Loop
CD player speed control: Closed-Loop
Doritos
John MacGregor (Chem Eng): Visual feedback control of avours
Structured approach to design
Disk drive: Intro
Spins at between 1800 and 7200 rpm,
perhaps even 10,000

Head height: 100nm A key issue: Seek time
Disk drive: Establish goals
Move from track a to track b

with accuracy of at least1m within 50ms
Disk drive: Control architecture
EE 3CL4, L 3 1 / 20 Tim Davidson Modelling physical systems

Translational Newtonian Mechanics Rotational Newtonian Mechanics

McMaster University
Linearization Laplace transforms
Winter 2011

Outline
1 Modelling physical systems
2 Linearization
3 Laplace transforms

Differential equation models
Most of the systems that we will deal with are dynamic Differential equations provide a powerful way to
describe dynamic systems

Will form the basis of our models We saw differential equations for inductors and
capacitors in 2CI, 2CJ

What about mechanical systems?
both translational and rotational

Translational Spring
F (t ): resultant force in direction x Recall free body diagrams and action and reaction
Spring. k : spring constant, xr : relaxed length of spring
F (t ) = k [x2 (t ) x1 (t )] xr

Translational Damper
F (t ): resultant force in direction x
Viscous damper. b : viscous friction coefcient
F (t ) = b
dx2 (t ) dx1 (t ) = b v2 (t ) v1 (t ) dt dt

Mass
F (t ): resultant force in direction x
Mass: M
F (t ) = M
d 2 xm (t ) dvm (t ) =M = Mam (t ) dt dt 2

Rotational spring
T (t ): resultant torque in direction
Rotational spring. k : rotational spring constant,
r : rotation of relaxed spring
T (t ) = k [2 (t ) 1 (t )] r

Rotational damper
Rotational viscous damper.
b: rotational viscous friction coefcient
T (t ) = b
d 2 (t ) d 1 (t ) = b 2 (t ) 1 (t ) dt dt

Rotational inertia
Rotational inertia: J
T (t ) = J
d 2 m ( t ) d m ( t ) =J = J m (t ) 2 dt dt

Example system (translational)

Horizontal. Origin for y : y = 0 when spring relaxed
(t ) F = M dv dt (t ) v (t ) = dy dt (t ) F (t ) = r (t ) b dy dt ky (t )
d 2 y (t ) dy (t ) +b + ky (t ) = r (t ) dt dt

Example, continued
Resembles equation for parallel RLC circuit.

Example, continued
Stretch the spring a little and hold. Assume an under-damped system. What happens when we let it go?

Taylors series
Nature does not have many linear systems However, many systems behave approximately linearly
in the neighbourhood of a given point

Apply rst-order Taylors Series at a given point Obtain a locally linear model

Pendulum example
Torque due to gravity: T = MgL sin Linearize around = 0. At that point, T = 0 Linearized model
T MgL
d sin d
=0
= MgL

Laplace transform
Once we have a linearized differential equation we can
take Laplace Transforms to obtain the transfer function

We will consider the one-sided Laplace transform, for
signals that are zero to the left of the origin.
F (s) =
0
f (t )est dt
What does
mean? limT
Does this limit exist? If |f (t )| < Met , then exists for all Re(s ) > .
Includes all physically realizable signals

Note: When multiplying transfer function by Laplace of input, output
is only valid for values of s in intersection of regions of convergence

Laplace transform pairs
can be tabulated

Laplace transform pairs
Recall that complex poles come in conjugate pairs.
EE 3CL4, L 4 1 / 29 Tim Davidson Laplace transforms Laplace transforms in action

McMaster University
Winter 2011
Outline
1 Laplace transforms
2 Laplace transforms in action
Laplace transform
We will consider the one-sided Laplace transform, for
signals that are zero to the left of the origin.
F (s) =
0
f (t )est dt
Key properties
df (t ) sF (s) f (0 ) dt F (s ) 1 f (x ) dx + s s
t 0
f (x ) dx
Final value theorem
If F (s ) has all its poles in the left half plane, except,
perhaps, for a single pole at the origin,

then we can compute the nal value of f (t ), namely
limt f (t ) without inverting the Laplace Transform.

In particular,
t
lim f (t ) = lim sF (s)

s0
Common application: Steady state value of step
response
Mass-spring-damper system
Horizontal (no gravity) Set origin of y where spring is relaxed

(t ) F = M dv dt (t ) v (t ) = dy dt (t ) F (t ) = r (t ) b dy dt ky (t )
MSD system
M d 2 y (t ) dy (t ) +b + ky (t ) = r (t ) dt dt
Consider t 0 and take Laplace transform M s2 Y (s)sy (0 ) dy (t ) dt +b sY (s)y (0 ) +kY (s) = R (s)
t =0
Hence Y (s ) = 1/M R (s) s2 + (b/M )s + k /M (s + b/M ) + 2 y (0 ) s + (b/M )s + k /M 1 dy (t ) + 2 s + (b/M )s + k /M dt
t =0
Response to static init. cond.
Spring stretched to a point y0 , held, then let go at time t = 0 Hence, r (t ) = 0 and Hence, Y (s ) = s2 (s + b/M ) y0 + (b/M )s + k /M
dy (t ) dt t =0
=0
What can we learn about this response without having to invert Y (s)
Standard form
(s + b/M ) y0 + (b/M )s + k /M (s + 2n ) y 2 0 + 2n s + n
b 2 kM
Y (s ) =
s2
= where n =
s2
k /M and =
Poles: s1 , s2 = n n
2 1
> 1 (equiv. b > 2 kM ): distinct real roots, overdamped = 1 (equiv. b = 2 kM ): equal real roots, critically damped < 1 (equiv. b < 2 kM ): complex conj. roots, underdamped
Overdamped case
s1 , s2 = n n y (t ) = c1 es1 t + c2 es2 t y (0) = y0 = c1 + c2 = y0
(t ) dy dt t =0
2 1
Overdamped response: > 1 (equiv. b > 2 kM )
= 0 = s1 c1 + s2 c2 = 0
What does this look like when strongly overdamped s2 is large and negative, s1 is small and negative Hence es2 t decays much faster than es1 t Also, c2 = c1 s1 /s2 . Hence, small Hence y (t ) c1 es1 t Looks like a rst order system!
Critically damped case
s1 = s2 = n y (t ) = c1 en t + c2 ten t y (0) = y0 = c1 = y0
(t ) dy dt t =0
= 0 = c1 n + c2 = 0
Underdamped case
s1 , s2 = n j n 1 2 Therefore, |si | = n : poles lies on a circle Angle to negative real axis is cos1 ( ).
Underdamped case
Dene = n , d = n
1 2 . Response is:
y (t ) = c1 et cos(d t ) + c2 et sin(d t ) = Aet cos(d t + ) Homework: Relate A and to c1 and c2 . Homework: Write the initial conditions y (0) = y0 and
dy (t ) dt t =0
= 0 in terms of c1 and c2 , and in terms of A and
Numerical examples
Y (s) =
(s+2n ) 2 s2 +2n s+n
y0 , where n = 2 1
k /M , =
b 2 kM
Poles: s1 , s2 = n n
> 1: overdamped; < 1: underdamped Consider the case of M = 1, k = 1. Hence, n = 1, b = 3 0. Hence, = 1.5 0 Initial conds: y0 = 1,
dy (t ) dt t =0
=0
Poles and transient response, b=3
Poles and transient response, b = 2.75
EE 3CL4, L 5 1 / 30 Tim Davidson Transfer function Step response Transfer function of DC motor

McMaster University
Winter 2011
Outline
1 Transfer function
2 Step response
3 Transfer function of DC motor
Transfer function
Denition: Laplace transform of output over Laplace transform of input when initial conditions are zero
Recall the mass-spring-damper system,
Transfer function, MSD system

For the mass-spring-damper system, Y (s ) = 1/M R (s) s2 + (b/M )s + k /M (s + b/M ) + 2 y (0 ) s + (b/M )s + k /M 1 dy (t ) + 2 s + (b/M )s + k /M dt
t =0
Therefore, transfer function is: s2 1 1/M = 2 + (b/M )s + k /M Ms + bs + k
Step response
Recall that u (t ) 1 s Therefore, for transfer function G(s ), the step response
is: L 1
G(s) s
For the mass-spring-damper system, step response is
L 1
1 s(Ms2 + bs + k )
What is the nal position for a step input?
Recall nal value theorem. Final position is 1/k .

Consider again the case of M = k = 1, b = 3 0.
n = 1, = 1.5 0.
Poles and step response, b = 3
Poles and step resp., b = 2.75
Poles and step resp., b = 2
A DC motor
We will consider linearized model for each component Flux in the air gap: (t ) = Kf if (t ) (Magnetic cct, 2CJ4) Torque: Tm (t ) = K1 (t )ia (t ) = K1 Kf if (t )ia (t ). Is that linear? Only if one of if (t ) or ia (t ) is constant We will consider armature control: if (t ) constant
Armature controlled DC motor
if (t ) will be constant (to set up magnetic eld), if (t ) = If Torque: Tm (t ) = K1 Kf If ia (t ) = Km ia (t ) Will control motor using armature voltage Va (t ) What is the transfer function from Va (s ) to angular
position (s)?
Origin?
Towards transfer function
Tm (t ) = Km ia (t ) Tm (s ) = Km Ia (s ) KVL: Va (s ) = (Ra + sLa )Ia (s ) + Vb (s ) Vb (s ) is back-emf voltage, due to Faradays Law Vb (s ) = Kb (s ), where (s ) = s (s ) is rot. velocity Remember: transfer function implies zero init. conds
Torque on load: TL (s ) = Tm (s ) Td (s ) Td (s ): disturbance. Often small, unknown (e.g., wind) Load torque to angle (Newton plus friction):
TL (s) = Js2 (s) + bs(s)

Now put it all together
(s)Vb (s) Tm (s ) = Km Ia (s ) = Km Va R a +sLa Vb (s ) = Kb (s )
TL (s ) = Tm (s ) Td (s ) TL (s ) = Js 2 (s ) + bs (s ) = Js (s ) + b (s )
L (s ) Hence (s ) = T Js+b (s ) = (s )/s
Block diagram
(s)Vb (s) Tm (s ) = Km Ia (s ) = Km Va R a +sLa
Vb (s ) = Kb (s ) TL (s ) = Tm (s ) Td (s ) TL (s ) = Js 2 (s ) + bs (s ) = Js (s ) + b (s )
L (s ) Hence (s ) = T Js+b (s ) = (s )/s
Transfer function
Set Td (s ) = 0 and solve (you MUST do this yourself)
G(s) =
(s) Km = Va (s) s (Ra + sLa )(Js + b) + Kb Km Km = 2) 2 s(s + 2n s + n
Third order :(
Second-order approximation
G (s ) =
(s) Km = Va (s) s (Ra + sLa )(Js + b) + Kb Km
Often armature time constant, a = La /Ra , is negligible Hence (you MUST derive this yourself)
G(s)
Km Km /(Ra b + Kb Km ) = s(1 s + 1) s Ra (Js + b) + Kb Km
where 1 = Ra J /(Ra b + Kb Km ) Using a power balance can show that Kb = Km
EE 3CL4, L 6 1 / 10 Tim Davidson Block diagram models

Example: Loop transfer function
Block diagram transformations

McMaster University
Winter 2011

Outline
1 Block diagram models
2 Block diagram transformations

Bock diagram models

A convenient way to represent a transfer function is via
a block diagram
In this case, U (s ) = Gc (s )R (s ) and Y (s ) = G(s )U (s ) Hence, Y (s ) = G(s )Gc (s )R (s ) Consistent with the engineering procedure of breaking
things up into little bits, studying the little bits, and then put them together

Simple example
Y1 (s ) = G11 (s )R1 (s ) + G12 (s )R2 (s ) Y2 (s ) = G21 (s )R1 (s ) + G22 (s )R2 (s )

Ea (s ) = R (s ) B (s ) = R (s ) H (s )Y (s ) Y (s ) = G(s )U (s ) = G(s )Ga (s )Z (s ) Y (s ) = G(s )Ga (s )Gc (s )Ea (s ) Y (s ) = G(s )Ga (s )Gc (s ) R (s ) H (s )Y (s )
Y (s) G(s)Ga (s)Gc (s) = R (s) 1 + G(s)Ga (s)Gc (s)H (s)

Each transfer function is a ratio of polynomials in s What is Ea (s )/R (s )?


Using block diagram transformations

Using block diagram transformations
EE 3CL4, L 7 1 / 18 Tim Davidson Transfer function of armature controlled DC motor Application to disk drive read system Our rst control system design Characteristics of feedback

McMaster University
Winter 2011
Outline
1 Transfer function of armature controlled DC motor
2 Application to disk drive read system
3 Our rst control system design
4 Characteristics of feedback control systems, Ch 4
Last week
We constructed a block diagram for armature-controlled
DC motor
if (t ) constant; motor controlled using va (t )
(s) Determined the transfer function V a (s )
Transfer function
G (s ) =
Km (s) = Va (s) s (Ra + sLa )(Js + b) + Kb Km Km = 2) s(s2 + 2n s + n
Model for a disk drive read system
Uses a permanent magnet DC motor Can be modelled using arm. contr. model with Kb = 0 Hence, motor transfer function:
G (s ) =
(s) Km = Va (s) s(Ra + sLa )(Js + b)
Assume for now that the arm is stiff
Typical values
G (s ) = (s) Km = Va (s) s(Ra + sLa )(Js + b)
G(s) =
5000 s(s + 20)(s + 1000)
Time constants
Initial model
G(s) =
5000 s(s + 20)(s + 1000)
Motor time constant = 1/20 = 50ms Armature time constant = 1/1000 = 1ms Hence
G(s)
5 s(s + 20)
A simple feedback controller

Now that we have a model, how to control?
Here, Y (s) = (s).
Simplied block diagram
What is the transfer function from command to
position? Derive this yourself Y (s) K a G (s ) = R (s) 1 + Ka G(s)

For second-order G(s), and a gain of Ka = 40,
Y (s ) =
s2
200 R (s ) + 20s + 200
What is the response for R (s ) = 0.1/s ? Does it meet our design criteria?
Within 1m within 50ms?
Step response
Response to r (t ) = 0.1u (t )
Closed and open loop
The error signal
Error signal E (s) = R (s) Y (s) For the case where H (s) = 1 (derive this for yourself): E (s) = 1 R (s) 1 + Gc (s)G(s) G(s) Td (s) 1 + Gc (s)G(s) Gc (s)G(s) + N (s ) 1 + Gc (s)G(s)
Loop gain
Again, set H (s) = 1 Dene loop gain: L(s) = Gc (s)G(s) E (s ) = 1 G(s) L(s) R (s) Td (s ) + N (s) 1 + L(s) 1 + L(s) 1 + L(s)
Sensitivities
Again, set H (s) = 1 Dene sensitivity: S (s) = L(s) 1 + L(s) 1 1 + L(s)
Dene complementary sensitivity: C (s) =
E (s) = S (s)R (s) S (s)G(s)Td (s) + C (s)N (s) Note that S (s) + C (s) = 1. Trading S (s) against C (s): a key to the art of control design
EE 3CL4, L 8 1 / 11 Tim Davidson Characteristics of feedback

McMaster University
Winter 2011
Outline
1 Characteristics of feedback control systems
The closed loop
For the whole of this lecture we will consider unity feedback: H (s) = 1
The output signal
Under standing assumption of H (s) = 1, what is Y (s)? Y (s ) = Gc (s)G(s) R (s ) 1 + G c (s )G (s ) G(s) + Td (s) 1 + Gc (s)G(s) Gc (s)G(s) N (s ) 1 + Gc (s)G(s)
The error signal
Dene the error signal: E (s) = R (s) Y (s) For H (s) = 1, what is E (s)? E (s ) = 1 R (s ) 1 + Gc (s)G(s) G (s ) Td (s) 1 + Gc (s)G(s) + Gc (s)G(s) N (s) 1 + Gc (s)G(s)
Note: For H (s) = 1, contr. input Ea (s) = R (s) Y (s) + N (s)
Loop gain
Dene loop gain: L(s) = Gc (s)G(s) 1 G(s) L(s) R (s) Td (s ) + N (s) 1 + L(s) 1 + L(s) 1 + L(s)
E (s ) =
What do we want?
A taste of loop shaping
E (s ) =
1 G(s) L(s) R (s) Td (s ) + N (s) 1 + L(s) 1 + L(s) 1 + L(s)
What do we want? Good tracking: E (s ) does depend only weakly on R (s ) = L(s) large where R (s) large
Good disturbance rejection:
= L(s) large where Td (s) large

Good noise suppression:
= L(s) small where N (s) large
A taste of loop shaping

Possibly easier to understand in pure freq. domain, s = j Recall that L(s) = Gc (s)G(s), G(s): xed; Gc (s): controller to be designed Good tracking: = L(s) large where R (s) large |L(j )| large in the important frequency bands of r (t ) Good dist. rejection: = L(s) large where Td (s) large |L(j )| large in the important frequency bands of td (t ) Good noise suppr.: = L(s) small where N (s) large |L(j )| small in the important frequency bands of n(t ) Typically, L(j ) is a low-pass function How big should |L(j )| be? Any other constraints? Stability!
Sensitivities
Dene sensitivity: S (s) =
1 1 + L(s) L(s) 1 + L(s)
Dene complementary sensitivity: C (s) =
E (s) = S (s)R (s) S (s)G(s)Td (s) + C (s)N (s) Note that S (s) + C (s) = 1. Trading S (s) against C (s), with stability is the essence of the art of control design
Model sensitivities
In practice we rarely model G(s ) exactly, and it may age

(s) How does T (s ) = Y R (s) change as G(s ) changes? G(s ) + G(s ). T (s)/T (s) ln T (s) limG(s)0 = G(s)/G(s) ln G(s) For an open loop system this sensitivity =1
For the closed loop system, with H (s ) = 1,
S (s) =
1 1+Gc (s)G(s) ,
as earlier
EE 3CL4, L 9 1 / 18 Tim Davidson More advantages of feedback Price of feedback Example: English Channel boring machines Example: Disk drive read system

Lecture 9 (Given as the 10th lecture in Winter 2011)
Tim Davidson
McMaster University
Winter 2011
Outline
1 More advantages of feedback
2 Price of feedback
3 Example: English Channel boring machines
4 Example: Disk drive read system
Advantages of feedback so far
reduced sensitivity to disturbances reduced sensitivity to model variation can also manipulate the transient response,
to some degree (next week)

can also control steady-state response to certain
inputs, without tight calibration
Steady-state error
Recall standing assumption of H (s) = 1 Consider R (s) only; set Td (s) and N (s) to zero E (s) = R (s) Y (s) = 1 R (s ) 1 + Gc (s)G(s)
So what is the steady state error? If E (s) = 1+Gc (1 s)G(s) R (s ) has no poles in the close right half plane, except, perhaps for a simple pole at the origin,
t
lim e(t ) = lim sE (s)

s0
Steady-state error for step
Consider the case where r (t ) = u (t ). = R (s ) = 1/s . E (s ) =
1 1 1 + Gc (s)G(s) s
1 1 + Gc (s)G(s) 1 Hence, steady-state error is 1 + Gc (0)G(0) How to make this small? Large loop gain at DC
lims0 sE (s ) = lims0
Steady-state error for ramp
Consider r (t ) = tu (t ). = R (s ) = 1/s 2 .
1 1 E (s ) = 1+G ( c s )G(s ) s 2 1 1 lims0 sE (s ) = lims0 1+G ( c s )G(s ) s How to make this nite? n(s) c (s ) Let Gc (s ) = n dc (s) , G(s ) = d (s) dc (s)d (s) 1 sE (s ) = d (s)d (s)+nc (s)n(s) s c For nite SS error, dc (s )d (s ) must contain a factor s
i.e., either Gc (s ) or G(s ) must have pole at origin
integration
Price of feedback
more components than open loop less gain than open loop
Gc (s)G(s) 1+Gc (s)G(s)
instead of Gc (s)G(s)
potential for instability
English Channel boring machines
Y (s) =
1 K + 11s R (s ) + 2 Td (s) s2 + 12s + K s + 12s + K
Lets consider step and step disturbance responses for two values of K , 100 and 20.
English Channel boring machine

K = 100

K = 20
Disk drive read system
Y (s ) =
s3
1020s2
5000Ka R (s) + 20000s + 5000Ka s + 1000 + 3 Td (s ) 2 s + 1020s + 20000s + 5000Ka
with Td (s) = 0, compare step resps with Ka = 10 and 80.
Step responses for Ka = 10, 80
Disturb. step resp. for Ka = 80
To reduce this response need larger Ka but larger Ka will result in more oscilliatory response
EE 3CL4, L 10 1 / 19 Tim Davidson Performance of feedback control systems Performance of second-order systems

Lecture 10 (Given as the 9th lecture in Winter 2011)
Tim Davidson
McMaster University
Winter 2011
Outline
1 Performance of feedback control systems
2 Performance of second-order systems
Basic performance criteria
Stability (next week) Steady-state response to chosen inputs Transient response to chosen inputs Compromises: the art of design
Typical test signals
Step, ramp, parabolic
A second-order system
Now examine, in detail, a particular class of second order systems
Y (s ) =
2 n G(s) R (s ) R (s ) = 2 2 1 + G (s ) s + 2n s + n
Step response
What is the step response? Set R (s) = 1/s; take inverse Laplace transform of Y (s) Y (s) =
2 n 2 s s2 +2n s+n
For the case of 0 < < 1, y (t ) = 1 where = 1 n t e sin(n t + )
1 2 and = cos1 .
G(s) 1+G(s)
Recall pole positions of
(ignore the zero):
Typical step responses, xed n
Typical step responses, xed
Key parameters of (under-damped) step response

With = 1 2 and = cos1 , y (t ) = 1 1 n t e sin(n t + )
Peak time and peak value
y (t ) = 1
1 n t e sin(n t + )
Peak time: rst time dy (t )/dt = 0 Can show that this corresponds to n Tp = Hence, Tp =
1 2
2 Hence, peak value, Mpt = 1 + e / 1
Percentage overshoot
Let fv denote the nal value of the step response. Percentage overshoot dened as: P.O. = 100 Mpt fv fv
In our example, fv = 1, and hence P.O. = 100 e

/
1 2
Overshoot vs Peak Time

This is one of the classic trade-offs in control
Steady-state error, ess
In general this is not zero. However, for our second-order system, y (t ) = 1 1 n t e sin(n t + )
Hence ess = 0
Settling time
y (t ) = 1
1 n t e sin(n t + )
How long does it take to get within x % of nal value? Approx. when en t < x /100 When x = 2, that corresponds to n Ts 4;
that is, 4 time constants 4 In that case, Ts = n
Rise time (under-damped)
y (t ) = 1
1 n t e sin(n t + )
How long to get to the target (for rst time)? Tr , the smallest t such that y (t ) = 1
10%90% Rise time
What is Tr in over-damped case? Hence, typically use Tr 1 , the 10%90% rise time
10%90% Rise time
Difcult to get an accurate formula Linear approx. for 0.3 0.8 (under-damped),
EE 3CL4, L 11 1 / 19 Tim Davidson Step response of a class of second-order systems (review) A taste of poleplacement design Transient performance, poles and zeros Summary and plan

Lecture 11 (Given as the 12th lecture in 2011)
Tim Davidson
McMaster University
Winter 2011
Outline
1 Step response of a class of second-order systems
(review)
2 A taste of pole-placement design
3 Transient performance, poles and zeros
4 Summary and plan
Class of systems
We considered the step response of an (under-damped) second order system
Y (s) = T (s)R (s) =
2 n R (s ) 2 s2 + 2n s + n
2 n T (s) = 2 2 s + 2n s + n
Poles of T (s)
Where are the poles? s1 , s2 = n j n 1 2 = n cos() j n sin()
Step response
Y (s ) = T (s )
2 1 1 n = 2 2 s s + 2n s + n s
= y (t ) = 1 where =
1 n t e sin(n t + )
1 2 and = cos1 .
Properties of step response
Settling time, Ts ,: 2% settling time
4 n
2 Percentage overshoot: P.O. = 100 e / 1
Design problem
For what values of K and p is

the settling time 4 secs, and the percentage overshoot 4.3%?
2 K n G(s) = 2 = 2 , 2 1 + G (s ) s + ps + K s + 2n s + n where n = K and = p/(2 K )
T (s ) =
Pole positions
Ts
4 n
P.O. = 100 e
1 2
For Ts 4, n 1 For P.O. 4.3%, 1/ 2
Where should we put the poles of T (s)?
Pole positions
n 1 s1 , s2 = n j n where = cos1 ( ). 1/ 2 1 2 = n cos() j n sin()
Final design constraints

n 1 p2 1/ 2 2K
Caveat
Our work on transient response has been for systems
with T (s) =
2 n 2 s2 + 2n s + n
What about other systems?
Poles, zeros and transient response

(s) Consider a general transfer function T (s ) = Y R (s )
Step response: Y (s ) = T (s ) 1 s Consider case with DC gain = 1; no repeated poles Partial fraction expansion
Y (s) =
1 + s
Ai + s + i
s2
Bk s + Ck 2 + 2) + 2k s + (k k
Step response
y (t ) = 1 +
i
Ai ei t +
k
Dk ek t sin(k t + k )
Summary: Desirable properties
With H (s) = 1, E (s) = R (s) Y (s), L(s) = Gc (s)G(s), E (s) = G (s ) L(s) 1 R (s ) Td (s) + N (s) 1 + L(s) 1 + L(s) 1 + L(s)
Stability Good tracking in the steady state Good tracking in the transient Good disturbance rejection (good regulation) Good noise suppression Robustness to model mismatch
Plan: Analysis and design techniques

Rest of course: about developing analysis and design techniques to address these goals
Routh-Hurwitz: Enables us to determine stability without having to nd the poles of the denominator of a transfer function Root locus Enables us to show how the poles move as a single design parameter (such as an amplier gain) changes Bode diagrams There is often enough information in the Bode diagram of the plant/process to construct a highly effective design technique Nyquist diagram More advanced analysis of the frequency response that enables stability to be assessed even for complicated systems
EE 3CL4, L 12 1 / 15 Tim Davidson Steady-state error

Lecture 12 (Given as the 11th lecture in 2011)
Tim Davidson
McMaster University
Winter 2011
Outline
1 Steady-state error
Steady-state error
E (s) = R (s) Y (s) =
1 R (s ) 1 + Gc (s)G(s)
If the the conditions are satised, the nal value theorem gives steady-state tracking error: ess = lim e(t ) = lim s
t s 0
1 R (s) 1 + Gc (s)G(s)
One of the fundamental reasons for using feedback, despite the cost of the extra components, is to reduce this error. We will examine this error for the step, ramp and parabolic inputs
Step input
ess = lim e(t ) = lim s
t s 0
1 R (s) 1 + Gc (s)G(s)
Step input: R (s ) = A s sA/s ess = lims0 1+G (s)G(s) = 1+lim A c s0 Gc (s )G(s ) Now lets examine Gc (s )G(s ). Factorize num., den.
Gc (s)G(s) = where zi = 0 and pk = 0.
K sN
M i =1 (s + zi ) Q k =1 (s + pk )
Limit as s 0 depends strongly on N . If N > 0, lims0 Gc (s )G(s ) and ess = 0 If N = 0,
ess =
A 1 + Gc (0)G(0)
System types
Since N plays such a key role,
it has been given a name

It is called the type number Hence, for systems of type N > 1,
ess for a step input is zero

A For systems of type 0, ess = 1+G ( c 0)G(0)
Position error constant

A For type-0 systems, ess = 1+G ( c 0)G(0) A Sometimes written as ess = 1+ Kp
where Kp is the position error constant

Recall Gc (s )G(s ) =
Q K M i =1 (s +zi ) Q sN Q k =1 (s +pk )
Therefore, for a type-0 system
Kp = lim Gc (s)G(s) =
s 0
M i =1 (zi ) Q k =1 (pk )
Note that this can be computed from positions of the
non-zero poles and zeros
Ramp input
The ramp input, which represents a step change in
velocity is r (t ) = At .
A Therefore R (s ) = s 2 Assuming conditions of nal value theorem are
satised, ess = lim A s(A/s2 ) = lim s0 s + sGc (s )G(s ) s0 1 + Gc (s )G(s ) A = lim s0 sGc (s )G(s )
Again, type number will play a key role.
Velocity error constant

For a ramp input ess = lims0 Recall Gc (s)G(s) =
A sGc (s)G(s)
Q K M i =1 (s +zi ) Q sN Q k =1 (s +pk )
For type-0 systems, Gc (s)G(s) has no poles at origin. Hence, ess For type-1 systems, Gc (s)G(s) has one pole at the origin. Q K i zi A Q Hence, ess = K , where K = v pk v
k
Note Kv can be computed from non-zero poles and zeros Suggests formal denition of velocity error constant Kv = lim sGc (s)G(s)
s0
For type-N systems with N 2, for a ramp input ess = 0
Parabolic input
The parabolic input, which represents a step change in
acceleration is r (t ) = At 2 /2.
A Therefore R (s ) = s 3 Assuming conditions of nal value theorem are
satised, ess = lim A s(A/s3 ) = lim 2 s0 s Gc (s )G(s ) s0 1 + Gc (s )G(s )
Again, type number will play a key role.
Acceleration error constant

For a parabolic input ess = lims0 Recall Gc (s)G(s) =
Q K M i =1 (s +zi ) Q sN Q k =1 (s +pk ) A s 2 Gc (s )G (s )
For type-0 and type-1 systems, Gc (s)G(s) has at most one pole at origin. Hence, ess For type-2 systems, Gc (s)G(s) has two poles at the origin. Q K i zi A Q Hence, ess = K , where K = a pk a
k
Again, Ka can be computed from non-zero poles and zeros Suggests formal denition of acceleration error constant Ka = lim s2 Gc (s)G(s)
s0
For type-N systems with N 3, for a parabolic input ess = 0
Summary of steady-state errors
Robot steering system
Lets examine a proportional controller: Gc (s) = K1
Gc (s )G(s ) = K1 K /( s + 1) Hence, type-0 system. Hence, for a step input,
ess = where Kp = K1 K .
A 1 + Kp
Robot steering system
Lets examine a proportional-plus-integral controller: Gc (s) = K1 + K2 K1 s + K2 = s s
(K1 s+K2 ) When K2 = 0, Gs (s )G(s ) = K s ( s+1) Hence, type-1 system.
Hence, for a step input, ess = 0 For ramp input,
ess =
A , Kv
where Kv = lims0 sGc (s)G(s) = KK2
Typical response
to a sawtooth input
EE 3CL4, L 13 1 / 12 Tim Davidson Stability

Condition in terms of poles Condition in terms of denominator coefcients

McMaster University
Winter 2011

Outline
1 Stability

Stability
A systems is said to be stable if all bounded inputs r (t ) give rise to bounded outputs y (t ) Counterexamples
Albert Collins, Jeff Beck (Yardbirds),
Pete Townshend (The Who), Jimi Hendrix, Tom Morello (Rage Against the Machine), Kurt Cobain (Nirvana)
Tacoma Narrows

Conditions for stability
y (t ) =
g ( )r (t ) d
Let r (t ) be such that |r (t )| r
|y (t )| =

g ( )r (t ) d g ( )r (t ) d

g ( ) d

Using this: system G(s) is stable iff
g ( ) d is nite

Condition in terms of poles?

We want

g ( ) d to be nite
Can we determine this from G(s)? We can write a general rational transfer function in the form G (s ) = K sN
k (s i (s + zi ) 2 m (s + 2m s 2 + 2 )) + ( m m
+ k )
Poles: 0, k , m j m Assuming N = 0 and no repeated roots, the impulse response is g (t ) =

k
Ak ek t +
m
Bm em t sin(m t + m )
Stability requires |g (t )| dt to be bounded; that requires k > 0, m > 0 In fact, system is stable iff poles have negative real parts

Marginal stability
Consider G(s ) = 1/s , simple pole at origin y (t ) = r (t ) dt if r (t ) = cos (t ), which is bounded, then y (t ) = sin(t ). Bounded If r (t ) = u (t ), which is bounded, then y (t ) = t . Not bounded Consider G(s ) = 1/(s 2 + 1), simple poles at s = j 1 Unit step response: u (t ) cos(t ). Bounded What if r (t ) is a sinusoid of frequency 1/(2 ) rad/sec? Not bounded
If G(s) has a pole with positive real part, or a repeated pole on j -axis output is always unbounded

Routh-Hurwitz condition
We have seen how to determine stability from the poles. Much easier than having to determine impulse response Can we determine stability without having to determine the poles? Yes! Routh-Hurwitz condition

Let G(s) =
p (s ) q (s) ,
where
q (s) = an sn + an1 sn1 + . . . a1 s + a0 = an (s r1 )(s r2 ) . . . (s rn ) where ri are the roots of q (s) = 0. By multiplying out, q (s) = 0 can be written as q (s) = an sn an (r1 + r2 + + rn )sn1 + an (r1 r2 + r2 r3 + . . . )sn2 an (r1 r2 r3 + r1 r2 r4 + . . . )sn3 + + (1)n an (r1 r2 r3 . . . rn ) = 0 If all ri are real and in left half plane, what is sign of coeffs of sk ? the same!

That observation leads to a necessary condition. Hence, not that useful for design A more sophisticated analysis leads to the Routh-Hurwitz condition, which is necessary and sufcient Hence, can be quite useful for design
EE 3CL4, L 13 11 / 12 Tim Davidson
R-H cond: A rst look

Consider an sn + an1 sn1 + an2 sn1 + + a1 s + a0 = 0
Stability
Construct a table of the form sn sn1 sn2 sn3 . . . s0 where bn1 = b n 3 = 1 an1 an1 an2 an an3 1 = a n 1 a n 1 an an1 a n 4 a n 5 cn1 = an a n 1 1 bn1 an2 an3 an1 bn1 an3 bn3 an an1 bn1 cn1 . . . hn1 an2 an3 bn3 cn3 . . . an4 an5 bn5 cn5 . . . ... ... ... ... ...
R-H cond: A rst look

Consider an sn + an1 sn1 + an2 sn1 + + a1 s + a0 = 0 Construct a table of the form sn sn1 sn2 sn3 . . . s0 Loosely speaking: Number of roots in the right half plane is equal to the number of sign changes in the rst column of the table Stability iff no sign changes in the rst column More sophisticated statement in next lecture an an1 bn1 cn1 . . . hn1 an2 an3 bn3 cn3 . . . an4 an5 bn5 cn5 . . . ... ... ... ... ...
Stability
EE 3CL4, L 14 1 / 19 Tim Davidson Routh Hurwitz condition

Disk drive read control

McMaster University
Winter 2011

Outline
1 Routh Hurwitz condition

Stability
Let G(s) =
p (s ) q (s) ,
where
q (s) = an sn + an1 sn1 + . . . a1 s + a0 System is stable iff all poles of G(s) have negative real parts Recall, poles are solutions to q (s) = 0 Can we nd a necessary and sufcient condition that depends only on ak so that we dont have to solve q (s) = 0?

1
Consider, with an > 0 an sn + an1 sn1 + an2 sn2 + . . . a1 s + a0 = 0 Construct a table of the form Row n Row n 1 Row n 2 Row n 3 . . . Row 0 an an1 bn1 cn1 . . . hn1 an2 an3 bn3 cn3 . . . an4 an5 bn5 cn5 . . . ... ... ... ... ...
Procedure provided on the following slides

3 4
Count the sign changes in the rst column That is the number of roots in the right half plane
Stability (poles in LHP) iff ak > 0 and all terms in rst col. > 0

Constructing RH table
an sn + an1 sn1 + an2 sn2 + . . . a1 s + a0 = 0
Step 2.1: Arrange coefcients of q (s) in rst two rows

Row n Row n 1 an an1 an2 an3 an4 an5 ... ...

Interlude
Determinant of a 2 2 matrix: a b c d = ad cb

Step 2.2: Construct 3rd row using determinants of 2 2 matrices constructed from rows above
Row n Row n 1 Row n 2 an an1 bn1 1 a n 1 an2 an3 an4 an5 ... ...
bn1 =
an a n 1
an2 an3

Step 2.2, cont: Construct 3rd row using determinants of 2 2 matrices constructed from rows above
Row n Row n 1 Row n 2 an an1 bn1 1 a n 1 an2 an3 bn3 an4 an5 ... ... ...
bn3 =
an a n 1
an4 an5

Step 2.3: Construct 4th row using determinants of 2 2 matrices constructed from rows above
Row n Row n 1 Row n 2 Row n 3 an an1 bn1 cn1 1 bn1 an2 an3 bn3 ... an4 an5 ... ... ...
cn1 =
a n 1 b n 1
an3 bn3
Step 2.4: Continue in this pattern. Caveat: If all elements of rst column are non-zero Will come back to that. Lets see some examples, rst

RH table, second-order system
q (s) = a2 s2 + a1 s + a0
Row 2 Row 1 Row 0 b1 = 1 a1 a2 a1
a2 a1 b1 a0 0
a0 0
= a0
Therefore, second order system is stable iff all three denominator coefcients have the same sign

RH table, third order system

q (s) = a3 s3 + a2 s2 + a1 s + a0
Row 3 Row 2 Row 1 Row 0
a3 a2 b1 c1
a1 a0 0 0
b1 =
1 a2
a3 a2
a1 a0
c1 =
1 b1
a2 b1
a0 0
= a0
Therefore, if a3 > 0, necessary and sufcient condition for third-order system to be stable is that a2 > 0, b1 > 0 and a0 > 0. b1 > 0 is equiv. to a2 a1 > a0 a3 , and this implies a1 > 0.

RH table, higher order systems

Consider a general case:
n sn + an1 sn1 + an2 sn2 + + a1 s + n =0
= s/n Normalize to natural frequency by dening s

2 n 2 n1 n +(an1 /n )s n1 +(an2 /n s )s + +(a1 /n )s +1 = 0

RH table, dealing with zeros
The Routh-Hurwitz table encounters trouble when there
is a zero in the rst column

The next row involves (1/0) times a determinant When some other elements in that row are not zero, we
can proceed by replacing the zero by a small positive number , and then taking the limit as 0 after the table has been constructed.
When a whole row is zero, we need to be a bit more
sophisticated (see next lecture)

RH table, zero rst element in non-zero row

As an example, consider q (s) = s5 + 2s4 + 2s3 + 4s2 + 11s + 10 Routh table
Row 5 Row 4 Row 3 Row 2 Row 1 Row 0 c1 = 4 12 = 1 2 0 c1 d1 10 2 4 6 10 0 0 d1 = 11 10 0 0 0 0 6c1 10 6 c1
12
Two sign changes, hence unstable with two RHP poles


Add velocity feedback (switch closed)
Using block diagram manipulation
G1 (s) =
5000 s + 1000
Gs (s) =
1 s(s + 20)

Closed loop
T (s ) =
Y (s) Ka G1 (s)G2 (s) = R (s) 1 + Ka G1 (s)G2 (s)(1 + K1 s)
Hence, char. eqn: s3 + 1020s2 + (20000 + 5000Ka K1 )s + 5000Ka = 0

Stabilizing values of K1 and Ka

s3 + 1020s2 + (20000 + 5000Ka K1 )s + 5000Ka = 0 Routh table
Row 3 Row 2 Row 1 Row 0 b1 = 1 1020 b1 5000Ka 20000 + 5000Ka K1 5000Ka
1020(20000 + 5000Ka K1 ) 5000Ka 1020
For stability we require b1 > 0 and Ka > 0 For example, Ka = 100 and K1 = 0.05. That pair gives a 2% settling time of 260ms
EE 3CL4, L 15 1 / 16 Tim Davidson Routh Hurwitz revision

Dealing with zero rows
Applications of Routh Hurwitz condition

Turning control of a tracked vehicle

McMaster University
Winter 2011

Outline

1 Routh Hurwitz revision
2 Applications of Routh Hurwitz condition

Recall Routh Hurwitz condition

1
Determine the characteristic polynomial (denominator of transfer function), with an > 0 q (s) = an sn + an1 sn1 + an2 sn2 + . . . a1 s + a0

Construct the Routh Table Row n Row n 1 Row n 2 Row n 3 . . . Row 0 an an1 bn1 cn1 . . . hn1 an2 an3 bn3 cn3 . . . an4 an5 bn5 cn5 . . . ... ... ... ... ...
Construction procedure reviewed on next slide

3
System is stable iff ak > 0 and all terms in rst col. > 0

Construction procedure
Row k + 2 Row k + 1 Row k To compute r3 , multiply
1 1 rst element of previous row = q1 by determinant of 2 2 matrix formed in the following way: The rst column contains the rst elements of the two rows above the element to be calculated The second column contains the elements of the two rows above that lie one column to the right of the element to be calculated

p1 q1 r1
a3 q3 r3
p5 q5 r5
... ... ...
Therefore
r3 =
1 q1
p1 p5 q1 q5
1 p1 q5 q1 p5 q1

Zero in the rst column

When there is a zero in the rst column, but other
elements in the row are not zero

Replace the zero by a small positive number, say , and
once the table has been constructed, take the limit as 0. (See previous lecture)

Zero row
It is possible that the Routh Hurwitz procedure can
produce a zero row

While this complicates the procedure, it yields useful

information for design

Zero rows occur when polynomial has either equal and
opposite roots on the real axis, or a pair of complex conjugate roots on the imaginary axis. The latter is more common, and more useful in design
So how can we deal with this? Routh Hurwitz procedure provides an auxiliary polynomial that contains the roots of interest as factors The coefcients of this polynomial appear in the row above the zero row We replace the zero row by the coefcients of the derivative of the auxiliary polynomial

Zero row example

q (s ) = s 5 + 2s 4 + 24s 3 + 48s 2 25s 50 = 0 Construct table

Row 5 Row 4 Row 3
1 2 0
24 48 0
25 50
(s) = 2s4 + 48s2 50 Auxiliary polynomial: q This is actually a factor of q (s). (s) are s2 = 1, 25 Using quadratic formula, roots of q (s) are s = 1, j 5 Hence roots of q
(s) 3 dq ds = 8s + 96s .
Replace zero row by these coefcients Row 5 1 24 25 Row 4 2 48 50 Row 3 8 96

Zero row example, cont

Now complete the table in the usual way

Row 5 Row 4 Row 3 Row 2 Row 1 Row 0
1 2 8 24 112.7 50
24 48 96 50 0
25 50
One sign change in rst column.
Indicates one root in right half plane. (s) is a factor of q (s). Recall q (s) Indeed, by polyn division q (s) = (s + 2)q (s) are 1 and j 5. We have seen that roots of q
Hence q (s ) does indeed have one root with a positive
real part.


Select K and a so that the closed-loop is stable, and the steady-state error due to a ramp is at most 24% of the magnitude of the command

Deal with stability rst

Transfer function: T (s) =
Gc (s)G(s) 1+Gc (s)G(s)

Char. equation: s4 + 8s3 + 17s2 + (K + 10)s + Ka = 0 Routh table

Row 4 Row 3 Row 2 Row 1 Row 0 b3 = 126 K 8 1 8 b3 c3 Ka 17 K+10 Ka Ka
c3 =
b3 (K + 10) 8Ka b3
For stability we require b3 > 0, c3 > 0 and Ka > 0

Stability region
These constraints can be rewritten as K < 126 Ka > 0 (K + 10)(126 K ) 64Ka > 0 For positive K ,
)(126K ) last constraint becomes a < (K +10 64K Region of stable parameters


Steady-state error to ramp

For a ramp input r (t ) = At , we have ess = A/Kv , where

Kv = lim sGc (s)G(s) = Ka/10

s0
Therefore, ess = 10A/(Ka) To obtain ess < 0.24A, we need Ka > 10/0.24 41.67, Any (K , a) pair in stable region with Ka > 41.67 will
satisfy design constraints

Set of parameters with desired performance

For positive K , stability region is below the blue solid curve desired steady-state error region is above the red dashed curve and below the blue solid curve Design example: (70,0.6)

Ramp response
Ramp response for case of K = 70 and a = 0.6

EE 3CL4, L 16 1 / 18 Tim Davidson The Root Locus Procedure

Preliminary examples Formal Procedure

McMaster University
Winter 2011

Outline
1 The Root Locus Procedure

Simple example
Transfer function T (s) =
KG(s) 1+KG(s)
Char. eqn: s2 + 2s + K = 0 Closed-loop poles: s1 , s2 = 1 1K
What does this look like as K goes from 0 to +?

Simple example

Another example
Transfer function T (s) = Consider K to be xed
KG(s) 1+KG(s)
Char. eqn: s2 + as + K = 0 Closed-loop poles: s1 , s2 = (a
a2 4K )/2
What does this look like as a goes from 0 to +?

Another example

What to do in the general case?
In the previous examples we exploited the simple factorization of second order polynomials To be truly useful, we need a more general procedure

Principles of general procedure
Transfer function T (s) =
KG(s) 1+KG(s)
p (s ) q (s )
Closed loop poles are solutions to q (s) = 0 These are also solutions to 1 + KG(s) = 0 In polar form, |KG(s)|KG(s) = 1 + j 0 = 1(180 + k 360 ) Therefore, for s0 to be a closed-loop pole, we must have |KG(s0 )| = 1 where k is any integer We will also keep in mind that R (s) and Y (s) correspond to real signals. Hence, closed-loop poles are either real or occur in complex-conjugate pairs and KG(s0 ) = (180 + k 360 )

In terms of poles and zeros

For s0 to be a closed-loop pole, we must have |KG(s0 )| = 1 and KG(s0 ) = (180 + k 360 )
Write G(s) =
Q KG M (s+zi ) Qn i = 1 , ( s j =1 +pj )
which means that the
open loop zeros are zi s; open loop poles are pi s For s0 to be a closed-loop pole |KKG |
M n n j =1 |s0 M i =1 |s0
+ zi | =1 + pj |
K + K G +
i =1
(s0 + zi )
j =1
(s0 + pi ) = 180 + k 360
Can we interpret these expressions in a geometric way?

Vector difference
Let u and v be complex numbers. Can you describe v u in geometric terms? Use the fact that v = u + (v u ). That means that v u is the vector from u to v
v u = ej . That is,
|v u | is the length of the vector from u to v . (v u ) is the angle of the vector from u to v
In our expressions we have terms of the form s0 + zi = s0 (zi ) and s0 + pj = s0 (pj )
Geometric interpretation
Magnitude criterion: |KKG |
n j =1 M i =1
The Root Locus Procedure

|s0 + zi | =1 |s0 + pj |
|KKG |
M i =1 distances n j =1 distances
from zeros of G(s) to s0 =1 from poles of G(s) to s0
Phase criterion:
M n
K + KG +
i =1 M
(s0 + zi )
j =1
(s0 + pi ) = 180 + k 360
K + KG +
i =1 n
angles from zeros of G(s) to s0 angles from poles of G(s) to s0 = 180 + k 360
j =1

Formal Procedure
We will rst consider the case of K going from 0 to +

Step 1
Write the characteristic equation as 1 + F (s ) = 0 Rearrange so that the parameter of interest is
contained in the multiplier K in an exprn of the form 1 + KP (s) = 0, where the numerator and denominator of P (s) are monic polynomials (the coefcient of the highest power of s is 1).
i =1 (s +zi ) Factorize P (s ) into poles and zeros, P (s ) = Qn (s+p ) j =1 j
QM
Hence characteristic equation is equiv. to

n j =1 (s
+ pj ) + K
M i =1 (s
+ zi ) = 0
Where does the locus start? Where are poles for K = 0? They are the poles of P (s ). Mark each with an

Step 1
n M
(s + pj ) + K
j =1 i =1
(s + zi ) = 0
Where do the poles end up? Where are poles for K ? Rewrite as (1/K )
n j =1 (s
+ pj ) +
M i =1 (s
+ zi ) = 0
The zeros of P (s ). Mark each with a Since M n there will often be zeros at , too
Summary: Root locus starts at poles of P (s) and ends at zeros of P (s) Note: Often P (s) = Gc (s)G(s) and K is an amplier gain. In that case, root locus (of the closed loop) starts at the open loop poles and ends at the open loop zeros.

Step 2
Phase condition:
M n
K +
i =1
(s0 + zi )
j =1
(s0 + pj ) = 180 + k 360
Recall that for K > 0, K = 0. What does this tell us when s0 is on the real axis?
Any complex conjugate pairs have no impact
Step 2, cont.
Phase condition for K > 0:
M i =1
The Root Locus Procedure

(s0 + zi )
n j =1
(s0 + pj ) = 180 + k 360
Lets examine effects of poles on the real axis
For s0,1 , all angles from poles to s0,1 are zero For s0,2 , right pole generates an angle of 180 , others zero For s0,3 , For s0,4 ,
n j =1 n j =1
(s0 + pj ) = 360 (s0 + pj ) = 540
Something similar for zeros. Therefore: sections of real axis on the locus must lie to left of odd number of (real-valued) poles and (real-valued) zeros of P (s)

Example
P (s ) =
2(s+2) s(s+4)
Step 1: Poles s = 0, 4; Zeros s = 2 Step 2: Determine segments on real axis In this case, this is enough to generate the complete root locus
EE 3CL4, L 17 1 / 15 Tim Davidson Sketching the Root Locus

Review of Principles Outline of steps Review of Steps 1 and 2 Step 3 Step 4

McMaster University
Winter 2011

Outline
1 Sketching the Root Locus

Principles
We would like to know where the closed-loop poles go
as a parameter of the loop (typically a controller design parameter) is changed.

We would like to gain insight from how the closed-loop
poles move in order to guide our design of the controller

Sketching the Root Locus

Step 1
Write the denominator of the closed-loop transfer
function in the form 1 + KP (s) = 0, where P (s) =

QM i =1 (s +zi ) Qn j =1 (s +pj )
K contains the parameter of interest

We will focus on the case in which K 0 We will discuss the negative root locus case later Root loci start at poles of P (s ) and end at zeros of P (s ),
including the zeros of P (s) at innity

Mark the poles of P (s ) with an Mark the (nite) zeros of P (s ) with a

Step 2
Using the phase condition, we showed that for K > 0,

any part of the root locus on the real axis lies to the left
of an odd number of (real-valued) poles and (real-valued) zeros of P (s)

Step 3
P (s ) =
QM i =1 (s +zi ) Qn j =1 (s +pj )
closed-loop characteristic equation

n M
(1/K )
j =1
(s + pj ) +
i =1
(s + zi ) = 0
As K +, there are M nite values of s that satisfy
the equation
+... How many zeros at innity? Recall that P (s ) = s sn +...
M
Therefore, n M zeros at innity
How do the loci approach the zeros at innity? Along equi-angular rays that intersect somewhere on
the real axis

Step 3, Angles
Consider a point s0 on the root locus far from the poles of P (s) and the nite zeros of P (s) Phase condition (for positive K ):
M i =1
(s0 + zi )
n j =1
(s0 + pj ) = 180 + k 360
Since the point s0 is far away from all zi and pj , all angles are approximately the same, say
Hence, phase cond. is approx: (M n) = 180 + k 360 Re-arranging, and using multiples of 360 , =
2k +1 nM 180
for k = 0, 1, . . . , (n M 1)

Step 3, Centroid
From where do these rays eminate?
i =1 (s +zi ) Recall P (s ) = Qn (s+p ) j =1 j
QM
For large s , effects of nite zeros almost cancelled out
by that of M of the nite poles

Therefore, as s gets large, the roots follow a similar
(s) = 1/(s A )nM . path to those of P

By equating rst couple of terms of Taylors expansion,
A =
poles of P (s) zeros of P (s) nM

n j =1 (pj )
M i =1 (zi )
nM

Example
+1 =0 Sketch the root locus of the char. eqn: 1 + K s(s+s 2)(s+4)2
Step 1: Poles of P (s): s = 0, 2, 4, 4; Zeros of P (s): s = 1 Step 2: Intervals on real axis:

Order poles and zeros of P (s ): -4, -4, -2, -1, 0 Examine from the right for intervals that are to the left of
an odd number of poles and zeros

[1, 0], [4, 2],[, 4]

Example
Partial root locus after Step 2

Example
Step 3: Asymptotes:
Angles: n M = 4 1 = 3.
Hence, angles are 60, 180, 300 Note that we already knew 180!
Centroid: A = 3

Example
Hence the complete root locus
What is the largest gain for which system is stable?

Step 4
Find values of K for which closed-loop poles lie on
imaginary axis.
Also nd the positions of these closed-loop poles How can we do this? Routh-Hurwitz table (as in tutorial) Gains of interest correspond to zero rows,
but remember not all zero rows correspond to closed-loop poles on j -axis
Find the closed-loop pole positions by factorizing the
auxiliary polynomial (polynomial with coeffs in row above zero row)
EE 3CL4, L 18 1/8 Tim Davidson Sketching the root locus Compensator design for VTOL aircraft

McMaster University
Winter 2011
Outline
1 Sketching the root locus
2 Compensator design for VTOL aircraft
Procedure
Steps 1 to 4
1
i i =1 1 + KP (s), with P (s) = Qn j =1 (s +pj ) Put an at the pj s; put a at the zi s Loci start at the s and end at the s or at innity
Write the denominator of the closed loop as Q

M
(s+z )
Parts of loci on real axis: to the left of an odd number of (real-valued) poles and (real-valued) zeros of P (s) n M asymptotes as K gets large: Angles = Centroid: A = 2k + 1 180 nM for k = 0, 1, . . . , (n M 1)
n j =1 (pj ) M i =1 (zi )
nM
Roots on j -axis and corresponding K s from zero rows and auxiliary polynomial of Routh-Hurwitz procedure
Using root locus for design
For this loop, P (s ) in root locus procedure is G(s ) What can we do if the root locus is not to our liking. Can we use the insight that we have developed to Q
(s+zi ) design a compensator Gc (s) = Q( j ) that we insert s+p between the amplier and G(s) so that the root locus with P (s) = Gc (s)G(s) is more to our liking?
Note that in the compensated system the zeros of P (s ) are the zi s from G(s ) i s from Gc (s) and the z the zeros of P (s ) are the pj s from G(s ) j s from Gc (s) and the p Lets attempt this for a VTOL aircraft
Compensator design for VTOL aircraft

In this experiment we will work with a model for the
vertical control system for a VTOL aircraft, such as the Harrier jump jet
The transfer function of the process/plant can be
approximated by G(s) =
1 s(s1)
Do you notice anything interesting about this model? Tasks: Sketch the root locus of a proportional controller Highlight some features of that root locus If the proportional root locus is not satisfactory, use insight from the root locus sketching procedure to choose a compensator so that the closed-loop has a satisfactory root locus

General Procedure Review of Steps 14 Step 5 Step 6 Step 7
Example

McMaster University
Winter 2011

Outline
Example
2 Example

General Procedure
Example

Steps 1 to 4
1
i i =1 1 + KP (s), with P (s) = Qn j =1 (s +pj ) Put an at the pj s; put a at the zi s Loci start at the s and end at the s or at innity
Write the denominator of the closed loop as Q

M
(s+z )
Example
Parts of loci on real axis: to the left of an odd number of (real-valued) poles and (real-valued) zeros of P (s) n M asymptotes as K gets large: Angles = Centroid: A = 2k + 1 180 nM for k = 0, 1, . . . , (n M 1)
n j =1 (pj ) M i =1 (zi )
nM
Roots on j -axis and corresponding K s from zero rows and auxiliary polynomial of Routh-Hurwitz procedure

Step 5
Since complex poles appear in conjugate pairs,
the root locus can leave the real axis only in even multiplicities; often just a pair
Due to phase criterion, angles of break away are evenly
Example
spaced; when a pair, they depart at 90 ; Examples:
What is the point of departure?

Step 5
Recall that the characteristic equation is 1 + KP (s ) = 0 Rewrite as K = p (s ), i.e., p (s ) = 1/P (s ) We want to nd the largest K such that there real
Example
solutions to K = p(s) in the neighbourhood of interest

(s ) This will occur at the solutions of dp ds = 0 that are real

Step 5, Example
(s+1) Root locus of 1 + KP (s ), with P (s ) = s(s+ 2)(s+3) . Outcome of Steps 1-3 of root locus sketching procedure
(Step 4 is not relevant in this case)
Example
Step 5: Dene p (s ) = 1/P (s ).

(s) dp ds = 0
2s3 + 8s2 + 10s + 6 = 0
The only real root in the interval [3, 2] is 2.46.
This is the breakaway point

Step 6
Determine angle of departure from (complex) poles
and angle of arrival to (complex) zeros

Lets consider a particular pole, p1 ; e.g.,
Example
In which direction does the locus leave p1 ? Use the fact that the phase condition must hold at any
point on the root locus.

Apply that to test points close to p1 ; e.g., s1

Step 6, cont
Example
Phase condition at test point s1 :
sum of angles from zeros to s1 sum of angles from other poles to s1 angle from (p1 ) to s1 = 180 + k 360
When s1 is close to p1 the angles from zeros and
angles from other poles are essentially the same as the angles to p1

Step 6, cont
Example
Phase condition at test point s1 : 2 3 1 = 180 2 90
3 from triangle with corners p3 , p1 and Re(p1 ) + j 0. Hence, 1 = 90 3 .

Step 6, cont
Example
Note that conjugate pole moves in a direction that preserves conjugate symmetry

Step 6, Summary
Using the phase condition,
angle of departure from pole at pj0 is j 0 = angles from zeros to (pj0 ) angles from other poles to (pj0 ) 180 + k 360
Example
Conjugate pairs move in complementary directions Angle of arrival at zeros is calculated in the same way

Step 7
Example
Join the segments that have been drawn with a smooth curve Curve should be as simple as possible Curve must respect conjugate symmetry of poles and
zeros of a system with real inputs and real outputs

General Procedure
Example

Example
Sketch root locus of 1 + KP (s) = 0 for K 0, where P (s) = 1 s4 + 12s3 + 64s2 + 128s
Example
poles: 0, 4, 4 j 4; zeros: no nite zeros; n M = 4 0 = 4 asymptotes Segments of real axis: [4, 0] Angles of asymptotes: 45 , 135 , 225 , 315 Centroid: (4 4 4)/4 = 3
2 3

Example, cont
Partial sketch from Steps 13
Example

Example, cont
Example
Closed loop denom: s4 + 12s3 + 64s2 + 128s + K = 0 Routh table implies stability for K < 568.89. Poles on j axis at j 3.266 To nd breakaway point, p(s) = (s4 + 12s3 + 64s2 + 128s). Set deriv. to zero: 4s3 + 36s2 + 128s + 128 = 0 Breakaway point 1.577

Example, cont
Example

Example, cont
Example
Angle of departure from 4 + j 4: Angle from pole at 4: 90 Angle from pole at 4 j 4: 90 Angle from pole at origin: 3 = 135 Hence angle of departure: 1 = 90 90 135 180 225

Example, cont
Example

Actual Root Locus
Example
EE 3CL4, L 20 1 / 13 Tim Davidson Sketching the Root Locus Parameter Design Negative Root Locus

McMaster University
Winter 2011
Outline
2 Parameter Design
3 Negative Root Locus
General Procedure (K 0)
Parameter Design
In the examples so far, 1 + KP (s ) has been the
denominator of a closed loop with

negative feedback proportional control, with positive gain, possibly, some compensation; i.e., P (s ) = Gc (s )G(s ).
However, same principles can also be applied to some
other design parameters

The key step is to rewrite the characteristic polynomial
of the closed loop in the form 1 + P (s), where is the (non-negative) parameter to be designed
This is not always possible, but when it is possible it
can be very useful
Parameter Design: Example 1

Suppose that the characteristic equation of the closed
loop is: s 3 + ( 3 + )s 2 + 3s + 6 = 0
Suppose we are interested in root locus for > 0 Rewrite as s 3 + 3s 2 + 3s + 6 + s 2 = 0. Hence,
1+
s2 =0 s3 + 3s2 + 3s + 6
Now sketch the root locus of 1 + P (s ), where
P (s) =
s2 (s + 1)3
Parameter Design: Example 2

We can adapt this to designs with two parameters, too. Consider a characteristic equation of the form
s3 + s2 + s + = 0.
The effect of varying from zero to innity for a given
value of corresponds to the root locus for s 1 + P (s) = 0, where P (s) = s3 +s 2 +

Note that, among other things, determines the
starting point of the root locus for ; i.e., the roots for = 0
With = 0, the root locus for is the roots of
(s), where P (s ) = 1 + P
1 s2 (s+1)
Sketches
Locus for with = 0 1 i.e., roots of 1 + s2 (s +1) as : 0 + Locus for with = 1 i.e., roots of 1 + s3 +ss 2 + 1 as : 0 +
Negative Root Locus

Our root locus procedure has been for parameters that
change from 0 to +
What if our parameter of interest goes from 0 to ? The underlying principles remain the same For s0 to be on the root locus, 1 + KP (s0 ) = 0. This implies Magnitude condition: |KP (s0 )| = 1 Phase condition KP (s0 ) = 180 + k 360 However, since K is now negative, its phase is 180 .
Therefore, some of the interpretations change.

That said, interpretations can be derived in the same
way as they were for the case of positive K , and they are quite familiar
Sketching Negative Root Locus

From the 12th edition of the textbook
Sketching Negative Root Locus
EE 3CL4, L 22 1 / 18 Tim Davidson Bode Diagrams

Quick overview Example: Disk drive read system
Compensators
Lead compensators

McMaster University
Winter 2010

Outline
Compensators
Lead compensators
1 Bode Diagrams
2 Compensators
Lead compensators

Frequency response
For a stable LTI system with transfer function G(s ),
Compensators
Lead compensators
the frequency response is G(s)|s=j .

Bode magnitude plot
20 log10 |G(j )| against log10
Bode phase plot
G(j ) against log10

Poles, zeros and frequency response

Consider a transfer function
Compensators
Lead compensators
G (s ) = K
+ zi ) j (s + pj )
i (s
Zeros: zi ; Poles: pj Recall that if system is stable,
frequency response is G(s)|s=j . Hence,

Magnitude:
|G(j )| = |K |
Phase:
prod. distances from zeros to j prod. distances from poles to j
G(j ) = K + sum angles from zeros to j sum angles from poles to j

Sketching Bode Diagrams

Convert G(s) = K sN
i (s j (s
Compensators
Lead compensators
+ zi )
+ pj )
2 2 k (s + 2k n,k s + n,k ) 2 2 r (s + 2d ,k nd ,r s + nd ,r )
where zi and pj are real, to G(j ) = K i (1 + j /zi ) N (j ) j (1 + j /pj )

r k
1 + 2k (j /n,k ) + (j /n,k )2 1 + 2d ,k (j /nd ,r ) + (j /nd ,r )2
Then apply rules learnt in EE2CJ4.

Compensators
Lead compensators
Sketch the Bode diagram of the open loop transfer function

Compensators
Lead compensators

Compensators
Compensators
Lead compensators
We have seen how the performance of a closed loop
can be adjusted by varying a parameter What happens if we are unable to obtain that performance that we want by doing this?
Ask ourselves whether this is really the performance
that we want
Ask whether we can change the system,
say by buying different components

seek to compensate for the undesirable aspects of the
plant

Cascade compensation
Compensators
Lead compensators
Usually, the plant is a physical process If commands and measurements are made electrically,
compensator is often an electric circuit

General form of the compensator is
Gc (s) =
n j =1 (s
M i =1 (s
+ zi ) + pj )
Therefore, the cascade compensator adds open loop
poles and open loop zeros These will change the shape of the root locus

Compensator design
Compensators
Lead compensators
Where should we put new poles and zeros to achieve desired performance? That is the art of compensator design We will consider rst order compensators of the form Gc (s) = K (s + z ) (s + p)
with the pole p in the left half plane and the zero, z in the left half plane, too
For reasons that will soon become clear when |z | < |p |: phase lead network when |z | > |p |: phase lag network

Lead compensation
Compensators
Lead compensators
Gc (s) =
K (s + z ) (s + p )
with |z | < |p|. That is, zero closer to origin than pole

Lead compensation
Compensators
Lead compensators
Gc (s) =
If |p |
K (s + z ) (s + p )
|z | then pole can be neglected, Gc (s) K (s + z )/p If z close to origin, compared to plant dynamics, Gs (s) K ps What is this? Differentiator, Gdiff = KD s Note that Gdiff (j ) = KD j . Hence, Gdiff (j ) = +90 Hence the name phase lead

Frequency response
Compensators
Lead compensators
Gc (j ) = where
(Kz /p) 1 + j /z K (j + z ) K1 (1 + j ) = = (j + p) 1 + j /p 1 + j
= 1/p : time constant = p /z : ratio of magnitudes of pole and zero K1 = K /: For phase lead, > 1

Bode diagram
Gc (j ) = Magnitude
Low frequency gain: K1 Corner frequency in numerator at z = z = 1/( ) Corner frequency in denominator at p = p = 1/ z < p High frequency gain: K1
K1 (1 + j ) 1 + j
Compensators
Lead compensators
Phase
( ) = atan( ) atan( ) At low frequency: ( ) = 0 At high frequency: ( ) = 0 In between: positive, with peak at =
zp

Diagram
Compensators
Lead compensators

A passive phase lead network
Compensators
Lead compensators
Homework: Show that characteristic
V2 (s) V1 (s)
has the phase lead
EE 3CL4, L 23 1 / 20 Tim Davidson Lead Compensation, a revision Lag Compensation Design of Lead, Lag and Lead-Lag Compensators using Root Locus
Lead Compensator example

McMaster University
Winter 2010
Outline
1 Lead Compensation, a revision
2 Lag Compensation
3 Design of Lead, Lag and Lead-Lag Compensators using
Root Locus Lead Compensator example
If commands and measurements are made electrically, compensator is often an electric circuit We will consider rst order compensators of the form Gc (s) = K (s + z ) (s + p)
with the pole, p, and the zero, z , both in the left half plane when |z | < |p|: phase lead network when |z | > |p|: phase lag network
Lead compensation
Gc (s) =
K (s + z ) (s + p )
with |z | < |p|. That is, zero closer to origin than pole
Bode Diagram
A passive phase lead network
Homework: Show that characteristic
V2 (s) V1 (s)
has the phase lead
Lag Compensators
Gc (s) =
1 + s 1 (s + z ) = 1 + s (s + p)
with z = 1/ , p = 1/( ) and = z /p > 1. Hence, pole closer to origin than zero
Frequency response
Gc (j ) = Magnitude
Low frequency gain: 1 Corner frequency in denominator at p = p = 1/( ) Corner frequency in numerator at z = z = 1/ p < z High frequency gain: 1/
1 + j 1 + j
Phase
( ) = atan( ) atan( ) At low frequency: ( ) = 0 At high frequency: ( ) = 0 In between: negative, with max. lag at =
zp
Bode Diagram
A passive phase lag network
Active lead and lag networks
Heres an example of an active network architecture.
RL design: Basic procedure

1
Translate design specications into desired positions of dominant poles Sketch root locus of uncompensated system to see if desired positions can be achieved If not, choose the positions of the pole and zero of the compensator so that the desired positions lie on the root locus (angle criterion), if that is possible Evaluate the gain required to put the poles there (magnitude criterion) Evaluate the total system gain so that the steady-state error constants can be determined If the steady state error constants are not satisfactory, repeat
This procedure enables relatively straightforward design of systems with specications in terms of rise time, settling time, and overshoot; i.e., the transient response. For systems with steady-state error specications, Bode (and Nyquist) methods may be more straightforward (later)
Lead Comp. example
Consider a case with G(s) = s(s1 +2) . Design a lead compensator to achieve damping coefcient = 0.45 and velocity error constant Kv > 20 What to do?
Lets draw something. Plot poles of G(s ). Where should desired roots be? Note that the settling time is not specied. We are free
to choose it. Lets start with Ts = 1, so n = 4. cos1 (0.45) 64 ; desired roots 4 j 8
Lead Comp. example

Now where to put the zero and pole? Rule of thumb: put zero under desired root, or just to the left Determine position of the pole using angle criterion sum angles from open-loop zeros sum angles from open-loop poles = 180 90 (116 + 104 + p ) = 180 = p = 50 Hence pole at 10.6 Gain of compensated system: 9(8.25)(10.4) Prod. dist. from open-loop poles = 96.5 Prod. dist. from open-loop zeros 8 Hence compensated open loop: Gc (s)G(s) =
96.5(s+4) s(s+2)(s+10.6)
Velocity constant: Kv = lims0 sGc (s)G(s) = 18.2 :(
Compensated root locus

Root locus and the step response with the dominant roots at their desired locations
What to do now?
We tried hard, but did not achieve the design specs Lets go back and re-examine our choices Zero position of compensator was chosen via rule of
thumb
Can we do better? We need the gain required to put dominant poles in
desired position to increase.

Most effective if we can move the compensator pole to
the left
That requires moving the zero to the left By how much? Show that for n = 10, z = 4.5, p = 11.6 and Kv = 22.7
Outcomes
Root locus approach to phase lead design was
reasonably successful in terms of putting dominant poles in desired positions; e.g., in terms of and n
However, root locus approach does not provide
independent control over steady-state error constants

That said, since lead compensators reduce the DC
gain, they are not normally used to control steady-state error.

The goal of our lag compensator design will be to
increase the steady-state error constants, without moving the dominant poles too far
EE 3CL4, L 25 1 / 12 Tim Davidson Design of Lag Compensators using Root Locus

Steady-state errors Lag compensator design Lag compensator example

McMaster University
Winter 2010

Outline
1 Design of Lag Compensators using Root Locus

We will consider rst order compensators of the form Gc (s) = K (s + z ) (s + p)
with the pole, p, and the zero, z , both in the left half plane when |z | < |p|: phase lead network when |z | > |p|: phase lag network

Steady-state errors
Consider the uncompensated loop (i.e., Gc (s) = KP ). If closed loop stable, steady state error for input R (s): ess = lim e(t ) = lim s
t s 0
R (s ) 1 + KP G(s)
For a type 1 uncompensated process, ess for r (t ) = At is A/Kv ,unc , where the velocity constant is Kv ,unc = lim sKP G(s)
s 0
For such a system, G(s) =
Q KG i (s+zi ) Q s j (s+pj ) .
Hence, Kv ,unc =
KP KG Q
Q pj
zi
Aside: What about ess for ramp of Type 0 and Type 2 processes

Steady-state errors
Now, for a Type 1 G(s), what about Kv of a compensated closed loop, for a compensator that does not have a zero at s = 0. Kv ,comp = lim sGc (s)G(s) =
s0
Kv ,unc lim Gc (s) KP s0

Lag compensator design

Consider a lag compensator of the form, Gc (s) = Kc Since it is a lag compensator, |z | > |p|. Design Principles We dont try to reshape the uncompensated root locus. We just try to increase the value of the desired error constant, without moving the poles (well not much) Reshaping was the goal of lead compensator design As a result, compensator pole and zero are usually close together (so that angle criterion nearly satised desired poles of uncompensated closed loop) Kc = KP (so that uncompensated closed loop poles are not moved on their root locus)
z In that case, Kv ,comp = (Kv ,unc /KP ) lims0 Gc (s) = p Kv ,unc s+z s +p .

Lag compensator design

1 2
4 5
Obtain the root locus of uncompensated system From transient performance specs, locate suitable dominant pole positions on that locus Obtain the loop gain for these points, K = KP KG ; hence the (closed-loop) steady-state error constant Calculate the necessary increase. Hence = z /p Place pole and zero close to the origin (with respect to desired pole positions), with z = p. Typically, choose z and p so that their angles to desired poles differ by less than 1 . Set Kc = KP
What if there is nothing suitable at step 2? Perhaps do lead compensation rst, then lag compensation on lead compensated plant. i.e., design a lead-lag compensator

Example
Lets consider, again, the case with G(s) = s(s1 +2) . Design a lag compensator to achieve damping coefcient = 0.45 and velocity error constant Kv > 20 Note: we will get a different closed loop from our lead design. First step, obtain uncompensated root locus, and locate desired dominant pole locations

Example
Gain required to put closed loop poles in desired position = prod. distances from open loop poles That is, K = 2.242 = 5. Therefore KP = K /KG = 5 Velocity error const: Kv ,unc = lims0 sKP G(s) = K /2 = 2.5 The increase required is 20/2.5 = 8 That implies must choose p = z /8, where z is chosen to be close to the origin with respect to dominant closed-loop poles

Example
Lets choose z = 0.1. Hence, p = 1/80

Example
Compensated open loop transfer function is now Gc (s)G(s) = 5(s + 0.1) s(s + 2)(s + 1/80)
Closed-loop poles: 0.95 j 1.98, 0.10
EE 3CL4, L 30 1 / 14 Tim Davidson Frequency response techniques

Mapping contours

McMaster University
Winter 2010

Mapping contours
Outline
1 Frequency response techniques
Mapping contours

Mapping contours
Introduction
We have seen techniques that determine stability of a
system:
Routh-Hurwitz root locus
However, both of them require a model for the plant Today: frequency response techniques Although they work best with a model For an open-loop stable plant, they also work with measurements Key result: Nyquists stability criterion Design implications: Bode techniques based on gain
margin and phase margin

Mapping contours
Characteristic equation
To determine the stability of the system we need to
examine the characteristic equation: F (s) = 1 + L(s) = 0 where L(s) = Gc (s)G(s)H (s).
The key result involves mapping a closed contour of
values of s to a closed contour of values of F (s).

We will investigate the idea of mappings rst

Mapping contours
Simple example
Set F (s ) = 2s + 1 Map the square in the "s-plane" to the contour in the
"F (s)-plane"

Mapping contours
Area enclosed
How might we dene area enclosed by a closed
contour?
We will be perfectly rigorous, but will go against
mathematical convention
Dene area enclosed to be that to the right when the
contour is traversed clockwise

What you see when moving clockwise with eyes right Sometimes we say that this area is encircled by the
clockwise contour

Mapping contours
Example with rational F (s)
s A mapping for F (s ) = s+ 2 Note that s -plane contour encircles the zero of F (s )
How many times does the F (s )-plane contour encircle
the origin?

Mapping contours
Cauchys Theorem
Nyquists Criterion is based on Cauchys Theorem: Consider a rational function F (s )

If the clockwise traversal of a contour s in the s-plane
encircles Z zeros and P poles of F (s)

and does not go through any poles or zeros then the corresponding contour in the F (s )-plane, F
encircles the origin N = Z P times in the clockwise direction A sketch of the proof next week. First, some examples

Mapping contours
Example 1
A mapping for F (s ) = s+s 1/2 s -plane contour encircles a zero and a pole Theorem suggests no clockwise encirclements of origin
of F (s)-plane
This is what we have!

Mapping contours
Example 2
s -plane contour encircles 3 zeros and a pole Theorem suggests 2 clockwise encirclements of the
origin of the F (s)-plane

Mapping contours
Example 3
s -plane contour encircles one pole Theorem suggests -1 clockwise encirclements of the
origin of the F (s)-plane

That is, one anti-clockwise encirclement

Mapping contours
Nyquists criterion
Nyquist was concerned about testing for stability How might one use Cauchy Theorem to examine this? Perhaps choose F (s ) = 1 + L(s ), as this determines
stability
Which contour should we use?

Mapping contours
Nyquists contour
EE 3CL4, L 31 1 / 14 Tim Davidson Nyquists Stability Criterion

McMaster University
Winter 2010
Outline
1 Nyquists Stability Criterion
Conformal Mappings
For a linear system with transfer function F (s ) Consider a clockwise closed contour in s -plane Map each point s on the contour to F (s ) Plot real and imaginary parts of F (s ) Forms a contour in the F (s ) plane Region encircled: to the right as contour traversed
Cauchys Theorem
Consider a rational function F (s ) If the clockwise traversal of a contour s in the s-plane
encircles Z zeros and P poles of F (s)

and does not go through any poles or zeros then the corresponding contour in the F (s )-plane, F
encircles the origin N = Z P times in the clockwise direction
Informal Justication
(s+z1 )(s+z2 ) Consider the case of F (s ) = ( s+p1 )(s+p2 ) F (s1 ) = z1 + z2 p1 p2
As the contour is traversed the nett contribution from
z1 is 360 degrees
As contour is traversed, the nett contribution from other
angles is 0 degrees
Hence, as contour is traversed, F (s ) changes by 360
degrees. One encirclement!
Informal Justication
Extending this to any number of poles and zeros inside
the contour
For a closed contour, the change in F (s ) is
360Z 360P
Hence F (s ) encircles origin Z P times
Towards Nyquists Criterion
Typically we are interested in stability of the closed loop In particular, the zeros of F (s ) = 1 + L(s ),
where L(s) is the open loop transfer function, are the closed loop poles
What should the contour be if we want to relate system
stability to the number of encirclements of the origin made by F (s)?
Nyquists contour
Coarse Applic. of Cauchy

Recall that the zeros of F (s ) = 1 + L(s ) are the poles of
the closed loop

Let P denote the number of right half plane poles of
F (s )
The number of right half plane zeros of F (s ) is the
N + P , where N is the number of clockwise encirclements of the origin made by the image of Nyquists contour in the F (s) plane.
A little difcult to parse. Perhaps we can apply Cauchys Theorem in a more
sophisticated way.
Towards Nyquists Criterion
F (s ) = 1 + L(s ), where L(s ) is the open loop transfer
function
Encirclement of the origin in F (s )-plane is the same as
encirclement of 1 in the L(s)-plane

This is more convenient, because L(s ) is often
factorized, and hence we can easily determine P

Now that we are dealing with L(s ), P is the number of
right-half plane poles of the open loop transfer function
Nyquists Criterion: Stable open loop
For a stable open-loop transfer function L(s ), the closed-loop system is stable if and only if the image
of Nyquists Contour in the L(s)-plane does not encircle the point (1, 0).
Nyquists Criterion: General case

For a general open-loop transfer function L(s ), the closed-loop system is stable if and only if the
number of counter-clockwise encirclements of the point (1, 0) made by the image of Nyquists Contour in the L(s)-plane is equal to the number of right half plane poles of L(s) Proof:
Based on the fact that the number of right half plane
zeros of F (s) is Z = N + P , where N is the number of clockwise encirclements of (1, 0) by the image of Nyquists Contour in the L(s)-plane, and P is the number of right half plane poles of L(s).
For the closed-loop to be stable, Z must be zero.
Example: Stable L(s)
L(s) =
100 (s + 1)(s/10 + 1)
For 0 < , think about the Bode mag. & phase sketches Recall that L(j ) = L(j ) Remember to examine the r part of the curve Note: No encirclements of (1, 0) = closed loop is stable

McMaster University
Winter 2010
Outline
1 Nyquists Stability Criterion
Nyquists Criterion
Consider a unity feedback system with an open loop transfer function L(s) = Gc (s)G(s)H (s),with no zs or ps on j -axis Let PL denote the number of poles of L(s) in RHP Consider the Nyquist Contour in the s-plane Let L denote image of Nyquist Contour under L(s) Let NL denote the number of clockwise encirclements that L makes of the point (1, 0) Nyquists Stability Criterion: Number of closed-loop poles in RHP = NL + PL This statement is equivalent to those in previous lecture, but often more convenient
Example: Pole of L(s) at origin
Consider
L(s) =
Like in servomotor
K s( s + 1)
Problem with Nyquists Contour It goes through a pole! Cauchys Theorem does not apply Must modify Nyquist Contour to go around pole Then Nyquist Criterion can be applied
Example: Pole of L(s) at origin
Now three aspects of the curve

Around the origin Positive frequency axis;
remember negative freq. axis yields conjugate

At
Around the origin
L(s ) = s( K s+1) Around the origin, s = ej ,
where goes from 90 to 90

In the L(s ) plane: lim 0 L( ej ) K K j This is: lim 0 e e j = lim 0
Up positive j -axis
For 0 < < , L(j ) = K
1+ 2 2
ej (90
atan( ))
For small , L(j ) is large with phase 90 For large , L(j ) is small with phase 180 For = 1/ , L(j ) = K / 2 ej 135
For s = rej for large r
For s = rej with large r , and from +90 to 90 , limr L(rej ) = K e j 2 r2 How many encirclements? None Implies that closed loop is stable for all positive K Consistent with what we know from root locus (Lab. 2)
Example with open loop RHP pole

Consider
L(s) =
PL = 1
K1 s(s 1)
For 0 < < , L(j ) = K1 2
+ 4
90 atan( )
For = 0+ , L(j ) is large with angle +90 . For , L(j ) is small with angle +180 Conjugate for < < 0 What about when s = ej for 90 90 ? L(s ) = K1 (180 )
Recall PL = 1 Number clockwise encirclements of (1, 0) is 1 Hence there are two closed loop poles in the RHP Consistent with root locus analysis (sketch)
We showed at the time that the closed-loop can be
stabilized by a PD controller
Can we see that in the Nyquist diagram? Plot the Nyquist diagram of Gc (s )G(s ), where
G(s) =
K1 s(s1)
and Gc (s) = 1 + K2 s
Recall that PL = 1 If K1 K2 > 1, there is one anti-clockwise encirc. of 1 In that case, number closed-loop poles in RHP is
1 + 1 = 0 and the closed loop is stable Again, consistent with root locus analysis (sketch)
EE 3CL4, L 33 1 / 11 Tim Davidson Nyquists Stability Criterion as a Design Tool

Relative Stability Gain margin and Phase margin Relationship to transient response

McMaster University
Winter 2010

Outline
1 Nyquists Stability Criterion as a Design Tool

Nyquists Criterion
Consider a unity feedback system with an open loop transfer function L(s) = Gc (s)G(s)H (s), Let PL denote the number of poles of L(s) in (open) RHP Consider the (modied) Nyquist Contour in the s-plane (looping to the right of any poles or zeros on the j -axis) Let L denote image of (mod.) Nyquist Contour under L(s) Let NL denote the number of clockwise encirclements that L makes of the point (1, 0) Nyquists Stability Criterion: Number of closed-loop poles in (open) RHP = NL + PL

One more example

L(s) = K (s 2) (s + 1)2
Open loop is stable, but has non-minimum phase (RHP) zero K 2 + 4 L(j ) = 180 atan(/2) 2 atan( ) 2 + 1 For small positive , L(j ) 2K 180 For large positive , L(j )
K
90
In between, phase decreases monotonically, 180 90 . magnitude decreases monotonically (Bode mag dia.) L(j ) =
2K 2 2 1+j (5 2 ) (1+ 2 )2
; When =
5, L(j ) = K /2
When s = rej with r and : 90 90 , L(s) (K /r )ej

Nyquist plot of L(s)/K
Number of open loop RHP poles: 0 Number of clockwise encirclements of 1: if K < 1/2: 0; if K > 1/2: 1 Hence closed loop is stable for K < 1/2; unstable for K > 1/2
This is what we would expect from root locus (sketch)

Relative Stability: Introductory Example

Consider L(s) = Nyquist Diagram: K s(1 s + 1)(2 s + 1)

Zooming in
Since L(s) is minimum phase (no RHP zeros), we can zoom in
For a given K , how much extra gain would result in instability? we will call this the gain margin how much extra phase lag would result in instability? we will call this the phase margin

Formal denitions
Gain margin:
where x is the frequency at which L(j ) reaches 180 amplifying the open-loop transfer function by this amount would result in a marginally stable closed loop
Phase margin:
1 |L(j x )| ,
180 + L(j c ), where c is the frequency at which |L(j )| equals 1 adding this much phase lag would result in a marginally stable closed loop
These margins can be read from the Bode diagram

Bode diagram
L(j ) =
1 j (1 + j )(1 + j /5)
Gain margin 15 dB Phase margin 43

Phase margin and damping

Consider a second-order open loop of the form 2 n L(s) = s(s+2 , with < 1 n s) Closed-loop poles s1 , s2 = n j n
4 2 2 4 Can show that c + 4 2 n c n = 0;
1 2
Let c be the frequency at which |L(j )| = 1 Equivalently,

2 c 2 n
4 4 + 1 2 2
By denition, pm = 180 + L(j c ) Hence pm = atan 2 (4 + 1/ 4 )1/2 2
Phase margin is an explicit function of damping ratio! Approximation: for < 0.7, 0.01pm , where pm is measured in degrees

Relative Stability Gain margin and Phase margin Relationship to transient response Peak frequency response and bandwidth

McMaster University
Winter 2010

Outline
1 Nyquists Stability Criterion as a Design Tool

Nyquists Criterion
Consider a unity feedback system with an open loop transfer function L(s) = Gc (s)G(s)H (s), Let PL denote the number of poles of L(s) in (open) RHP Consider the (modied) Nyquist Contour in the s-plane (looping to the right of any poles or zeros on the j -axis) Let L denote image of (mod.) Nyquist Contour under L(s) Let NL denote the number of clockwise encirclements that L makes of the point (1, 0) Nyquists Stability Criterion: Number of closed-loop poles in (open) RHP = NL + PL

Relative Stability: Introductory Example

Consider L(s) = Nyquist Diagram: K s(1 s + 1)(2 s + 1)

Zooming
Since L(s) is minimum phase (no RHP zeros), we can zoom in
For a given K , how much extra gain would result in instability? we will call this the gain margin how much extra phase lag would result in instability? we will call this the phase margin

Formal denitions
Gain margin:
where x is the frequency at which L(j ) reaches 180 amplifying the open-loop transfer function by this amount would result in a marginally stable closed loop
Phase margin:
1 |L(j x )| ,
180 + L(j c ), where c is the frequency at which |L(j )| equals 1 adding this much phase lag would result in a marginally stable closed loop
These margins can be read from the Bode diagram

Bode diagram
L(j ) =
1 j (1 + j )(1 + j /5)

Phase margin and damping

Consider a second-order open loop of the form 2 n L(s) = s(s+2 , with < 1 n s) Closed-loop poles s1 , s2 = n j n 1 2
Let c be the frequency at which |L(j )| = 1

4 2 2 4 Square and rearrange: c + 4 2 n c n = 0;
Equivalently,
2 c 2 n
4 4 + 1 2 2
By denition, pm = 180 + L(j c ) Hence pm = atan 2 (4 + 1/ 4 )1/2 2
Phase margin is an explicit function of damping ratio! Approximation: for < 0.7, 0.01pm , where pm is measured in degrees

Previous example
L(j ) =
Phase margin 43
1 j (1 + j )(1 + j /5)
Damping ratio 0.43

Frequency response: open and closed loops

Nyquist diagram is a plot of the frequency response of
the open loop transfer function, L(s) = Gc (s)G(s) (for the case where H (s) = 1).
However, the transient response of the closed loop is
related to the frequency response of the closed loop.

Can we relate them?
Gc (j )G(j ) T (j ) = 1+ Gc (j )G(j ) Let Gc (j )G(j ) = u + jv ,
where variation of u and v with is implicit.

Then
M ( ) =
Gc (j )G(j ) u + jv = 1 + Gc (j )G(j ) 1 + u + jv

M -circles
From previous slide,
M=
u + jv 1 + u + jv
Algebraic manipulation leads to
M2 u 1 M2
+ v2 =
M 1 M2
What kind of object is this in the (u , v )-plane?

M M A circle, with center 1 , 0 and radius 1 M2 M2
2

M -circles
The M -circle that is tangent to the Nyquist plot
corresponds to the peak magnitude of the closed loop The corresponding frequency is the peak frequency M -circle crossing points also provide points on the magnitude response of closed loop

M -circles and Nyquist diagram
Left Nyquist plot (open loop) of a given L(j ),
for two values of the open loop gain

L(j ) Right: Bode Magnitude plot of the closed loops, 1+ L(j ) Roughly speaking, speed of response of closed loop
increases with increasing bandwidth
EE 3CL4, L 35 1 / 14 Tim Davidson Frequency Domain Approach to Compensator Design

McMaster University
Winter 2010
Outline
1 Frequency Domain Approach to Compensator Design
Frequency domain analysis
Analyze closed loop using open loop transfer function
L(s) = Gc (s)G(s)H (s).

Nyquists stability criterion Gain margin: |L(j1 x )| , where x is
the frequency at which L(j ) reaches 180 the frequency at which |L(j )| equals 1
Phase margin, pm : 180 + L(j c ), where c is Damping ratio: pm = f ( ), Settling time related to the bandwidth of the loop
Bode diagram
L(j ) =
1 j (1 + j )(1 + j /5)
Compensators and Bode diagram

We have seen the importance of phase margin If G(s ) does not have the desired margin,
how should we choose Gc (s) so that L(s) = Gc (s)G(s) does?

To begin, how does Gc (s ) affect the Bode diagram Magnitude:
20 log10 |Gc (j )G(j )| = 20 log10 (|Gc (j )| + 20 log10 |G(j )|

Phase:
Gc (j )G(j ) = Gc (j ) + G(j )
Lead Compensators
(s+z ) Gc (s ) = Kcs +p , with |z | < |p |, alternatively,
c 1+s Gc (s ) = K 1+s , where p = 1/ and = p /z > 1 Bode diagram:
Lead Compensation
What will lead compensation, do? Phase is positive: might be able to increase phase
margin pm
Slope is positive: might be able to increase the
cross-over frequency, c , (and the bandwidth)
Lead Compensation
c 1+s Gc (s ) = K 1+s By making the denom. real, can show that
Gc (j ) = atan
(1) 1+( )2
1 Max. occurs when = m = = Max. phase angle satises tan(m ) =
zp
1 2
1 Equivalently, sin(m ) = +1 At = m , we have |Gc (j m )| = Kc /
Bode Design Principles
Set the loop gain so that desired steady-state error
constants are obtained Insert the compensator to modify the transient properties:
Damping: through phase margin Response time: through bandwidth
Compensate for the attenuation of the lead network, if
appropriate To maximize impact of phase lead, want peak of phase near c of the compensated open loop
Design Guidelines
1
3 4 5
7 8 9
For uncompensated (i.e., proportionally controlled) closed loop, set gain Kp so that steady-state error constants of the closed loop meet specications Evaluate the phase margin, and the amount of phase lead required. Add a little safety margin to the amount of phase lead 1 From this, determine using sin(m ) = +1 Determine the frequency at which open-loop frequency response has magnitude 10 log10 () If we set m to be this frequency, then m will be the cut-off frequency of the compensated loop, and hence we will have maximum phase contribution to the compensated closed loop at the appropriate frequency Choose = 1/(m ) and hence p = m . Choose z = p/. Set Kc = Kp . Compensator: Gc (s) =
Kc (s+z ) s +p .
Example
Type 1 plant of order 2: G(s ) = s(s5 +2) Design goals: Steady-state error due to a ramp input less than 5% of velocity of ramp Phase margin at least 45 (implies a damping ratio) Steady state error requirement implies Kv = 20. For prop. controlled Type 1 plant: Kv = lims0 sKp G(s ).
Hence Kp = 8.
To nd phase margin of prop. controlled loop we need
to nd c , where |Kp G(j c )| =

6.2rad/s
40 j c (j c +2)
=1
Kp G(j c ) = 90 atan(/2) Hence pm, prop = 18
Example
pm, prop = 18 . Hence, need 27 of phase lead Lets go for a little more, say 30 So, want peak phase of lead comp. to be 30
1 Solving +1 = sin(30 ) yields = 3 Since 10 log10 (3) = 4.8 dB we should choose m to be
where 20 log10
40 j m (j m +2)
= 4.8 dB
Solving this equations yields m = 8.4rad/s Therefore z = m / = 4.8, p = z = 14.4,
Kc = 3 8
(s+4.8) Gc (s ) = 24 s+14.4 (s+4.8) Gc (s )G(s ) = s(s24 +2)(s+14.4) , actual pm = 43.6 Goal can be achieved by using a larger target for
additional phase, e.g., = 3.5
Bode Diagram

Lag Compensators

McMaster University
Winter 2010

Lag Compensators
Outline
1 Frequency Domain Approach to Compensator Design
Lag Compensators

Lag Compensators
Frequency domain analysis
Analyze closed loop using open loop transfer function
L(s) = Gc (s)G(s)H (s).

Nyquists stability criterion Gain margin: |L(j1 x )| , where x is
the frequency at which L(j ) reaches 180 the frequency at which |L(j )| equals 1
Phase margin, pm : 180 + L(j c ), where c is Damping ratio: pm = f ( ), Settling time related to the bandwidth of the loop

Lag Compensators
Bode diagram
L(j ) =
1 j (1 + j )(1 + j /5)

Lag Compensators
Compensators and Bode diagram
How does Gc (s ) affect the Bode diagram Magnitude:
20 log10 |Gc (j )G(j )| = 20 log10 (|Gc (j )| + 20 log10 |G(j )|

Phase:
Gc (j )G(j ) = Gc (j ) + G(j )

Lag Compensators
Lag Compensators
Gc (s ) = Gc (s ) = >1 We will consider case where Kc = Bode diagrams of lag compensators for two different s
Kc (s+z ) s+p , with |p | < |z |, alternatively, Kc 1+s 1+s , where z = 1/ and = z /p

Lag Compensators
What will lag compensation do?
Since zero and pole are typically close to the origin,
phase lag is not really used. What is useful is the attenuation above = 1/ : gain is 20 log10 (), with little phase lag Can reduce cross-over frequency, c , without adding much phase lag Tends to reduce bandwidth

Lag Compensators
Qualitative example
Uncompensated system has small phase margin Phase lag of compensator does not play a large role Attenuation of compensator does:
c reduced by about a factor of a bit more than 3 Increased phase margin is due to the natural phase characteristic of the plant

Lag Compensators
Lag Compensation
+s Gc (s ) = 11 +s Attenuation above = 1/ is 20 log10 ();
i.e., gain is 20 log10 ()

Phase lag above = 1/ is small

Lag Compensators
Bode Design Principles
For lag compensators:

Set the loop gain so that desired steady-state error
constants are obtained Insert the compensator to modify the phase margin:
Do this by reducing the cross-over frequency Observe the impact on response time
Basic principle: Set attenuation to reduce c far enough so that uncompensated open loop has desired phase margin compensated open loop

Lag Compensators
Design Guidelines
1
For uncompensated (i.e., proportionally controlled) closed loop, set gain Kp so that steady-state error constants of the closed loop meet specications Obtain the Bode diagram, evaluate the phase margin. If that is insufcient. . . Determine c , the frequency at which the uncompensated open loop has a phase margin equal to the desired phase margin plus 5 . Now design a lag compensator so that the gain of the compensated open loop at this frequency is 0 dB
Place the zero of the compensator around c /10 to
ensure we get almost full attenuation by the compensator at c Choose so that 20 log10 () is the attenuation needed to reduce the gain of the uncompensated closed loop at c to 0 dB Place the pole at p = z /

Lag Compensators
Example, same set up as lead design

Type 1 plant of order 2: G(s ) = s(s5 +2) Design goals: Steady-state error due to a ramp input less than 5% of velocity of ramp Phase margin at least 45 (implies a damping ratio) Steady state error requirement implies Kv = 20. For prop. controlled Type 1 plant: Kv = lims0 sKp G(s ).
Hence Kp = 8.
To nd phase margin of prop. controlled loop we need
to nd c , where |Kp G(j c )| =

6.2rad/s
40 j c (j c +2)
=1
Kp G(j c ) = 90 atan(/2) Hence pm, prop = 18

Lag Compensators
Example
Since want phase margin to be 45 , we set c such that G(j c ) = 180 + 45 + 5 = 130 . = c 1.5 Required attenuation is 20 dB. Actual curves are around 2 dB lower than the straight line approximation shown Hence = 10. Zero set to be one decade below c ; Pole is z / = 0.015 z = 0.15

Lag Compensators
Example: Compd open loop
4(s+0.15) Compensated open loop: Gc (s )G(s ) = s(s+ 2)(s+0.015) Numerical evaluation: new c = 1.58 new phase margin = 46.8 By design, Kv remains 20

Lag Compensators
Step response
Observe impact of reduction in bandwidth on response time

McMaster - Intro To Linear Control Systems

Încărcat de

Informații document

Descriere originală:

Titlu original

Drepturi de autor

Formate disponibile

Partajați acest document

Partajați sau inserați document

Opțiuni de partajare

Vi se pare util acest document?

Este necorespunzător acest conținut?

Drepturi de autor:

Formate disponibile

McMaster - Intro To Linear Control Systems

Încărcat de

Drepturi de autor:

Formate disponibile

EE 3CL4, L 1 1 / 21 Tim Davidson Why are you here? What is a control system? What tools will we use?

Administrative details Parting shot

EE3CL4: Introduction to Linear Control Systems

Why are you here?

Why are you here?

You might be interested in:

Why are you here?

Why are you here?

Designing disk drives

Developing technologies for enhanced power

distribution (Smart Grid)

Why are you here?

Why are you here?

What is a feedback control system?

What is a feedback control system?

What tools will we use?

(Math 2P04/2Z03, EE 2CI5, EE 2CJ4)

(Math 2P04/2Z03, EE 2CJ4, EE 3TP4)

(EE 2CI5, EE 2CJ4, EE 3TP4, . . . )

ITBA310 Ext. 27352 davidson@mcmaster.ca, EE3CL4 in subject line

Here we will focus on some key points

all laboratory reports in order to be eligible for a nal grade

have not seen before

EE3CL4: Introduction to Linear Control Systems

1 What is control engineering

4 Disk drive example

What is control engineering

system that controls the behaviour of the rst

inaccurate, or subject to noise/disturbances

Model, open-loop, closed-loop

Model, open-loop, closed-loop

What about disturbances? Mismatch (temperature, age)

Model, open-loop, closed-loop

Something for fourth year!

Watts yball governor

CD player speed control: Open-Loop

CD player speed control: Closed-Loop

John MacGregor (Chem Eng): Visual feedback control of avours

Structured approach to design

Disk drive: Intro

Spins at between 1800 and 7200 rpm,

perhaps even 10,000

Disk drive: Establish goals

Move from track a to track b

Disk drive: Control architecture

EE 3CL4, L 3 1 / 20 Tim Davidson Modelling physical systems

EE3CL4: Introduction to Linear Control Systems

Linearization Laplace transforms

EE 3CL4, L 3 2 / 20 Tim Davidson Modelling physical systems

1 Modelling physical systems

Linearization Laplace transforms

Translational Newtonian Mechanics Rotational Newtonian Mechanics

EE 3CL4, L 3 4 / 20 Tim Davidson Modelling physical systems

Differential equation models

Linearization Laplace transforms

describe dynamic systems

capacitors in 2CI, 2CJ

both translational and rotational

EE 3CL4, L 3 5 / 20 Tim Davidson Modelling physical systems

Linearization Laplace transforms

EE 3CL4, L 3 6 / 20 Tim Davidson Modelling physical systems

Linearization Laplace transforms