Summary: Compensator design using the Root Locus; State Space

controller / plant Proportional (P)

R(s) + C(s)
Gc (s) Gp (s) K
−
/ compensator
choice of Proportional-Derivative (PD)
sensor / compensators
/ transducer H(s) K(s + z)
(usually, H(s) = 1).
Gc (s) 2.0
1.8
generic system block diagram with Proportional-Integral (PI) 1.6

controller/compensator and feedback 1.4

Ideal integral
s+z 1.2

c(t)
compensated
1.0
purpose: to improve the system’s dynamics K 0.8 Uncompensated

by proper choice of the controller TF and gain s 0.6

0.4

KGp (s)Gc (s) 0.2

Open—Loop TF: KGp (s)Gc (s)H(s) Closed—Loop TF: … and others that we haven’t 0
0 5 10 15 20

1 + KGp (s)Gc (s)H(s) seen: PID, lead, lag, lead-lag Time (Seconds)

Figure by MIT OpenCourseWare.

Root Locus sketching rules (negative feedback)
• Rule 1: # branches = # open-loop poles What is Root Locus?
• Rule 2: symmetrical about the real axis RL: locations on the s-plane
where the closed-loop poles move
• Rule 3: real-axis segments are to the left of an odd number of real-axis finite as we vary the feedback gain K
open-loop poles/zeros open-loop
poles/zeros
• Rule 4: RL begins at open-loop poles (K=0), ends at open-loop zeros (K=∞)
• Rule 5: Asymptotes: real-axis intercept σa, angles θa

jω ≡ jIm {s}
P P
finite poles − finite zeros (2m + 1)π
σa = θa = m = 0, ±1, ±2, . . .
#finite poles − #finite zeros #finite poles − #finite zeros

• Rule 6: Real-axis break-in and breakaway points

1 dK(σ) origin
Found by setting K(σ) = − (σ real) and solving =0 for real σ.
G(σ)H(σ) dσ s=0
• Rule 7: Imaginary axis crossings (transition to ⎧
instability)
£ ¤
⎨ Re KG(jω)H(jω) = −1,
Found by setting KG(jω)H(jω) = −1 and solving £ ¤
⎩
Im KG(jω)H(jω) = 0. σ ≡ Re {s}

2.004 Fall ’07 Lecture 26 – Wednesday, Nov. 7 #1

 Summary: Compensator design using the Root Locus; State Space
open-loop
poles/zeros
s on the RL ⇒ 1 + KGc (s)Gp (s)H(s) = 0 ⇒ KGc (s)Gp (s)H(s) = −1
2 ½
|KGc (s)Gp (s)H(s)| = 1
jω ≡ jIm {s}

l p1
lz l
p2 θp2 θp1
⇒ 6 {KGc (s)Gp (s)H(s)} = (2m + 1) π
θz
⎧
⎪ 1
−z −p2 p1 = 0 ⎪
⎨ K= sums taken over
⇒ |G c (s)G p (s)H(s)| Open Loop zeros/poles
⎪
⎪ P
⎩ P6
−4 −2 (s + z) − 6 (s + p) = (2m + 1)π
s-space geometry and transient characteristics
σ ≡ Re {s} jω
ωn2
TF = .
Here, Open Loop poles are p1 = 0, −p2 = −2, Open Loop zero is −z = −4. X
2
+ jωn 1 − ζ = jωd s2 + 2ζωn s + ωn2
Geometrical interpretation of the amplitude and phase contributions to s: ωn s-plane • Settling time
θ
− ζ ωn = − σ d
σ Ts ≈ 4/(ζωn );
lp1 = |s + p1 | = |s| ; lp2 = |s + p2 | = |s + 2| ; lz = |s + z| = |s + 4| ;
θp1 = 6 (s + p1 ) = 6 s; θp2 = 6 (s + p2 ) = 6 (s + 2) ; θz = 6 (s + z) = 6 (s + 4) . • Damped osc. frequency
X
2
− jωn 1 − ζ = −jωd p
ωd = 1 − ζ 2 ωn
Since the point s shown as crimson block belongs to the Root Locus,
p • Overshoot %OS
Figure by MIT OpenCourseWare.
à !
⎧ 1 − ζ2 ζπ
⎪ |s| |s + 2| lp1 lp2 cos θ = ζ tan θ = %OS = exp − p
⎪ K=
⎨ = ζ 1 − ζ2
⇒ |s + 4| lz
⎪ State Space & Phase Space
⎪
⎩ 6 From the Equation of Motion to the State—Space representation:
(s + 4) − 6 s − 6 (s + 2) = θz − θp1 − θp2 = −π µ ¶ µ ¶
x q1
mẍ(t)+bẋ(t)+kx(t) = w(t) → ≡ q(t) = state, y(t) ≡ ẋ(t) output
The crimson block is at s = −2 + j2 on the Root Locus. Using geometry, ẋ q2
µ ¶ µ ¶µ ¶ µ ¶ µ ¶
√ √ q̇1 0 1 q1 0 q1
lp1 = |s| = 2 2; lp2 = |s + 2| = 2; lz = |s + 4| = 2 2 ⇒ q̇(t) = = + w(t); y(t) = (0 1) ≡ cq.
q̇2 −k/m −b/m q2 1 q2
θp1 = 6 s = 3π/4; θp2 = 6 (s + 2) = π/2; θz = 6 (s + 4) = π/4. µ
0 1
¶
Solution to the state equations: 2 q t
A=
−k/m −b/m
We can see that indeed the angular contributions add up as µ ¶ sq̂(s) = Aq̂(s) + bW (s) ⇒ 1 q
0
θz − θp1 − θp2 = −π,
¡ √ ¢ ¡ √ ¢ b=
1
−1
q̂(s) = (sI − A) bW (s).
while the amplitude contributions give K = 2 2 × 2 / 2 2 ⇒ K = 2. −1
Y (s) = cq̂(s) = c (sI − A) bW (s).

2.004 Fall ’07 Lecture 26 – Wednesday, Nov. 7 #2