Republic of the Philippines

Tarlac State University

COLLEGE OF ENGINEERING AND TECHNOLOGY
Department of Mechanical Engineering
Romulo Blvd, Tarlac City, Tarlac

CLASSICAL CONTROL TECHNIQUES

A Report Paper Submitted to Engr. Marc Florenz P. Arnaldo

In Partial Fulfilment of the Requirements of the Course Subject

MEF382:
Control Engineering – Lecture

by Group No. 4:

FIDER, Christian
RODRIGUEZ, Jayson Rei
TIPAY, Geremie
BALABAG, Joseph Christopher
ROSARIO, Ralph Vincent
TORRES, Dienies
BRIONES, Azhar Iman
GOH, Qing Mie .
GUTIERREZ, Jamea Lei

Bachelor of Science in Mechanical Engineering

A.Y. 2021 – 2022

 Report Paper No. 2:

Classical Control Techniques

This topic is about the classical control theory, a strategy for developing control systems, in this
chapter. During the early phases of the development of feedback and control engineering, in the
1930s and 1940s, The techniques created in the 1950s are still crucial for designing control
systems, particularly for single-input, single-output linear systems. Classical findings serve as
the foundation for much of the contemporary notion of robust control.

The classical field of electrical engineering started with the work of H. Nyquist, H. S. Black, H.
W. Bode, and W. R. Evans. These techniques have some heuristic characteristics that explain
both their strengths and their weaknesses, These methods are used by designers to build a
compensation network or controller that will assist the closed-loop system in meeting its
performance objectives. The terms then used came from the field of amplifier circuit design
(Boyd and Barratt, 1991).

We will examine some of the traditional techniques for situational control in this chapter. The
presentation does not claim to be complete. Both classical and contemporary textbooks include
comprehensive introductions, as well as a wealth of supplementary information. In 1945, Bode
wrote that James et al.'s work was important. (1947), Evans (1954), Truxall (1955), Savant
(1958), Horowitz (1963), Ogata (1970), Thaler (1973), Franklin et al.(1986), D'Azzo en Houpis
(1988), Van de Vegte (1990), Franklin et al. There are many sources that support the idea that
1991 was a significant year. Dorf is one of these sources.

We go through the error characteristics of feedback control systems in a steady state in section
2.2, In Chapter 2, the concept of integrated control is thoroughly examined.

In classical control theory, frequency domain techniques are emphasized most, The Bode,
Nyquist, and Nichols plots, as well as other traditional visual representations of frequency
responses, are examined in this section.

In 2.5, the design goals and standards of classical control theory are taken into account while
Section 2.6 discusses the fundamental, traditional lead, lag, and lag lead corrective approaches.
In 2.7, a quick summary of the root locus theory as a technique for choosing compensator
parameters is provided in Section 2.8 describes the historically fascinating Guillemin-Truxal
design method. In the 1970s, Horowitz (1982) invented the quantitative feedback theory (QFT)
while 2.9 provides an explanation for this effective evolution of the traditional frequency domain
feedback design process.
Every design technique is model-based, they rely on implicit and explicit models of controllable
plants. The experimental Ziegler-Nichols rule for adjusting the PID controller, which is given in
2.3, is the exception.

Steady State Error Behavior

The fact that the operation of the control system reduces steady-state errors is one of the main
reasons for introducing feedback control into the system. Figure 2.1 depicts a typical single-loop
control system. When the reference input r is a polynomial time function of degree n, asymptotic
behavior in the time domain of t → ∞ is seen. consequently,

tn
r ( t )= l ( t ) , t ≥ 0.
n!

1 (t) = 1 (for t ≥ 0) and 1 (t) = 0 (for t & lt;) 0, where 1 is a unit step function. There is a unit
amplitude step if n = 0. The input is a ramp with a unit gradient if n = 1, and a parabola with a
second derivative of the unit if n = 2.

The Laplace transform of the reference input is rˆ (s) = 1 / sn + 1, Control deviation is defined
by signal ε

ε ( t )=r ( t )−z ( t ) , t ≥ 0.

If a consistent mistake exists,

ε ∞=lim ε (t )
t →∞

Laplace transform of control error,

1 H (s ) 1−H ( s)
εˆ ( s ) =r s− y s= n+1
− n+1
=
s s sn +1

the function
 PCF L
H= = F
1+ PC 1+ L

a closed-loop transmission mechanism. The loop gain is L = PC.

All H poles should be in the system's left side if the closed-loop model is stable.

At the level, you can use the final value theorem of the Laplace transform theory. Therefore, if
there is a steady-state error, it is given as follows:

1−H (s)
ε (n)
∞ =lim ,
s↓0 sn

where n is the degree of the reference input polynomial. Let's assume at first that the prefilter is
not installed. F (s) Equals 1 Then

1
1−H ( s )=
1+ L( s)

In addition, steady-state error

(n) 1
ε ∞ =lim n
¿
s↓0 s ¿¿

This equation may be used to determine the steady-state error of the response of the control
loop system to the reference input polynomial.

Type k systems

k
If the limit lim s ↓ 0 s L(s) exists and is non-zero for an integer k, then the closed system of loop
.
gain L is type k. However, in order for the system to be k-shaped, the loop gain L at the origin
must be exactly the k pole. In a type k system

{
¿ ∞ for 0 ≤ n< k ,
n
lim s L(s) ≠ 0 for n=k
s↓ 0
¿0 for>k .

A system of type k from (2.8) to F (s) = 1 without prefiltering is the end outcome.

{
¿ ∞ for 0 ≤ n<k ,
(n )
lim ε ∞ ≠ 0 for n=k
s↓ 0
¿0 for >k .

If the system is stable and of type k, there is a zero steady error for inputs less than k, a finite
non-zero steady error for inputs k, and an infinite steady error for inputs higher than k. For
ramp and higher-order inputs, Type 0 systems have infinite steady-state errors; however, step
reference inputs have non-zero steady-state errors.

k
A closed-loop system has a loop gain L, a limit lim s ↓ 0 s L(s) , and if it is an integer k, it will be
.
type k if it is non-zero. The loop gain L at the origin must be precisely k poles for the system to
be of type k. In a type k system
Type 1 systems have infinite steady-state errors for secondary and higher inputs, zero steady-
state errors for secondary inputs, and finite steady-state errors for ramp inputs. The steady-state
errors of the step and ramp inputs in a Type 2 system are zero, those of the second-order input
are finite, and those of the third-order input and above are infinite.

Figure 2.2 shows the relationship between system type and steady-state error.

Error Constants

The steady-state error is for the step input if the system type is 0.

1 1
ε (0)
∞ = =
1+ L(0) 1+ K P

The constant for position error is K p = L (0).

The steady-state error of the ramp input in a Type 1 system is:

(0) 1
ε ∞ =lim ¿
s↓0 s¿¿

The rate constant is given by the number Kv = lims↓0 sL(s).

The steady-state error of a Type 2 system for a secondary input is:

lim 1
(2) s↓0 1
ε =
∞ =
2
s L(s) 1+ K a
 2
The value K a =lim s L(s ) is the acceleration constant.
s↓0

The numerous steady-state errors are compiled in Table 2.1. The higher the error constant, the
smaller the associated steady-state error.

Depending on its functional purpose, a servo system must conform to certain fundamental
constraints regarding position, velocity, and acceleration faults. The results of steady-state
errors are robust in the sense that the quality of steady-state errors remains zero as The
transfer function’s coefficient of the system changes, unless the type of system changes. As a
result, the speed error constant for Type 1 systems might change because to unpredictability in
system parameters. Until the type 1 property is deleted, the system remains at error zero.

Integral Control

A proven method for reducing low frequency interference is integral control. It also provides
useful resilient properties.

To lessen interference, the loop gain is raised. By adding a factor of 1 / s, the loop gain L (s) = P
(s) C (s) can be made huge at infinite zero and low frequencies. If P (s) does not have a
"natural" factor of 1 / s, the factor is added to the transfer function compensator’s C by
choosing:

C 0 (s)
C ( s) =
s

The logical function C 0(s) was chosen is still adamant. A transfer function C 0(s) and a system
with transfer function 1 / s can be connected in series with compensator C. Because a system
with a compensator of this kind is though to have a integrating activity, and 1/s as an integrator

The system will be type 1 if the coefficient of loop gain L (s) is 1 / s, in accordance with the
wording of 2.2 and there is no steady-state uncertainty in response to the jump reference input.

This shows that the loop gain at L0 ( s). has no coefficient s.

L0 (s)
L(s) =
s

At zero frequencies the continuity is minimal and at low frequencies it is low. It is obvious that
any disturbances at that frequency or constant are fully removed because S is 0 at zero
frequency, It continuously employed denoising.
  A compensator with an integrated effect is simple to construct. Integral control is a
popular design strategy for real-world control systems due to its reliability and efficiency
in reducing low frequency interference. The following versions exist:
 Pure integral control with a transfer function for the compensator

C(s) = g ¿) (2.26)

provides more diversity.

Finally, the transfer function of the compensator serves as the foundation for PID
(Proportional-Integral-Derivative) control.

C ( s ) =g ¿ ), (2.27)

The lead time is T d. Response time can be sped up because bias operation tends to lessen the
impedance of the control loop system to high frequency disturbances. In any case, it increases
high frequency noise greatly, which is undesirable. Therefore, equation uses the "tamed"
discriminant function of the derivative sT d . (2.27).

sTd
(2.28)
1+ sT

T = small time constant.

Commercially accessible standard PID controllers are available, Using the Ziegler and Nichols
concepts, they are frequently altered experimentally in actual settings (for an illustration, see
Franklin et al., 1991). The Ziegler-Nichols rule was created with the presumption that plants'
transfer functions are low-pass filters with good attenuation (Ziegler and Nichols, 1942), The P
controller is linked to the system first for making Ziegler-Nichols adjustments to a P, PI, or PID
controller and the gain of the controller g is raised until there is unattenuated vibration. T 0 is the
vibration time, whereas g0 represents the related amplification. Afterward, the PID parameter
controller is made available by

P-controller: g = 0.5 g0 , T i=∞ , T d=0 ,

PI-controller: g = 0.45 g0, T i=0.85 T 0 , T d=0 ,

PID-controller: g = 0.6 g0 , T i=0.5 T 0 , T d=0.125T 0 .

A peak closed-loop step response to around 0.4 disturbances and approximately 0.25 relative
attenuation may be seen in the related closed-loop system.
Additionally beneficial for non-linear systems is integral control, In Fig. 1's block diagram, In
Figure 2.4 of the block diagram of Figure 1, the plant can support any constant output w 0, and
each constant input u0 has a specific constant-steady-state output w 0.

Frequency Response Plots

The visual depiction of frequency response is a crucial tool for comprehending and researching
the dynamics of control systems in both classical and contemporary control engineering. Three
popular techniques for charting frequency response will be discussed in this section: Bode,
Nyquist, and Nichols.

Take into account a stable system that is linear in time and has a transfer function called L,
inputs u, and outputs y. The sine wave input u (t) = uˆsin (ωt), t ≥ 0 gives the steady state sine
wave output y (t) = yˆsin (ωt + φ), t ≥ 0. Output made by

yˆ=¿, φ=argL(jω).

Gain at frequency ω is represented by quantity |L(jω)| Frequency response function L(jω). A

phase shift is its argument arg L(jω).

L(s) = k( s−z 1) ( s−z 2) … ¿ ¿

Where k is a constant, p1, p2, ···, pn is the system's pole, , z1, z2, ···, zm is a zero. The
amplitude and phase of L(jω) may therefore be calculated for each instance of s = jω by
measuring the length and angle of the vector derived from the pole-zero pattern, as illustrated in
Figure 2.8. The length is suitably multiplied and divided to increase the size of L. Angle addition
and subtraction continue the L phase.

For the value of close to Im pi, the pole pi near the imaginary axis has a small vector length s −
pi. The reason the poles close to the virtual axis contribute to the resonant peak of the
frequency response is because this causes a significant quantity of |L(jω)| to exist at this
frequency.
Bode plots

You can plot the frequency response functions L(jω), “ω ∈ R” using two independent graphs
showing amplitude, frequency and phase vs. frequency. The pair of graphs is known as the
Bode plot of L if the frequency and magnitude scales are both logarithmic. A Bode plot of the
quadratic frequency response function for various relative attenuations is shown in Figure 2.9.

After creating a Bode plot of the rational transfer function using the basic method, the frequency
response function should be expressed in a factorized form. The phase and amplitude
logarithms are:

log ¿ ¿ (2.39)
 m
argL ( jω )=arg k + ∑ arg ⁡( jω−¿ ¿ pi )¿ ¿ (2.40)
i=1

Without performing any mathematics, you may sketch the asymptotes of the terms (2.39) and
(2.40).

A main coefficient, s + ω 0, exists.

{
log | jω+ ω0|≈ log |ω 0| for 0 ≤ ω ≪|ω0| log ω for ω ≫|ω 0|
¿ (2.41)

arg( jω+ ω0 )≈
{arg ( ω 0 ) for 0 ≤ ω ≪|ω 0|
90° for ω ≫|ω0| (2.42)

The double log amplitude plot has a line as its asymptote, Asymptotes are continuous at low
frequencies and High frequency asymptotes have a slope of 10 years per 10 years. The slope is
20 dB per decade if the amplitude is shown in decibels At frequency |ω0|, the intersection of the
amplitude asymptotes happens then At low frequencies, the phase shifts from arg(ω0) (0◦ if ω0
is positive) to 90◦ (or 2 radians) at high frequencies. The linear component's amplitude and
phase curves, as well as its asymptote, are shown in Figure 2.10. (for positive 0).

The optimal format for combining components into quadratic elements is as follows.
 2 2
s +2 ζ 0 ω0 s+ω 0 (2.43)

Asymptotically,

{
log |( jω)2+ 2ζ 0 ω 0 ( jω )+ ω20|≈ 2 log|ω0| for 0 ≤ ω ≪|ω0|2 log ω for ω ≫|ω0|
¿
(2.44)

arg ¿ (2.45)

Once more, the low frequency amplitude asymptote is constant. For high frequency amplitude
asymptotes, the board size diagram shows a straight line with a slope of 20 years (40 dB /
decade) per 10 years. The frequency at which the asymptotes intersect is |ω0|. At low
frequencies the phase is 0 degrees, but at high frequencies it is 180 degrees. Relative
attenuation ζ0 (0 is a positive integer). The amplitude and phase diagram of the different values
are displayed in Figure 2.11. The asymptotic Bode plot of the higher-order transfer function can
be constructed by combining the logarithmic magnitude of the first and second-order
components with the phase contribution.

The asymptomatic high frequencies above the cutoff frequency are used to substitute the low
frequency values of magnitude and phase at frequencies above and below the cutoff frequency,
respectively, in the primary and secondary factor "proximity Bode plots." increase. First-order
and second-order transfer function component asymptotic plots are combined and subtracted to
create higher-order asymptotic board graphs. According to Figure 2.12, the loop gain frequency
response function's Bode plot may be used to quickly assess the stable feedback loop gain and
phase margin.

Nyquist Plots

The Nyquist plan has already been shown, The frequency response function's polar plot uses
frequency as a parameter. In situations when frequencies are not shown along the trajectory, as
in certain applications, Nyquist charts are less informative than Bode plots. Figure 2.15 depicts
the second-order frequency response function of the Nyquist diagram 2.9. Nyquist plots are
frequently employed exclusively for for 0 ≤ ω < ∞. The real axis-based mirroring is followed by
the presentation of the negative frequency graph.

The Nyquist plot approaches its source if L fits ω → ∞ precisely, Put it in terms of how it relates
to L's pole and zero (2.37). after that asymptotically

k
L ( jω ) ≈ for ω → ∞ (2.47)
( jω )n−m

The Nyquist plot approaches the origin at an angle of −(n − m) × 90◦ if k is positive, while the
system's pole-zero excess or relative order is represented by the quantity nm-m, The origin of
the loop gain L in a k-type control system has a degree k pole. As a response, at low
frequencies, the loop frequency response asymptotically takes the following form:

c
L ( jω ) ≈ for ω ↓ 0 , (2.48)
( jω )k

With respect to the value of k = 0, the Nyquist plot of L approaches the point on the real axis ω ↓
0 using the actual constant c. However, the Nyquist plot is rotated by -k x 90° to infinity if k > 0
and c is positive. Figure 2.16 illustrates this.
M- and N- circles

Consider the straightforward Unity Feedback Loop in Figure 2.17, which has a loop gain of L.
Loop-closure transfer, The system performs the same purpose as the complimentary sensitivity
function.

L
H=T = (2.49)
1+ L

The Nyquist diagram of loop gain L may be used to determine the closed-loop frequency
response function using the M- and N- circles, which are typical of the classical control era.
Point z on the complex plane is located at the M- circle. Where is the complex number's size.

z
(2.50)
1+ z

With M- squared, it is stable. M- center radius and circle.

center
( M2
1−M 2 )
,0 , |
radius
M
1−M 2 | (2.51)

The center and radius of the N- circle, which has an argument of constant number (2.50) and is
equal to Arctan N-, describe a complex planar location.

center ( −12 , 21N ) , radius

1
2√ 1
1+ 2
N
(2.52)

The location of the M- and N- circles on the complex plane is depicted in Figure 2.18. The
intersection of the loop gain L and M- circles' Nyquist diagrams yields the size of the closed-loop
frequency response and the corresponding sensitivity function T. Similar to this, the phase of T
continues from the point where the N- circle intersects. A typical Nyquist diagram with loop gain
L is shown in Figure 2.18. These are a few of the traits of the closed-loop response discovered
by analysis of the intersection with the M- circle.

• The highest value of M that can be found along the Nyquist diagram is the height of the
resonance peak.

• The frequency at which this maximum happens is known as the resonant frequency
ωm .

• The frequency at which the 0.707 circle meets the Nyquist plot is known as the
bandwidth (the -3 dB circle).
These and other findings offer helpful hints as to how the frequency response of the loop might
be altered and molded to enhance closed-loop characteristics. Compared to the Nyquist
diagram, the M- and N- loci are more frequently included in the Nichols diagram in the following
section.

Nichols Plots

The Nyquist display's linear scale can obscure a broad range of values if the loop gain's
magnitude fluctuates. Furthermore, it is not always simple to forecast how the frequency
response function P of the plant or the frequency response function C of the compensator would
modify the Nyquist diagram of loop gain L = PC. By using a Nichols diagram to visualize the
loop gain, both issues may be resolved (James et al., 1947). By contrasting the loop gain
frequency response function's logarithmic magnitude and phase, the Nichols plot is produced.
The circle of M and N represents their position at these coordinates. A Nichols plot is created by
a succession of M and N positions on a plane of magnitude relative to the position. The image
displays the Bode and Nyquist plots of the second-order frequency response functions, whereas
Section 2.19 displays the Nichols plot of these functions. 2.15 or 2.9.
Changes in gain cause vertical movements in the Nichols plot, whereas changes in phase
cause horizontal shifts. Therefore, it is straightforward to estimate how modifications to the plant
frequency response function P or the compensator frequency response function C would impact
the loop gain L = PC.

Classical Control System Design

A control system should not just do the job, it is not just supposed to have its transfer
function to be right. Just like any other machines and devices, it should also meet some certain
specific goals, which are the basic necessities in a single input and single output control system.
Design goals and criteria:
1. The transient response has to happen quickly enough.
2. The dampening in the transient is good.
3. The transient reaction complies with the accuracy standards, which are frequently
described in terms of the error parameters of 2.2 (p. 60)
4. The system is adequately resistant to shocks from the outside and changes in internal
Parameters.
Gain Margin – calculates the relative stability. It is the reciprocal of the magnitude of the niquist
at phase crossover frequency.
1
Gm =
M pC
Phase Margin – calculates relative stability too. It is equal to the sum of the angle 180 degrees
plus phase angle at the gain crossover frequency.
Gm = 180 + ᴓgc
Bandwidth – calculates the frequency – domain speed response. Labeled as B in the figure,
defined as frequency range that the closed-loop frequency response H occurs
has a magnitude that is at least one factor 1/2√2 = 0.707 (3 dB) is the value at zero frequency.
Resonance Peak - relative stability can be assessed using the peak value M of the
amplitude of the resonance's closed-loop frequency response H (in dB) frequency

Delay time (Td) – The total average delay between the reference and output is measured by
this. It is also described as the time where the return is 50 percent of the amplitude step or step
amplitude.
Rise time (Tr) – is represent the sharpness of the response. Or the time required to uplift the
10% to 90% in the end value.
Percentage overshoot (PO) – represents the utmost and least difference between the transient
and the steady-state response to an input step.
Settling time (Ts ) – described as the time needed for the reponse to step input to come to a
particular percentage ( normally 2 or 5 percent) and remain in it.

Lag, Lead, and Lag-Lead Compensation

Techniques in classical control also include Lag, Lead, and Lag-Lead Compensation. In
which will be discussed in this secition.

Lead Compensator
A lead compensator creates a sinusoidal output which has phase lead whenever a sinusoidal
input is put in. The purpose of a lead compensator is that is may add stability and the response
speed of a system.
Lag Compensator
When given sinusoidal input, a lag compensator is an electrical network that outputs sinusoids
with phase lag. A lag compensator's goal is to lower the steady-state inaccuracy. Different
combinations of one or more lead and lag compensators may be used, depending on the
desired outcome.
Lead-Lag Compensator
Lead-lag compensators are used to impact branches such as robotics, satellite
commands, automobile diagnostics, etc. A lead-lag compensator is made up of two
compensators: one for the lead cascaded with another one for the lag.
The figure above shows the Characteristics of lead and lag compensation design
Evans (1950) and Evans (1954) developed the root locus technique, which involves graphing
the locations of the roots of the characteristic equation of the closed-loop system as a function
of a proportionate gain factor. The stability characteristics of the system as a function of gain are
clearly depicted by this graphical manner. During design, it causes a decision regarding the
gain's worth. As a result, we can determine the closed loop poles' course and the control
system's type.
Root Locus Objectives
- To be aware if the system is totally stable

- To anticipate the results of altering the gain amount OR adding open-loop poles and/or
open-loop zeros on the positioning of the closed-loop poles.

The Guillemin-Truxal design procedure 

By proposing a network-theory focused approach towards the synthesis of control systems,
Truxal (1955). The concept is straightforward. Rather than building a compensator based on an
analysis, The closed-loop numerical solution H is simply chosen first from open-loop transfer
function such that. In general, an approximation is required to arrive at a workable low-cost
compensator. 

                                                                                                                 Unit Feedback System 

 
Procedure: 
H should represent the shuttered frequency response. For system evaluation loops such as
those depicted to the Unit frequency response system using a shrub and compensatory both P
and C are techniques commonly are commonly used. 
H=PC1 + PCH=PC1 + PC
 
 
When we solve for compensator transfer function C, we get 
C=1PH1−HC=1PH1−H
 
It is not easy to determine the desired H. H is sometimes chosen based of a first examination of
the performance of the only a limited control method is present in the closed circuit
 
L(s)=N(s)skD(s)L(s)=N(s)skD(s)
 
Considering N & D equations with zero roots. As a result, 
 
H(s)=L(s)1+L(s)=N(s)skD(s)+N(s)=amsm+aksk+bk−1sk−1+...+b0cnsn+cksk+bk−1sk−1+...
+b0H(s)=L(s)1+L(s)=N(s)skD(s)+N(s)=amsm+aksk+bk−1sk−1+...+b0cnsn+cksk+bk−1sk−1+...
+b0
 
 
Considering a first
kk
coefficients
bjbj
, on the other hand in the denominator polynomial equal to the use of a denominator polynomial
assures that the system is type k.  
         This still provides plenty of room for additional objectives. Assume we choose closed-loop
function as, 
H(s)=b0sn+bn−1sn−1+...+b0H(s)=b0sn+bn−1sn−1+...+b0
 
This implies a steady-state inaccuracy of 0 with phase inputs r(t) = 1(t). 
Practice: Zero steady-state error 
One method for determining the coefficients
b0, b1b0, b1
…,
 bn−1bn−1
is to evenly distribute the closed-loop poles. Dispersed here on left side of a circle having a
radius of zero and a center there at origin This produces closed loop replies with the desired
amount of damping. Butterworth polynomials are the resultant polynomials.  
    Another typical option is to select the coefficients so the integral of time multiplied by absolute
blunder 
∫∞0t|e(t)|dt∫0∞t|e(t)|dt
 
is small, with e representing the failure for the step input.

 
A Butterworth is an example of a signal conditioning screen, which aims to have a passband
frequency range that is as flat as possible.  
Doctor Graham and R.C. created the first ITAE coefficient table (ITAE stands for Integral of
Time multiplied by Absolute Error). Lathrop with a conventional computer. 
                                                              Butterworth pole patterns 
 
 
                                                                         ITAE criterion 

The denominator is an ITAE polynomial of second order. The numerator was already set to
zero. Type 1 control is position error. 
          The closed-loop transfer function H and the closed-loop step response of a
Guillemin-Truxal design for the cruise control system.  
 
 
Quantitative feedback theory (QFT) 
Horowitz (1982) coined the term quantitative feedback theory (QFT) (also see Horowitz and Sidi
(1972)). Lunze provides an informative account (1989). The method is firmly anchored in
traditional control. It seeks to satisfy quantitative limitations imposed on changes within closed-
loop transfer function caused by defined changes in the loop gain. The design method is based
on the Nichols chart's graphical representation of loop gain. 
Effect of parameter variations 
 Nichols plots can be used to examine the impact of parameter fluctuations and other
plant-specific uncertainties on closed-loop systems.
 The effect of disturbances within transfer function can be studied the breadth of the path
which is pushed beyond the disturbances   
             Nominal and perturbed complementary sensitivity functions and step
responses for the nominal design 
 
For instance: (2nd structure with unknowns) As an illustration, take the transfer-functioning
plant.  [Equation] 
g = 1 & = 0 are nominal values. The gain g fluctuates from 0.5 and 2 when perturbed. 
We suppose that preliminary research has resulted in a design phase in the format of a main
compensator with transfer function. 
[Equation] 

 
 
                   Nominal Nichols plot and uncertainty regions. (Shaded: forbidden region) 
depicts the nominal loop gain's Nichols plot, [Equation]with [Equation]The picture also includes
the uncertainty zones brought on by parameter changes occurring at various fixed frequency. 
correspond to extreme values of the parameters as follows:  
A: θ = 0, g = 0.5  
B: θ = 0.2, g = 0.5, 
 C: θ = 0.2, g = 2,  
D: θ = 0, g = 2. 
Stability and performance robustness 
If perturbations prevent the loop gain's Nichols plot from crossing the chart's center, the closed-
loop system will be stable and reliable. 
       The QFT technique specifies a banned region start point for loop gain to prevent
destabilization brought on by unmodeled disturbances. 
       Performance robustness is defined in the QFT technique as limitations on the change of the
size of a closed-loop spectral function H in the simplest case. 
 
Performance robustness of the design example 
According to a visual inspection of the plots, the disturbances move out a very small range of
variants of |T | at frequency range only about 0.2, a band with a size of approximately 5 dB at
intensity 1, a band with a size of approximately 10 dB between frequencies 2 and 10, and a
band with an increasing width for higher frequencies. 
 
Strong feedback systems designed with QFT 
It is simple for feedback developed system to fall short of the efficiency robustness
requirements. Often, this can be fixed by altering the loop gain L. 
       The procedure includes creating models of the uncertain regions at various frequencies
(often no more than five), then moving these about in Nichol’s diagram. 
 
Establishing the performance limits. 
The initial step of the process is to determine the location of the notional points for each chosen
crucial frequency so that the leniency band is fulfilled with the closest fit possible. The
effectiveness border designates this region. 
 
Establishing the robustness limits. 
The robustness border is then drawn by moving the pattern around it bidden zone, so it touches
but does not penetrate it. If a certain frequency occurs 
Gain shaping in loops. 
The design's most vital phase is to modify the loop gain so that 
 The associated loop gain L(ώ) is on or over the associated quality barrier at each crucial
wavelength. 
 The relevant loop gain for each crucial phase is from the east of the associated instability
resilience limit. 
 

 
                                                 Performance and Robustness boundaries 
 
Prefilter design  
The QFT layout must be finished with the prefilter layout again when the input absorber has
indeed been chosen. 
 
 
Concluding remarks 
The methods for designing a traditional absorber are covered in this chapter. The accessible
resonant frequency is being shaped with the goal of compensating. 
In the classical control techniques, the solid start for modifying loop gain are thoroughly
discussed. 

REFERENCES
Bosgra, O.H.,Kwakernaak, H., & Meinsma, G. (2007-2008) Design Methods for Control
Systems, Notes for a course of the Dutch Institute of Systems and Control Winter term
2007–2008.

Tutorials Point India (2016)

https://www.youtube.com/channel/UCVLbzhxVTiTLiVKeGV7WEBg/about

https://ctms.engin.umich.edu/CTMS/index.php?aux=Extras_Leadlag