ELEN90055 Control Systems

The University of Melbourne

Problem Set 1 with Reference Solutions

m1
x1

k2 b2

k1 b1

m2
x2

Figure 1: Passive vibration control system: schematic model.

Problem 1
Consider the passive vibration control system shown schematically in Figure 1. An application of
Newton’s second law leads to the following differential equation model:
m1 ẍ1 + b1 ẋ1 + k1 x1 = F − m1 g + k1 ℓ1 + b2 (ẋ2 − ẋ1 ) + k2 (ℓ2 + x2 − x1 )
(1)
m2 ẍ2 + b2 ẋ2 + k2 x2 = −m2 g − k2 ℓ2 + b2 ẋ1 + k2 x1
where mi , ki , ℓi , b1 > 0 and b2 ≥ 0 are constants, xi is the position of the centre of gravity of mass mi
for i = 1, 2, and g is the acceleration due to gravity. The external force F applied to mass m1 is the
system input. The system output is the deviation of x1 from its equilibrium position x1q . The signal
x2 is an ‘internal’ signal.
1. Suppose the equilibrium input force Fq = 0. Using the ODE model (1), derive expressions for
the corresponding equilibrium values x1q and x2q of the mass positions in terms of mi , ki , ℓi , bi .
2. Determine an incremental linear ODE model (2 equations) for the relationship between the
. . .
incremental input force δF = F −0 and the incremental signals δx1 = x1 −x1q and δx2 = x2 −x2q .
3. Figure 2 (overleaf) shows a block-diagram for the incremental ODE model in Part 2. F (s)
denotes the Laplace transform of the applied (incremental) force and ∆xi (s) the Laplace trans-
.
form of δxi (t) = xi (t) − xiq for i = 1, 2. Determine the transfer functions H(s), P (s) and Q(s).

4. Show that
m2 s2 + b2 s + k2
m1 m2 s4 + (m2 b1 + (m1 + m2 )b2 )s3 + (m2 k1 + (m1 + m2 )k2 + b1 b2 )s2 + (b2 k1 + b1 k2 )s + k2 k1
is the transfer function from ∆F (s) to ∆x1 (s).

1
 ∆F (s) + ∆x1 (s)
H(s) P (s)
_

Q(s)
∆x1 (s) − ∆x2 (s)

Figure 2: Passive vibration control system: block-diagram model.

Reference Solution:

1. At equilibrium all signals are constant, whereby derivatives of any order are all zero valued.
Setting all derivative terms in (1) to zero gives the following equations describing equilibrium

k1 x1q = Fq − m1 g + k1 ℓ1 + k2 (ℓ2 + x2q − x1q )

(2)
k2 x2q = −m2 g − k2 ℓ2 + k2 x1q .

With Fq = 0, it follows that

(m1 + m2 )
x1q = ℓ1 − g
k1
k2 m1 + (k1 + k2 )m2
x2q = (ℓ1 − ℓ2 ) − g.
k2 k1

2. Note that δ̇xi = ẋi − 0 = ẋi and δ̈xi = ẍi − 0 = ẍi . As such, it is possible to rewrite (1) as

m1 δ̈x1 + b1 δ̇x1 + k1 (δx1 + x1q ) = δF − m1 g + k1 ℓ1 + b2 (δ̇x2 − δ̇x1 ) + k2 (ℓ2 + (δx2 + x2q ) − (δx1 + x1q ))
m2 δ̈x2 + b2 δ̇x2 + k2 (δx2 + x2q ) = −m2 g − k2 ℓ2 + b2 δ̇x1 + k2 (δx1 + x1q )

Using (2) then leads to the incremental model

m1 δ̈x1 + b1 δ̇x1 + k1 δx1 = δF + b2 (δ̇x2 − δ̇x1 ) + k2 (δx2 − δx1 )

(3)
m2 δ̈x2 + b2 δ̇x2 + k2 δx2 = b2 δ̇x1 + k2 δx1 .

3. Assuming that initially the system is at equilibrium so that δx1 (0) = δx2 (0) = 0, taking the
Laplace transform of each side of the two equations in (3) gives

(m1 s2 + b1 s + k1 )∆x1 (s) = F (s) − (b2 s + k2 )(∆x1 (s) − ∆x2 (s))

(m2 s2 + b2 s + k2 )∆x2 (s) = (b2 s + k2 )∆x1 (s).

1 b2 s + k2 m2 s 2
This is consistent with H(s) = , P (s) = and Q(s) =
b2 s + k2 m1 s2 + b1 s + k1 m2 s2 + b2 s + k2
in the block diagram shown.

2
 4. Note that

∆x1 (s) = P (s) · (H(s) · ∆F (s) − (∆x1 (s) − ∆x2 (s)))

= P (s) · (H(s) · ∆F (s) − Q(s) · ∆x1 (s))
= P (s) · H(s) · ∆F (s) − P (s) · Q(s) · ∆x1 (s).

As such, (1 + P (s)Q(s))∆1 (s) = P (s)H(s)F (s), whereby

P (s)H(s)
∆x1 (s) = ∆F (s).
(1 + P (s)Q(s))

Substituting the specific transfer functions H(s), P (s) and Q(s) identified in Part 3 into this
∆ (s)
expression yields the transfer function ∆xF1(s) given in the question.

Problem 2
Consider the inverted pendulum system shown in Figure 3. The pendulum is mounted on the shaft
of an armature-controlled DC servomotor. A simplified schematic model for the motor is shown in
Figure 4.

mg
l θ

Figure 3: The Pendulum System.

I
b
R L J
θ
v e

Figure 4: The armature-controlled DC servomotor

In this system

• R = armature resistance (Ohms)

• L = armature inductance (H)
• I = armature current (A)
• v = applied armature voltage (V)
• e = back EMF (V)

3
 • T = torque developed by the motor (Nm)
• J = (Jm + mℓ2 ) = moment of inertia of rotor and pendulum load (kg · m2 )
• b = viscous damping of shaft (N·s/m)
• θ = angular displacement of the motor shaft (rad)
Suppose the motor is configured so that the current, measured positive as shown in Figure 4, yields
a torque
T = Kt I
with Kt > 0, positive in the anti-clockwise direction, so that it opposes the torque due to gravity
when the angular position θ of the pendulum is positive (i.e., displaced in the clockwise direction as
shown in Figure 3.) The back EMF, measured positive as shown, is such that

e = −Kb θ̇

with Kb > 0 (if the motor configuration were such that the torque generated by a positive current is
positive in the clockwise direction, then one would have Kb < 0.) In answering the following, use the
sign convention described above, noting that this is an arbitrary choice.

1. Derive differential equations to model the dynamics of the system. Consider using one equation
for the electrical components and one for the mechanical components.

2. Suppose that by applying a constant voltage vq to the motor, the pendulum is held steady at
angle θq . Derive an algebraic equation relating θq to vq .

3. Linearize the differential equations obtained in Part 1, about the equilibrium operating point
(vq , θq ). Express the equations in terms of the incremental variables δv := v − vq , δθ := θ − θq .
. .
4. Derive the transfer function from ∆v (s) = L[δv ](s) to ∆θ (s) = L[δθ ](s).

5. Suppose Kb ≈ 0. In this case, is the incremental linear model stable or unstable? (Hint: The
answer can depend on the operating point θq )

6. When the incremental model is stable, what is the steady state value of the response δθ of the
incremental model to a step input δv of magnitude 0.1 V?

Reference Solution:

1. For the electrical part apply Kichoff’s voltage law: LI˙ +RI +e−v = 0 ⇔ LI˙ +RI −Kb θ̇−v = 0
For the mechanical part apply Newton’s 2nd law:

J θ̈ + bθ̇ − mgl sin θ + T = 0 ⇔ (Jm + ml2 )θ̈ + bθ̇ − mgl sin θ + Kt I = 0.

2. At equilibrium, all quantities remain constant. Therefore,

RIq = vq , mgl sin θq = Kt Iq .

This implies mgl sin θq = Kt vq /R.

3. Let δI (t) = I(t) − Iq , δv (t) = v(t) − vq , and δθ (t) = θ(t) − θq . Then the linearized equations
take the form

Lδ̇I + RδI − Kb δ̇θ − δv = 0

J δ̈θ + bδ̇θ − mgl(cos θq )δθ + Kt δI = 0

4
 Differentiating the second equation gives
 ... 
δ̇I = (−1/Kt ) J δθ + bδ̈θ − mgl(cos θq )δ̇θ

which when substituted into the first equation, along with the expression for δI that can be also
obtained from the second, gives the third-order linear ODE model
...
LJ δθ + (Lb + RJ)δ̈θ + (Rb + Kt Kb − Lmgl(cos θq ))δ̇θ − Rmgl(cos θq )δθ + Kt δv = 0.

4. From the first equation in part 3. above, with ∆I (s) = L[δI ](s), ∆v (s) = L[δv ](s), and ∆θ (s) =
L[δθ ](s),

L(s∆I (s) − δI (0)) + R∆I (s) = (∆v (s) + Kb (s∆θ (s) − δθ (0)).

Since δI (0) = 0 and δθ (0) = 0, it follows that

1
∆I (s) = (∆v (s) + Kb s∆θ (s)).
(Ls + R)

From the second equation in part 3. above,

(Js2 + bs − mgl(cos θq ))∆θ (s) − J(sδθ (0) + δ̇θ (0)) − bδθ (0) = −Kt ∆I (s),

which with δθ (0) = δ̇θ (0) = 0 gives

−Kt
∆θ (s) = ∆I (s)
(Js2
+ bs − mgl(cos θq ))
−Kt 1
= 2
(∆v (s) + Kb s∆θ (s))
(Js + bs − mgl(cos θq )) (Ls + R)

and therefore,
−Kt
∆θ (s) = ∆v (s)
(Ls + R)(Js2
+ bs − mgl(cos θq )) − Kt Kb s
−Kt
= 3 2
∆v (s)
LJs + (Lb + RJ)s + (Rb + Kt Kb − Lmgl(cos θq ))s − Rmgl(cos θq )
= G(s)∆v (s).

It can be confirmed that this is consistent with taking Laplace transform of the third order
ODE model given in part 3. above.

5. With Kb = 0, the system transfer fuction is G(s) = −Kt /((Ls + R)(Js2 + bs − mgl cos θq )).
As such, stability depends on the sign of cos θq . If cos θq < 0 then the system is stable because
all poles of the the transfer
p function are in the open left hand side of the complex plane at
s = −R/L and s = (−b ± b2 + 4Jmgl cos θq )/2J. On the other hand, if cos θq ≥ 0, one root
of (Jm + ml2 )s2 + bs − mgl cos θq = 0 is either on the imaginary axis or in the right hand side
of the complex plane. In this case, the system is unstable.

6. If the incremental system is unstable, the steady state value does not exist. If it is stable,
the steady state value is limt→∞ δθ (t) = lims→0 s∆θ (s) = lims→0 s · G(s) · 0.1
s = 0.1G(0) =
−0.1Kt
rad.
mgl| cos θq |R