Exercise Set 6 - Control Systems (5ESD0) Sept 2021

1. Loop-shaping:
Consider a system with the transfer function
5
G(s) = .
(s + 1)(s + 10)

(a) Draw the Bode plot of G(s).

(b) Design a controller D(s) by hand, that satisfies the following requirements:
ˆ zero steady-state error when the reference is a step function;
ˆ phase margin PM ≥ 60◦ ;
ˆ closed-loop bandwidth ωbw ≥ 12 rad/sec.
(c) Calculate and plot using Matlab, the Bode plot for the sensitivity function S(s) for the com-
pensated system. Determine in which frequency domain disturbances are reduced by at least
1/2.

Solution:
For having zero steady-state error in response to a step reference, the open-loop transfer function
should have at least 1 pole at the origin, that is, the open-loop system must be at least of type
1. Since this is not the case for G(s), we have to place at least 1 pole at the origin for D(s). For
example, let us start from Dpi (s) = K s+1 s so that one stable pole of G(s) is canceled out, and
5
L(s) = Dpi (s) G(s) = K s (s+10) . The Bode plot for L(s) is in Figure 1, which shows P M ≃ 87◦
and |L(jω)| ≥ 0 dB for all ω ≤ 0.499.
 
5
Note that for ωbw = 12 rad/sec we have |L(jωbw )| =  j12(j12+10)  ≈ −31.5 dB. Thus we shall
 
increase the gain K and we can select K = 40 ≃ 32 dB. The resulting Bode plot is shown in Figure
2. It now follows that ωc = 12.5 rad/sec, hence ωbw > 12 rad/sec as required, but PM ≃ 38◦ < 60◦ .
Next we are going to add ϕmax = 30◦ via an additional lead compensator, that is, 1+T s
1+αT s , for some
1−sin(ϕmax )
0 < α < 1 and T > 0. By the standard formula, we select α = 1+sin(ϕmax )= 1/3. Since we want to
√
have the maximum phase increase ϕmax at frequency ωc = 12.5, we use the equation ωmax T = 1/ α,
with ωmax = ωc , which leads to T ≃ 0.1386. In this way, we have changed the gain at ωc to about
4.79 dB, so we have to adjust the open-loop gain of additional Klead = 0.5761 ≃ −4.79 dB. Thus,
we obtain the lead compensator Dlead (s) = Klead 1+0.1386s
1+0.0462s , and the final open-loop transfer function
reads as
1 + 0.1386s 5
Lfinal (s) = (Dlead (s)Dpi (s)) G(s) = 23.044 · · .
1 + 0.0462s s(s + 10)

Figure 3 shows that all the specifications are now met.

For the sensitivity, we have
1
S(s) = ,
1 + D(s) G(s)
whose Bode plot is in Figure 4. We note that disturbances below approximately ω0 = 6 rad/sec are
attenuated by at least 1/2 since |S(jω0 )| ≃ 1/2 ≃ −6 dB and |S(jω)| ≤ −6 dB for all ω ≤ ω0 .

1
Exercise Set 6 - Control Systems (5ESD0) Sept 2021

5
Figure 1: Bode plot for s(s+10) .

200
Figure 2: Bode plot for s(s+10) .

2
Exercise Set 6 - Control Systems (5ESD0) Sept 2021

Figure 3: Bode plot for Lfinal (s).

Figure 4: Bode plot for the sensitivity function 1/(1 + Lfinal (s)).

2. Problem 6.59 in the Franklin book:

Golden Nugget Airlines had great success with their free bar near the tail of the airplane. However,
when they purchased a much larger airplane to handle the passenger demand, they discovered that
there was some flexibility in the fuselage that caused a lot of unpleasant yawing motion at the
rear of the airplane when in turbulence and was causing the travelers to spill their drinks. The
approximate transfer function for the Dutch roll mode (see Section 10.3.1) is

r(s) 8.75 (4s2 + 0.4s + 1)

= ,
δr (s) (100s + 1)(s2 + 0.24s + 1)

where r is the airplane yaw rate and δr is the rudder angle.

In performing a Finite Element Analysis (FEA) of the fuselage structure and adding those dynamics
to the Dutch roll motion, they found that the transfer function needed additional terms that reflected
the fuselage lateral bending that occurred due to excitation from the rudder and turbulence. The

3
Exercise Set 6 - Control Systems (5ESD0) Sept 2021

revised transfer function is

r(s) 8.75 (4s2 + 0.4s + 1) 1
G(s) = = · ,
δr (s) (100s + 1)(s2 + 0.24s + 1) (s2 /ωb2 + 2ζs/ωb + 1)
where ωb = 10 rad/sec is the frequency of the bending mode and ζ = 0.02 is the bending mode
damping ratio.
Most swept wing airplanes have a “yaw damper” which essentially feeds back the measured yaw
rate by a rate gyro to the rudder with a simple proportional control law. For the new Golden
Nugget airplane, the proportional feedback gain is K = 1.
Remark: You may use Matlab for this exercise.

(a) Draw the Bode plot for the open-loop transfer function, determine the PM and GM for the
given controller design (D(s) = K = 1).
Plot the step response and the Bode magnitude of the closed-loop system. Which is the
frequency of the lightly damped mode that is causing troubles?
(b) Investigate remedies to quiet down the oscillations, but maintain the same low frequency gain
in order not to affect the quality of the Dutch roll damping provided by the yaw-rate feedback.
Specifically, investigate one at a time:
i. Increasing the damping of the bending mode from ζ = 0.02 to ζ̂ = 0.04 (which would
require adding energy absorbing material in the fuselage structure)
ii. increasing the frequency of the bending mode from ωb = 10 rad/sec to ω̂b = 20 rad/sec
(which would require stronger and heavier structural elements)
iii. adding a low-pass filter in the feedback loop, that is, replace the linear controller u(e) = K e
with some controller
1
Dlpf (s) = K
sτ + 1
pick τ so that the objectionable features of the bending mode are reduced while maintain-
ing PM ≥ 60◦ ;
iv. adding a notch filter in the feedback loop, that is, replace the linear controller u(e) = K e
with some controller

s2 /ωb2 + ζs/ωb + 1
Dnf (s) = 2
(1 + s/100)
pick the frequency of the notch zero to be at ωb with a damping of ζ = 0.04 and pick the
denominator poles to be (s/100 + 1)2 keeping the DC gain of the filter equal to 1.
(c) Investigate the sensitivity of the two compensated designs above (iii, iv) by determining the
effect of a reduction in the bending mode frequency of 10%. Specifically, re-examine the two
designs by looking at the GM, PM, closed-loop bending mode damping ratio and resonant
peak amplitude, and qualitatively describe the differences in the step response.
(d) What would you recommend to Golden Nugget to help their customers stop spilling their
drinks? (Make the recommendation in terms of improvements to the yaw damper.)

Solution:

(a) The Bode plot of the open-loop transfer function is shown in Figure 5. Note that GM ≃ 1 dB
is quite small due to the resonance effect at ω = 10 rad/sec which almost leading to instability.
G(s)
The step response of the closed loop system Gcl (s) = 1+G(s) is shown in Figure 6. From the
Bode plot of Gcl shown in Figure 7 we note that the frequency of the poorly damped mode is
ω = 10 rad/sec.
(b) i. The Bode plot for the open-loop transfer function with the bending mode damping in-
creased from ζ = 0.02 to ζ̂ = 0.04 is shown in Figure 8. We can see that the GM has
increased because the resonant peak is well below magnitude 1; thus the system would be
much better behaved.
ii. The Bode plot of the open-loop transfer function with ωb = 10 rad/sec replaced by ω̂b = 20
rad/sec is shown in Figure 9. Again, we note that the GM has increased and the resonant
peak has reduced.

4
Exercise Set 6 - Control Systems (5ESD0) Sept 2021

iii. We select τ = 1 as parameter of the low-pass filter and obtain the Bode plot shown in
Figure 10. Stability margins are satisfactory.
iv. The Bode plot of the open-loop transfer function with the given notch filter is shown in
Figure 11. We obtain PM = 97.5 (at 0.085 rad/sec) and GM = 55.1 (at 99.7 rad/sec).
These margins are the largest among all the considered control designs. The notch filter
has essentially cancelled the bending mode resonant peak out.
(c) Note that the notch filter is very sensitive to where to place the notch zeros to reduce the lightly
damped resonant peak. So if you want to use the notch filter, you must have a good estimation
of the location of the bending mode poles and the poles must remain at that location for all
aircraft conditions. On the other hand, the low pass filter is relatively robust to where to place
its break point.
Evaluation of the margins with the bending mode frequency lowered by 10%, that is, to
ω̃b = 9 rad/sec, will show a drastic reduction in the margins for the notch filter and very little
reduction for the low pass filter.
Let G̃(s) be the system transfer function with bending mode frequency ω̃b . Figures 12, 13
show the Bode plots of the two corresponding open-loop transfer functions.
(d) While increasing the natural damping of the system would be the best solution, it might be
difficult and expensive to carry out. Likewise, increasing the frequency typically is expensive
and makes the structure heavier, not a good idea in an aircraft.
Of the remaining two options, it might be better to go with the low pass filter since the exact
location of the bending mode might not be known but rather an estimate, choosing the low
pass filter is the more robust approach.

Figure 5: Bode plot for the open-loop transfer function in Problem 2.

5
Exercise Set 6 - Control Systems (5ESD0) Sept 2021

Figure 6: Step response for the closed-loop system with linear control law.

Figure 7: Bode plot for the closed-loop transfer function in Problem 2 with control law D(s) = K = 1.

6
Exercise Set 6 - Control Systems (5ESD0) Sept 2021

Figure 8: Bode plot for the open-loop transfer function in Problem 2 with damping parameter ζ̂ = 0.04
(rather than ζ = 0.02).

Figure 9: Bode plot for the open-loop transfer function in Problem 2 with resonance frequency ω̂b = 20
(rather than ωb = 10).

7
Exercise Set 6 - Control Systems (5ESD0) Sept 2021

1
Figure 10: Bode plot for the open-loop transfer function Dlpf (s) G(s) = 1+s G(s) in Problem 2.

Figure 11: Bode plot for the open-loop transfer function Dnf (s) G(s) in Problem 2.

8
Exercise Set 6 - Control Systems (5ESD0) Sept 2021

Figure 12: Bode plot for the open-loop transfer function Dlpf (s) G̃(s) in Problem 2.

Figure 13: Bode plot for the open-loop transfer function Dnf (s) G̃(s) in Problem 2.

9
Exercise Set 6 - Control Systems (5ESD0) Sept 2021

3. Sensor noise and output disturbance:

Consider the block diagram in Figure 14, where
1
G(s) = s

s

.
s 1+ 5 1+ 20

Design a compensator D(s) such that the closed-loop system satisfies the following specifications:
(a) the noise n introduced with the sensor signal at frequencies higher than 200 rad/s are to be
attenuated at the output by at least a factor of 100 (−40 dB);
(b) the disturbance d at the plant output acting at up to 10 rad/s shall be attenuated of at least
a factor 2 (1/2 ≃ −6 dB).

Figure 14: Block scheme for Problem 2.

Solution:
We look at the Bode plot of the open-loop transfer function G(s), see Figure 15, and at the sensitivity
plots for D(s) = 1, see Figure 16. From the Bode plot, we note that PM ≃ 76◦ at ωc ≃ 1 rad/s.
Since disturbances d up to 10 rad/s must be attenuated, we shall increase the bandwidth of the
system. One possible way of doing so is canceling the stable pole at −5 with the compensator
1 + s/5
D(s) = K ,
1 + s/50

where we tune K = 40 for achieving PM ≃ 15◦ at ωc = 23.5 rad/s, see Figure 17. Next we plot the
sensitivity functions for the compensated open-loop transfer function D(s)G(s) in Figure 18. We
note that |S(jω)| ≤ −9 dB for all ω ≤ 10 rad/s, and that |T (jω)| ≤ −45 dB for all ω ≥ 200 rad/s.

Figure 15: Problem 3: Bode plot for G(s).

10
Exercise Set 6 - Control Systems (5ESD0) Sept 2021

1 G(s)
Figure 16: Problem 3: Sensitivity functions S(s) = 1+G(s) , T (s) = 1+G(s) .

Figure 17: Problem 3: Bode plot for D(s)G(s).

(Another) Solution:
We compute the transfer function from d to y. y(s) = d(s) − D(s)G(s)y(s), hence y(s)/d(s) =
1
1+D(s)G(s) . From the low-frequency specification, we need that |1 + D(jω)G(jω)| > 2 for all ω ≤ 10
rad/s. The worst case situation is characterized by D(jω)G(jω) = −3. Thus, we shall design
|D(jω)G(jω)| > 3 for all ω ≤ 10 rad/s. First, we note that |D(jω)G(jω)| > 3 for all ω ≤ 10 rad/s
requires ωc > 10. The bode diagram for G(s) in Figure 15 shows us that we need to increase ωc ,
but also that we need to add phase in this frequency region in order to achieve a stable closed-loop
interconnection. For this reason a lead compensator will be designed.
T s+1
Let us then consider a lead compensator D(s) = K αT s+1 , for adding some phase (margin) at
ωmax = 10 rad/s. Let us choose α = 1/8√(this is a free choice, since no requirement on the PM
1 √1
is specified). By the formula ωmax T = 1/ α, we obtain T = ωmax α
= 0.2828. The last step for
the lead compensator design is to calculate K, which will be used to achieve |D(jω)G(jω)| > 3 for
all ω ≤ 10 rad/s. Observe that |D(jω)G(jω)| is decreasing for increase ω and therefore we require
that |D(j10)G(j10)| > 3, which gives us K > 26.52.

11
Exercise Set 6 - Control Systems (5ESD0) Sept 2021

1 D(s)G(s)
Figure 18: Problem 3: Sensitivity functions S(s) = 1+D(s)G(s) , T (s) = 1+D(s)G(s) .

−D(s)G(s)
We can compute the transfer function from n to y, which is y(s)/n(s) = 1+D(s)G(s) = −T (s). The
1
requirement specifies that |T (jω)| < 100 for all ω ≥ 200 rad/s. Observe that we must have that
1
|D(jω)G(jω)| ≪ 1. Therefore T (jω) can be approximated by |D(jω)G(ω)| < 100 for all ω ≥ 200
rad/s. Using this approximation we can transform our closed-loop noise requirement to a open-loop
requirement on |D(s)G(s)| again.
By choosing K = 26.52 the low-frequency specification is met and at the same time also the high-
frequency one, see Figure 19. For this reason a single lead compensator is sufficient to meet al
requirements. If the high-frequency specification was not met, then we could for example add a
pole, or a lag compensator.

1 D(s)G(s)
Figure 19: Problem 3 (2nd solution): Sensitivity functions S(s) = 1+D(s)G(s) , T (s) = 1+D(s)G(s) .

12
Exercise Set 6 - Control Systems (5ESD0) Sept 2021

4. Non-minimum phase system:

Consider the system with transfer function
s−2
G(s) =
(s + 1)(s + 3)

Design via SisoTool a controller D(s) such that the closed-loop system is stable and has “as high
as possible” bandwidth. Also describe what happens to the sensitivity function as you increase the
bandwidth.
Hint: Try (1) a proportional controller with positive gain, (2) a proportional controller with nega-
tive gain, and (3) a proportional-integral controller.

Solution:
(1) D(s) = Kpos (proportional controller)

The root locus shows that Kpos ∈ (0, 1.49) are the positive values that stabilize the closed-loop
system, see Figure 20. We select Kpos = 1.49 for the almost maximum bandwidth achievable with
proportional control, see the Bode plots in Figure 21.
(2) D(s) = Kneg (proportional controller)

On the other hand, a negative gain, e.g. Kneg = −3.99, would cancel the negative static gain in
G(s), ensure closed-loop stability and high bandwidth, see Figures 22 and 23.

(3) D(s) = K (s+p)

s (proportional-integral controller)

We choose a negative gain K for the controller to cancel out the negative gain in G(s), and a zero
at s = −1 near the stable pole (e.g. p = 1). One possible design is then D(s) = −2.99 (s+1)
s . See
Figure 24 for the corresponding root locus, and Figure 25 for the compensated Bode plot.

Observing the sensitivity plots for controllers (2), Figure 26, and (3), Figure 27 ,we note that as
we increase the bandwidth the peak in the sensitivity function increases in magnitude which can
severely effect robustness of our system. For controller (1), Figure 28, we see the low frequent part
of our sensitivity function increases which is also not desirable.
As we note from this exercise, RHP zeros pose a fundamental limitation on the bandwidth of the
system since the locations of our closed-loop poles move to the open-loop zeros. Since we cannot
cancel these zeros, we will always have a limitation on our gain and thus our bandwidth.

13
Exercise Set 6 - Control Systems (5ESD0) Sept 2021

Figure 20: Problem 4: Root Locus for G(s).

Figure 21: Problem 4: Bode plot for Kpos G(s).

14
Exercise Set 6 - Control Systems (5ESD0) Sept 2021

Figure 22: Problem 4: Root Locus for −G(s).

Figure 23: Problem 4: Bode plot for Kneg G(s).

15
Exercise Set 6 - Control Systems (5ESD0) Sept 2021

s+1
Figure 24: Problem 4: Root Locus for s G(s).

Figure 25: Problem 4: Bode plot for DP I (s)G(s).

16
Exercise Set 6 - Control Systems (5ESD0) Sept 2021

Bode Diagram
60

50

40
Magnitude (dB)

30

20

10

-10

-20
10-2 10-1 100 101 102
Frequency (rad/s)

Figure 26: Problem 4: Sensitivity function for Kneg G(s).

Bode Diagram
60

50

40

30
Magnitude (dB)

20

10

-10

-20

-30
10-1 100 101 102
Frequency (rad/s)

Figure 27: Problem 4: Sensitivity function for DP I (s)G(s).

17
Exercise Set 6 - Control Systems (5ESD0) Sept 2021

Bode Diagram
45

40

35

30
Magnitude (dB)

25

20

15

10

-5
10-4 10-3 10-2 10-1 100 101 102
Frequency (rad/s)

Figure 28: Problem 4: Sensitivity function for Kpos G(s).

5. Robustness margin:
Consider the block diagram in Figure 14, where
1000
G(s) = .
s2 (s2 + 10s + 3002 )

(a) Design a stabilizing controller D(s) such that the crossover frequency ωc ≥ 40rad/s and dis-
turbances are never amplified more than 6dB for all frequencies.
(b) What is the response of the system if
ˆ d(t) = sin (10t)
ˆ d(t) = sin (50t)
(c) What is the worst case response if d(t) = sin (f t) with f ∈ R+

Solution:

(a) We start by evaluating the bode plot of G(s), Figure 29. We note that at ω = 40 rad/s the
magnitude is -103dB, hence we increase the gain with 103 dB, e.g. K = 1.4125 · 105 , resulting
in the bode plot given in Figure 30.
To render the system stable we design a lead filter in the usual matter with ωmax = 40 rad/s,
ϕmax = 35◦ . This gives a stable system with a crossover frequency ωc ≥ 40 rad/s, Figure 31.
Disturbances may never be amplified more than 6dB, hence the sensitivity function cannot go
over 6dB. Looking at the peak of the sensitivity function, e.g. the worst case amplification,
we see that with the designed controller suffices the criteria, Figure 32.
(b) We see from the sensitivity function that for a 10 rad/s disturbance the amplification is ap-
proximately -18 dB which is an amplification of 0.1259. For a disturbance with frequency of
50 rad/s the amplification is approximately 5 dB which is an amplification of 1.7783.
(c) We designed our controller such that the sensitivity function is always below 6 dB, in fact, the
highest amplification for the designed system is 5.45 dB at 291 rad/s. Hence our worst case
amplification is 1.8728.

18
Exercise Set 6 - Control Systems (5ESD0) Sept 2021

Figure 29: Problem 5: Bode plot for G(s).

Figure 30: Problem 5: Bode plot for K G(s).

19
Exercise Set 6 - Control Systems (5ESD0) Sept 2021

Figure 31: Problem 5: Bode plot for K Dlead G(s).

Bode Diagram
10

-10
Magnitude (dB)

-20

-30

-40

-50

-60
100 101 102 103
Frequency (rad/s)

Figure 32: Problem 5: Sensitivity plot for K Dlead G(s).

20
Exercise Set 6 - Control Systems (5ESD0) Sept 2021

6. Disturbance reduction
Given the following loopgains
ˆ L1 (s) = 2
s2 +3s

ˆ L2 (s) = 2−2s
s2 +3s

ˆ L3 (s) = 2+2s
s2 +3s

ˆ L4 (s) = 4+4s
s2 −3s

For every loopgain

(a) Plot the sensitivity function and verify the Bode sensitivity integral
(b) Predict what the response of the system will be if it is excited by a disturbance signal

d(t) = sin(2t) + sin(0.1t)

(c) Verify your prediction using Matlab (Hint: you can use the function lsim.m)

Solution:
The sensitivity function is given by
1
S(s) =
1 + L(s)
Plotting the bode diagram of the sensitivity function gives Figure 33.

Bode Diagram
20
L_1
L_2
10
L_3
L_4
0
Magnitude (dB)

-10

-20

-30

-40

-50
10-2 10-1 100 101 102
Frequency (rad/s)

Figure 33: Sensitivity function for Loopgains L1 to L4 .

The bode sensitivity integral is given by

Z ∞ X π
ln |S(jω)|dω = π Re (pk ) − lim sL(s)
0 2 s→∞

where pk are the open loop RHP poles. Starting with L1 , we see this transfer function has no open
loop RHP poles and that it has two more poles than zeros, e.g. relative degree 2. We expect the

21
Exercise Set 6 - Control Systems (5ESD0) Sept 2021

integral to equal zero. Computing the integral for L1 gives

Z ∞ X π
ln |S(jω)|dω = π Re (pk ) − lim sL(s)
0 2 s→∞
π 2s
= 0 − lim 2
2 s→∞ s + 3s
π 2
= − lim
2 s→∞ s + 3
=0

L2 has a non-minimum phase zero and no RHP poles, we expect this function to equal a positive
constant indicating a peak in the Sensitivity function. Computing the integral for L2 gives
Z ∞ X π
ln |S(jω)|dω = π Re (pk ) − lim sL(s)
0 2 s→∞
π −2s(s − 1)
= − lim
2 s→∞ s(s + 3)
π −2s + 2
= − lim
2 s→∞ s + 3
π −2s
= − lim
2 s→∞ s
π
=2
2
=π

L3 has no RHP poles or zeros but is of relative degree 1, hence we expect it to equal a negative
constant indicating the sensitivity function does not peak.
Z ∞ X π
ln |S(jω)|dω = π Re (pk ) − lim sL(s)
0 2 s→∞
π 2s(s + 1)
= − lim
2 s→∞ s(s + 3)
π 2(s + 1)
= − lim
2 s→∞ s + 3
π 2s
= − lim
2 s→∞ s
π
= −2
2
= −π

L4 has one RHP pole and relative degree 1, hence we expect the Bode sensitivity integral to equal
a positive constant again indicating a peak in the sensitivity function.
Z ∞ X π
ln |S(jω)|dω = π Re (pk ) − lim sL(s)
0 2 s→∞
π 4s(s + 1)
= 3π − lim
2 s→∞ s(s − 3)
π 4(s + 1)
= 3π − lim
2 s→∞ s − 3
π
= 3π − 4
2
=π

To predict the response we look at the magnitude of the sensitivity function at the excited frequen-
cies. Starting with L1 we see that signals with a frequency of 0.1 rad/s are severely reduced, namely
by -16.5 dB, and we see a slight amplification for signals with a frequency of 2 rad/s. Plotting the
response confirms this, see Figure 34.

22
Exercise Set 6 - Control Systems (5ESD0) Sept 2021

2
Disturbance
System response
1.5

0.5

-0.5

-1

-1.5

-2
0 10 20 30 40 50 60 70 80 90 100
Time [s]

Figure 34: System response with loopgain L1 to the given disturbance signal.

3
Disturbance
System response
2

-1

-2

-3
0 10 20 30 40 50 60 70 80 90 100
Time [s]

Figure 35: System response with loopgain L2 to the given disturbance signal.

23
Exercise Set 6 - Control Systems (5ESD0) Sept 2021

Observing the sensitivity function of L2 , we see again that signals with a frequency of 0.1 rad/s are
reduced, but signals of 2 rad/s are amplified by 8 dB. This again can be verified by the response
given in Figure 35
Now taking a look at the sensitivity function for L3 we see that both excited frequencies are reduced.
This can again be verified by the system response shown in Figure 36

2
Disturbance
System response
1.5

0.5

-0.5

-1

-1.5

-2
0 10 20 30 40 50 60 70 80 90 100
Time [s]

Figure 36: System response with loopgain L3 to the given disturbance signal.

Lastly, looking at the sensitivity function for L4 , we see a reduction for signals with frequency of
0.1 rad/s. The sensitivity function peaks at 2 rad/s, hence signals with a frequency of 2 rad/s are
heavily amplified with a gain of 11 dB. This also shows from the response given in Figure 37.

24
Exercise Set 6 - Control Systems (5ESD0) Sept 2021

4
Disturbance
System response
3

-1

-2

-3

-4
0 10 20 30 40 50 60 70 80 90 100
Time [s]

Figure 37: System response with loopgain L4 to the given disturbance signal.

25