EE352 Lectures:04 Spring 2023

5 Block diagram representation of LTI systems

A single I/O LTI system with TF model G(s) can be represented by a simple block diagram (BD):

u(t) G(s) y(t)

Given a more complicated block diagram (many interconnected blocks). The following trans-
formations can simplify/reduce the BD to a smaller one between a specific input and output.

Block Reductive transformations

1. Series (cascade) transformation

R(s) G1 (s) G2 (s) Y (s) ⇒ R(s) G1 (s)G2 (s) Y (s)

2. Parallel transformation

G1 (s)

R(s) b
Y (s) ⇒ R(s) G1 (s) + G2 (s) Y (s)

G2 (s)

3. Feedback (negative) transformation

. G(s)
R(s) G(s) b
Y (s) ⇒ R(s) 1+H(s)G(s) Y (s)
−
Closed loop transfer function

H(s)

A standard (negative) feedback Control system’s block diagram:

Ea (s)
R(s) Gr (s) Gc (s) Gp (s) b
Y (s)
−

H(s)

Gc (s)Gp (s)
• 1+Gc (s)Gp (s)H(s) is the Closed-loop transfer function
G (s)G (s)
• T (s) = Y(s)/R(s) = Gr (s) 1+Gcc(s)Gpp(s)H(s) : is the total transfer function

Dr.Guernane 16
EE352 Lectures:04 Spring 2023

• Gr (s): is the prefilter(input transducer) transfer function

• Gc (s): is the controller/compensator transfer function

• Gp (s): is the plant transfer function

• H(s): is the sensor,feedback-path transfer function

Y(s)
• G(s) = Ea (s) = Gc (s)Gp (s) is the forward-path transfer function F-P TF.

• Gc (s)Gp (s)H(s): is the open-loop transfer function or loop gain.

• If Gr (s) = H(s) = 1, the control system is called unity feedback, otherwise it is called
non-unity feedback.

MATLAB tips:

1. There are commands to compute the equivalent of a series, parallel or feedback systems.
after defining (creating) them you can do:

>> G12=series(G1,G2) % G12 is the equivalent of G1 in series with G2

>> G12=parallel(G1,G2) % G12 is the equivalent of G1 in parallel with G2
>> G12=feedback(G1,G2) % equivalent of G1 (forward) with G2( negative feedback)

2. For example the total TF T (s) of the above block diagram can be computed in matlab using
the compound statement:

>> T=series(Gr,feedback(series(Gc,Gp),H)))

Dr.Guernane 17
EE352 Lectures:04 Spring 2023

A feedback control system subject to disturbance (Multiple input system):

D(s)

R(s) G1 (s) G2 (s) b

Y (s)
−

H(s)

Using superposition:

G1 (s)G2 (s) G2 (s)

Y(s) = YR (s)|D(s)=0 + YD (s)|R(s)=0 = R(s) + D(s)
1 + G1 (s)G2 (s)H(s) 1 + G1 (s)G2 (s)H(s)

But how to compute The transfer function of more complicated block diagrams when there are no
reductive transformation to apply directly? Consider the:

Block rearranging transformations

1. Moving a branch point forward

– X(s) b
G(s) Y (s) ⇐⇒ –
X(s) G(s) b
Y (s)

1
V (s) V (s) G(s)

2. Moving a Branch point backward

– X(s) G(s) b
Y (s) –
⇐⇒ X(s) b
G(s) Y (s)

V (s) V (s) G(s)

3. Moving a summing point forward

–X(s) G(s) Y (s) ⇐⇒

–
X(s) G(s) Y (s)
V (s)
G(s) V (s)

4. Moving the summing point backward

Dr.Guernane 18
EE352 Lectures:04 Spring 2023

– X(s) G(s) Y (s) –

⇐⇒ X(s) G(s) Y (s)

1
V (s) G(s) V (s)

Example: Using Block diagram transformations, compute the transfer function Y(s)/R(s)?

R(s) G1 (s) b
G2 (s) b
G3 (s) b
Y (s)
− − −
H2 (s) H3 (s)

H1 (s)

Y(s) G1 (s)G3 (s)[1 + G2 (s)]

T (s) = =
R(s) [1 + G2 (s)H2 (s) + G1 (s)G2 (s)H1 (s)][1 + G3 (s)H3 (s)]

Chapter highlights:

• Mathematical models approximate the physical relationship between the a system output
and input.

• Generally, the more accurate the model, the more complicated it is.

• A LTI system can be represented by LCCDEs, impulse responses, or transfer functions.

• All these models are input/output (or external) representations since they relate the output
to the input without any intermediate (or internal) variables.

• LCCDEs and impulse responses are time-domain representations.

• Transfer functions are s-domain (Laplace) representations

• Transfer functions models are more attractive to use in block diagrams.

• Complex block diagrams can be simplified to one block by using both reductive transforma-
tions (series, parallel, feedback) and rearranging transformations.

Dr.Guernane 19
Chapter III

Performance and characteristics of

feedback control systems

How to compare, or evaluate control systems?:

1. Stability: required for the system to be useful

2. Input tracking: the output should be as close to the input (desired output) as possible

(a) Steady-state tracking: reduce the error between input and output in steady-state
(b) Transient tracking: reduce the error between input and output at the transient-phase.

3. Insensitivity to system parameter: The system should be the least sensitive to parameter
changes or uncertainty.

4. Disturbance rejection: The system should be reject undesirable disturbance as much as

possible.

1 Stability of LTI systems

A fundamental requirement of any useful control system is stability.

Theorem 2. A LTI system is said to be (BIBO) stable if its impulse response is absolutely integrable:
Z +∞
|h(τ)| dτ < M
−∞

Let the LTI system be described by its transfer function H(s) = L(h(t)). What is the condition
on H(s) for the system to be stable?
X
h(t) = mode(pi )
i

Where pi is the pole of H(s). a mode is the time-response (or ILT) that correspond to a pole term in the PFE
of H(s), For instance e−2t , us (t) and et are the mode that corresponds to the pole −2, 0 and 1 respectively.
Condition:
modet→∞ (pi ) → 0 ∀pi

Theorem 3. A LTI system is (BIBO) stable if and only if the all the poles of H(s) lie in the LHP (σ < 0)

20
EE352 Lectures:04 Spring 2023

Notice How the pole location determine whether the corresponding time response (or mode)
is absolutely integrable or not.
In the following figure, its can be seen that only poles in the LHP results in absolutely integrable
modes.

Typical pole locations and the corresponding modes

Examples : Which of the following systems (represented by TF) is stable?

s−1
1. G(s) = (s+1)2 (s+3)
stable
s+21
2. G(s) = (s+1)(s2 +2)
unstable
2
3. G(s) = (s−1)(s+1)(s+2) unstable

Dr.Guernane 21
EE352 Lectures:04 Spring 2023

Definition: A system is said to be Marginally unstable (or marginally stable) iff it is an unstable system
with no poles in the RHP and no repeated poles on the jω axis

Examples:
s−1
1. G(s) = s(s+3) marginally stable
s−1
2. G(s) = (s+2)(s+3) stable
s−1
3. G(s) = (s−2)(s+3)(s2 +1)
unstable
s+21
4. G(s) = (s+1)(s2 +2)
marginally unstable
2
5. G(s) = s2 (s2 +2)
unstable

Marginally unstable systems can be used in practice with care

Example The integrator

1
Vo (s) = Vi (s) RC = 1
RCs
• its transfer function is 1/s, is unstable (strictly speaking),meaning there exists a bounded
input that produces an unbounded output: like a step input (us (t) ≤ 1 ∀t), but the output
Y(s) = 1/s2 is the unit ramp t which is not bounded).

• However, if one applies a sine (bounded) you get a cosine (bounded).

R
−

+ +
−vi (t) − vo (t)

Dr.Guernane 22