Unit 1: Introduction

Course Summary
Unit 1: Introduction
Unit 2: Modeling in the Frequency Domain
Unit 3: Time Response
Unit 4: Block Diagram Reduction
Unit 5: Stability
Unit 6: Steady-State Error
Unit 7: Root Locus Techniques
Unit 8: Design via Root Locus
Introduction
Course Summary
System Configurations
Open-loop vs. Closed-loop
Measuring Performance
Transient response, steady-state error, stability
The course cannot be summarized in one lecture. The intent here The Design Process
is to go through each unit of the course and mention some of the specifications
most important or poorly understood concepts. required performance measures
schematic
This brief tour is no replacement for the course itself! Many
transfer function for components
important details will be skipped. transfer function for system
analysis and design

ENGI 5821 Course Summary

Unit 2: Modeling in the Frequency Domain

Transfer Functions
Complex Frequency: x(t) = <{Xe st } Can be developed for LTI systems with zero-initial conditions
We can express constants, exponentials, sinusoids, and Transfer function G (s) gives the system output C (s) for any
exponentially decaying or growing sinusoids by varying s. The input R(s): C (s) = R(s)G (s)
actual amplitude and phase of a particular signal is expressed Electrical Systems
in X . Slow method: Develop DE’s, apply Laplace, form TF
Laplace Transform Fast method: Define transfer functions for individual
components and state the problem in the freq. domain
Transforms any signal into a combination of complex
Circuit analysis techniques developed for resistive circuits
exponentials automatically work for L’s and C’s in the freq. domain!
Tables of transform pairs and theorems
Op-Amps: Utilize ideal op-amp assumptions
Use partial-fraction expansion to decompose ratio of
polynomials
 Unit 3: Time Response
Mechanical Systems
Mechanical components
P P in translational and rotational forms
opposing forces = applied forces
One equation of motion for each linearly independent motion The poles give the form of the response, the zeros give the
In rotational systems torque replaces force (T = FR if F ⊥ weights
axis of rot.) and moment-of-inertia replaces mass
First-order systems yield exponential responses characterized
Gears linearly relate the motions of multiple shafts
Mechanical impedances can be reflected between shafts to by time constant
simplify the calculation of TF Second-order systems yield four different responses
Motors: Electromechanical systems characterized by ζ and ωn
Derived θm (s)/Ea (s) = K /s(s + α) Response specifications: Tp , %OS, and Ts
Jump back to the time-domain to run tests and evaluate Relationship between response specs. and pole position
motor parameters Additional poles or zeros
Real-axis poles or zeros far to the left have little effect
Linearization
A pole can cancel a nearby zero
Final exam will cover only the concept and application to a
simple DE

Unit 4: Block Diagram Reduction Unit 5: Stability

Definitions: Stability of the total response and the natural

response (we focused on the natural response)
RHP: Unstable, jω: Marginally stable, LHP: Stable
Recognize and reduce cascade, parallel, and feedback forms
All it takes is one RHP pole for instability (or one pair of jω
If none of the forms are apparent, blocks can be shift to the poles for marginal stability)
left or right of summing junctions and pickoff points
Characteristic coef’s missing or differing in sign ⇒ unstable
Moving blocks = algebraic manipulation
(opposite is not true)
Signal-flow graphs
Routh-Hurwitz
nodes are signals; edges are systems
Special case: zero in first column
Special case: ROZ ⇒ EP factor
EP has symmetric roots so system is either unstable or
marginally stable
Problem: find K such that...
Unit 6: Steady-State Error Unit 7: Root Locus Techniques

Vector representation of complex numbers

For any control system: E (s) = R(s) − C (s) Root locus: locations of closed-loop system poles as K is
Apply final value theorem to determine e(∞) varied
Unity feedback systems Properties of the RL
If ∠KG (s)H(s) = (2k + 1)180o we can find a K to satisfy
|KG (s)H(s)| = 1
Input Type 0: e(∞) Type 1: e(∞) Type 2: e(∞)
Step, u(t) 1
0 0 Sketching rules
1+Kp
Ramp, tu(t) ∞ 1
0 Refinements: breakaway and break-in points, jω crossings,
Kv
Para., t 2 u(t) ∞ ∞ 1 angles of departure and arrival
Ka
Search procedure required to find points with particular spec’s
Disturbances: e(∞) = eR (∞) + eD (∞)
Positive feedback requires changes to the RL definition and
sketching rules

Unit 8: Design via Root Locus

First check acceptable operating point on uncompensated RL
PI compensation: increase system type while maintaining
transient resp. PID: Design for transient response, then e(∞)
K (s + zc ) Analog PID implemented via op-amp
Gc (s) =
s Digital PID can be implemented in software (or digital
Choose zc as a small number; Requires active amplification hardware)
Lag compensation: increase static error constant (which PID tuning: Strategies to apply when system model is
reduces e(∞) unknown
K (s + zc ) Ziegler-Nichols (rules of thumb)
Gc (s) =
s + pc Method 1: Unit-step response is S-shaped
Method 2: System appears to involve integration and/or
Choose pc as a small number and adjust zc accordingly
underdamped poles
PD compensation: Adjust transient response Computational search
Gc (s) = K (s + zc )
Place zc to move RL to intersect desired operating point;
Requires active amplification
All of our design techniques rely on 2nd order approx.. Verify
approx. validity and simulate