Compensation of feedback control systems

Every Control system which has been designed for a specific application should meet certain
performance specification. However, there are always some constraints which are imposed on the control
system design in addition to the performance specification. The choice of a plant is not only dependent
on the performance specification but also on the size, weight & cost. Although the designer of the
control system is free to choose a new plant, it is generally not advised due to the cost & other
constraints. Under this circumstances it is possible to introduce some kind of corrective sub-systems in
order to force the chosen plant to meet the given specification. We refer to these sub-systems as
compensators or controllers whose job is to compensate for the deficiency in the performance of the
plant.
Setting the gain is the first step in adjusting the system for satisfactory performance. In many practical
cases, however, the adjustment of the gain alone may not provide sufficient alteration of the system
behavior to meet the given specifications. As is frequently the case, increasing the gain value will
improve the steady-state behavior but will result in poor stability or even instability. It is then necessary
to redesign the system (by modifying the structure or by incorporating additional devices or components)
to alter the overall behavior so that the system will behave as desired. Such a redesign or addition of a
suitable device is called compensation. A device inserted into the system for the purpose of satisfying the
specifications is called a compensator. The compensator compensates for deficient performance of the
original system, to improve system dynamic performance through the addition of dynamic elements in
order to mitigate some of the undesirable features of the control elements present in the system.
The PID-based compensational approach to control systems design is one way of improving the
performance of the system. To this day, PID control is still the predominant method in industry and is
found in over 95 percent of industrial applications. Its success can be attributed to the simplicity,
efficiency, and effectiveness of this design method. We discuss the basic controllers such as the
proportional, the derivative and the integral controllers.
Proportional Controller
The block diagram of the unity negative feedback closed loop control system along with the proportional
controller is shown in the following figure.
The proportional controller is used to change the transient response as per the requirement. The
proportional controller produces an output, which is proportional to error signal.
u(t )αe(t )
⇒u (t )=K P e (t )
Apply Laplace transform on both the sides -
U ( s)=K P E (s)
U (s)
⇒ =K P
E (s)

Therefore, the transfer function of the proportional controller is

K P.
Where,
U (s) is the Laplace transform of the actuating signal u (t)
E (s) is the Laplace transform of the error signal e (t)
K P is the proportionality constant
Derivative Controller
The block diagram of the unity negative feedback closed loop control system along with the derivative
controller is shown in the following figure.

The derivative controller is used to make the unstable control system into a stable one. The derivative
controller produces an output, which is time derivative of the error signal.
de (t )
u(t )= K D
dt
Apply Laplace transform on both sides.
U ( s)=K D sE (s)
U (s)
⇒ =K D s
E (s)

Therefore, the transfer function of the derivative controller is

K D s , where, K D is the derivative constant.
 Integral Controller
The integral controller produces an output, which is the time integral of the error signal. The block
diagram of the unity negative feedback closed loop control system along with the integral controller is
shown in the following figure. The integral controller is used to decrease the steady state error.

u(t )= K I ∫ e(t )dt

Apply Laplace transform on both the sides -
KI
U ( s )= E (s )
s
U (s ) K I
⇒ =
E (s ) s
KI
K
Therefore, the transfer function of the integral controller is s , where, I is the integral constant.
Let us now discuss about the combination of basic controllers.
Proportional Derivative (PD) Controller
The proportional derivative controller produces an output, which is the combination of the outputs of
proportional and derivative controllers.
de(t )
u(t )= K P e (t )+K D
dt u(t)
Apply Laplace transform on both sides -
U ( s )=( K P +K D s ) E(s )
U (s )
⇒ =( K P + K D s )
E (s )

Therefore, the transfer function of the proportional derivative controller is(

K P+K D s)
.
The block diagram of the unity negative feedback closed loop control system along with the proportional
derivative controller is shown in the following figure. The proportional derivative controller is used to
improve the stability of control system without affecting the steady state error.
 Proportional Integral (PI) Controller
The proportional integral controller produces an output, which is the combination of outputs of the
proportional and integral controllers.

u(t )= K P e(t )+K I ∫ e(t )dt

Apply Laplace transform on both sides -

(
U ( s )= K P +
KI
s ) E( s )

⇒
U (s)
E (s) (
= K P+
KI
s )
Therefore, the transfer function of proportional integral controller is
( K P+
KI
s ) . The block diagram of
the unity negative feedback closed loop control system along with the proportional integral controller is
shown in the following figure.

The proportional integral controller is used to decrease the steady state error without affecting the
stability of the control system.
Proportional Integral Derivative (PID) Controller
The proportional integral derivative controller produces an output, which is the combination of the
outputs of proportional, integral and derivative controllers.
de (t )
u(t )= K P e (t )+K I ∫ e(t )dt+ K D
dt

(
U ( s )= K P+
KI
s )
+ K D s E( s )

Apply Laplace transform on both sides -

⇒
U (s)
E (s) ( KI
= K P+ +K D s
s )
Therefore, the transfer function of the proportional integral derivative controller is
( K P+
KI
s
+K D s ) .
The block diagram of the unity negative feedback closed loop control system along with the proportional
integral derivative controller is shown in the following figure.

The proportional integral derivative controller is used to improve the stability of the control system and
to decrease steady state error.
Electronic Controllers using operational amplifiers.
In what follows we discuss electronic controllers using operational amplifiers.
Proportional Mode
Implementation of this mode requires a circuit that has a response given by:
u(t )= K P e (t )+u 0
Where,
e (t ) is error in percent of variable range
u(t ) is controller output
K P is proportional gain

u0 (t ) is controller output with no error

If we consider both the controller output and error to be expressed in terms of voltage, we see that the
equation is simply a summing amplifier. The op amp circuit in the figure below shows such an electronic
proportional controller.

In this case, the analog electronic equation for the output voltage is
V out =G P Ve +V 0 ,

where
V out is output voltage

G P =R2 / R 1 is the proportional voltage gain.

Ve is error voltage
V 0 is output with zero error

The design of a proportional controller calls for specification of the proportional gain described

by
K P that expresses the percent of output for an error of 1% of the measurement range. Alternatively, it

could be described as the proportional band,

PB=100 /K P . This must now be expressed in terms of the

voltage gain,
G P . The relationship between G P and K P is given by,

ΔV out
G P =KP
ΔV in
Where,
ΔV out is the range of output voltage
ΔV in is the range of measurement voltage

Integral Mode
Implementation of this mode requires a circuit that has a response given by:
t
u(t )= K I ∫ e(t )+u0
0

Where,
e (t ) is error in percent of variable range
u(t ) is controller output
K I is the integral gain (s−1 )

u0 (t ) is controller output at time t = 0

This function is easy to implement when op amps are used as the building blocks. A diagram of an
integral controller is shown in the figure below.
The corresponding equation relating input to output is
t
V out =G I∫ V e (t )+V out (0 )
0

Where,
V e is error voltage

V out is controller output voltage

G I=1/RC is the integral gain (s−1 )

V out (0 ) is controller output voltage at time t = 0

The values of R and C can be adjusted to obtain the desired integration time. The initial controller
output is the integrator output at t =0 . The integration time constant determines the rate at which

controller output increases when the error is constant. If

K I is made too large, the output rises so fast that

overshoots of the optimum setting occur and cycling or oscillations is produced. The actual value of
GI ,

and therefore of R and C, is determined from

K I and the input and output voltage ranges.
Derivative Mode
The derivative mode is never used alone because it cannot provide a controller output when the error is
zero. Nevertheless, we show here how it is implemented with op amps so it can be combined with other
modes in the next section. The control mode equation was given earlier as
de (t )
u(t )= K D
dt
where
e (t ) is error in percent of variable range
u(t ) is controller output
K D is the derivative time constant

In principle, this mode could be implemented by the op amp circuit shown in the figure below with
appropriate identifications in terms of circuit elements where:
dV e
V out =−RC
dt
The input voltage has been set equal to the controller error voltage. From a practical perspective, this
circuit cannot be used because it tends to be unstable; that is, it may begin to exhibit spontaneous
oscillations in the output voltage. The reason for this instability is the very large gain at high frequencies
where the derivative is very large. To study this effect, let us assume that the input voltage is given by a

sinusoidal voltage oscillating with some frequency, f; then Ve=V 0 sin (2 π ft ) . So, from the above equation

we can write the amplitude of the output as,

|V out|=2π fRC|V e|. This equation shows that the magnitude of
the output voltage increases linearly with frequency. So, in principle, as the frequency goes to infinity, so
does the output! Clearly this is unacceptable in our control system. A little high-frequency noise will
cause large excursions in output voltage.

In order to make a practical circuit, a modification is provided that essentially “clamps” the gain above
some frequency to a constant value. We make sure that the clamped frequencies are well above anything
that could occur in the actual control system. This way, the circuit provides a derivative output in the
frequencies of practical interest but simply acts like a fixed-gain amplifier at higher frequencies. Figure
below shows that the simple modification is to place a resistor in series with the capacitor.

The actual transfer function for this circuit can be shown to be given by
 dV out dV e
V out + R1 C =−R2 C
dt dt
It can see that the output depends upon the derivative of the input voltage, but there is now an extra term
involving the derivative of the output voltage. Essentially, we have a first-order differential equation
relating input and output voltage. For very high frequencies the impedance of the capacitor becomes very
small and can be neglected. Then the circuit becomes just an inverting amplifier with a gain (R2/R1). At
low frequency the impedance of the capacitor will be large so R1 can be neglected. Then, the response of
the first derivative equation prevails. The circuit exhibits a derivative response provided the following
inequality is satisfied,
2πfR1C << 1
Therefore, when using a derivative action circuit, we must estimate the maximum physical frequency at

which the system can respond, f max , and pick R1 so that for frequencies much higher than this, the
inequality of Equation 2πfR1C << 1 is satisfied.
Composite Controller Modes
The combination of several controller modes combine the advantages of each mode and, in some cases,
eliminate disadvantages. Composite modes are implemented easily using op amp techniques. Basically,
this consists of simply combining the mode circuits introduced in the previous section.
Proportional-Integral
A simple combination of the proportional and integral circuits provides the proportional-integral mode of
controller action. The resulting circuit is shown in the figure below.

For this case, the relation between input and output is most easily found by applying op amp circuit
analysis. We get (including the inverter)

( )
t
R2 1
V out = V + ∫V
R 1 e R1 C 0 e

Usually the definition of the proportional-integral controller mode includes the proportional gain
in the integral term, so we write
 ( ) ( )
R2 R2 1 t
V out = V + ∫ V +V ( 0)
R 1 e R 1 R 2 C 0 e out
R2
G p=
The adjustments of this controller are the proportional band through R1 , and the integration gain

1
G I=
Through
R2 C

Proportional-Derivative
This combination is implemented using a circuit similar to that shown in the figure below. Analysis
shows that this circuit responds according to the equation

( ) ( ) ( )
R1 dV out R2 R2 dV e
V out + R3 C = V e+ R3 C +V 0
R1 + R 3 dt R 1 + R3 R1 + R 3 dt

where the quantities are defined in the figure and the output inverter has been included.

This circuit includes the clamp to protect against high gain at high frequency in the derivative term. In
this case, an effective resistance given by
R1 R3
R=
R1 + R3 .Then the condition becomes as usual,2 πf max RC=0 .1 . Assuming this criterion has been

met, the equation for the proportional-derivative response becomes

( ) ( )
R2 R2 dV e
V out = V e+ R3 C +V 0
R 1 + R3 R1 + R 3 dt
R2
G p=
where the proportional gain is R1 + R 3 , and the derivative gain is G D=R3 C . Of course, this mode

still has the offset error of a proportional controller because the derivative term cannot provide reset
action.
PID (Three-Mode).
The ultimate process controller is the one that exhibits proportional, integral, and derivative response to
the process-error input. This mode is characterized by
 de (t )
u(t )= K P e (t )+K P K I ∫ e(t )dt +K P K D +u I (0)
dt
Where,
u(t ) is controller output in percent of full scale
K P is the proportional gain

e (t ) is the process error in percent of maximum

K I is the integral gain

K D is the derivative gain

u I (0 ) is the initial controller integral output

The zero-error term of the proportional mode is not necessary because the integral automatically
accommodates for offset and nominal setting. This mode can be provided by a straight application of op
amp circuits, resulting in the circuit of the figure below. It must be noted, however, that it is possible to
reduce the complexity of the circuitry of the figure 12 and still realize the three-mode action, but in these
cases an interaction results between derivative and integral gains.

We will use this circuit because it is easy to follow in illustrating the principles of implementing this
mode. Analysis of the circuit shows that the output is:

( ) ( ) ( )
R2 R 1 R dV
−V out =
R1
V e+ 2
R1 RI C I
∫ V e dt + 2 R D C D e +V out (0 )
R1 dt

NOTES:
These circuits have shown that the direct implementation of controller modes can be provided by standard
op amp circuits. It is necessary, of course, to scale the measurement as a voltage within the range of
operation selected by the circuit. Furthermore, the outputs of the circuits shown have been voltages that
may be converted to currents for use in an actual process-control loop. These circuits are only examples
of basic circuits that implement the controller modes. Many modifications are employed to provide the
controller action with different sets of components.
Uses of frequency domain compensating networks in the control systems.
The important uses of the compensating networks are written below.
Necessary of Compensation
1. In order to obtain the desired performance of the system, we use compensating networks.
Compensating networks are applied to the system in the form of feed forward path gain
adjustment.
2. Compensate an unstable system to make it stable.
3. A compensating network is used to minimize overshoot.
4. These compensating networks increase the steady state accuracy of the system. An important
point to be noted here is that the increase in the steady state accuracy brings instability to the
system.
5. Compensating networks also introduces poles and zeros in the system thereby causes changes in
the transfer function of the system. Due to this, performance specifications of the system change.
Methods of Compensation
1. Series Compensator: Connecting compensating circuit between error detector and plants known
as series or cascade compensation. The block diagram representation is shown.

2. Feedback Compensator: When a compensator used in a feedback manner called feedback

compensation.
 3. Load Compensator: A combination of series and feedback compensator is called load
compensation.

Now what are compensating networks? A compensating network is one which makes some adjustments
in order to make up for deficiencies in the system. Compensating devices may be in the form of electrical,
mechanical, hydraulic etc. Most electrical compensators are RC filters. The simplest network used as
compensators are known as lead, lag and lead-lag networks.
Phase Lead Compensation
The lead compensator is an electrical network which produces a sinusoidal output having phase lead
when a sinusoidal input is applied. The lead compensator circuit is shown in the following figure.

Here, the capacitor is parallel to the resistor R1 and the output is measured across resistor R2.
A system which has one pole and one dominating zero (the zero which is closer to the origin than all
other zeros is known as dominating zero.) is known as lead network. If we want to add a dominating zero
for compensation in control system then we have to select lead compensation network.
The basic requirement of the phase lead network is that all poles and zeros of the transfer function of the
network must lie on (-) ve real axis interlacing each other with a zero located at the origin of nearest
origin.
From above circuit we get,
Equating above expression of I we get,

Now let us determine the transfer function for the given network and the transfer function can be
determined by finding the ratio of the output voltage to the input voltage. So taking Laplace transform of
both side of above equations, we get:
1 1
E ( s )= [ Ei ( s )−E0 (s )]+Cs [ Ei (s )−E 0 ( s )] (neglecting initial conditions )
R2 0 R1
1 1 Ei ( s )
⇒ E0 (s )+ E 0 ( s )+CsE0 ( s )= +CsEi ( s )
R2 R1 R1

On substituting the α =( R 2 + R1 )/ R2 and T ={( R1 R2 )/( R1 + R 2 ) } in the above equation.

Where,T andα are respectively the time constant and attenuation constant,
we have

The above network can be visualized as an amplifier with a gain of 1/α.

Let us draw the pole zero plot for the above transfer function.
Pole Zero Plot of Lead Compensating Network
Clearly we can see that -1/T (which is a zero of the transfer function) is closer to origin than the -1/(αT)
(which is the pole of the transfer function). Thus we can say that in the lead compensator zero is more
dominating than the pole and because of this, the lead network introduces positive phase angle to the
system when connected in series.

If we substitute s= jω in the above transfer function and also we have 0 < α < 1. We can find the phase
angle function for the transfer function we as:

θ( ω)=tan−1 (ωT )−tan−1 (αω T )

We know that, the phase of the output sinusoidal signal is equal to the sum of the phase angles of input
sinusoidal signal and the transfer function. So, in order to produce the phase lead at the output of this
compensator, the phase angle of the transfer function should be positive. Therefore, zero will be nearer
to origin in pole-zero configuration of the lead compensator.

Now in order to find that the maximum phase lead θ m occurs at a frequency ω m let us differentiate this
phase function and equate it to zero. On solving the above equation, we get
1−sin θ m
α=
1+sin θm
Where, θm is the maximum phase lead angle. And the corresponding magnitude of the transfer function at
maximum θm is 1/ α .

Frequency response of G c ( jω)

A Bode plot of a phase-lead compensator G c ( jω) has the following form. The two corner frequencies
are at 1/ αT and 1/ T ; note the positive phase that is added to the system between these two frequencies.
The lead compensator increases the gain of the system at high frequencies (the amount of this gain is
equal to α ).

The maximum phase lead angle θm occurs at

ω m where,

1−α
sin θm =
1+ α , and
1
logω m= [log T +log αT ]
2
1
⇒ ωm=
T √α
Polar plot of a lead network
a ( jωT + 1 )
( jω aT + 1 ) , where 0 < a < 1 is given by:

Effect of Phase Lead Compensation

1. The velocity error constant Kv increases.
2. The slope of the magnitude plot reduces at the gain crossover frequency so that relative stability
improves and error decrease due to error is directly proportional to the slope.
3. Phase margin increases.
4. Response become faster.
Advantages of Phase Lead Compensation
Let us discuss some of the advantages of the phase lead compensation-
1. Due to the presence of phase lead network the speed of the system increases because it shifts gain
crossover frequency to a higher value.
 2. Due to the presence of phase lead compensation maximum overshoot of the system decreases.
Disadvantages of Phase Lead Compensation
Some of the disadvantages of the phase lead compensation –
1. Steady state error is not improved.
Lag Compensator
The Lag Compensator is an electrical network which produces a sinusoidal output having the phase lag
when a sinusoidal input is applied. A system which has one zero and one dominating pole (the pole
which is closer to origin than all other poles is known as dominating pole) is known as lag network. If we
want to add a dominating pole for compensation in control system then, we have to select a lag
compensation network. The basic requirement of the phase lag network is that all poles and zeros of the
transfer function of the network must lie in (-) ve real axis interlacing each other with a pole located or
on nearest to the origin. The lag compensator circuit is shown in the following figure

Here, the capacitor is in series with the resistor R2 and the output is measured across this combination.
The transfer function of this lag compensator is as follows:
We will have the output at the series combination of the resistor R2 and the capacitor C.
From the above circuit diagram, we get
1
e i=iR 1 + iR 2 +
C
∫ idt
1
e 0 =iR 2 +
C
∫ idt
Now let us determine the transfer function for the given network and the transfer function can be
determined by finding the ratio of the output voltage to the input voltage. Taking Laplace transform of
above two equation we get,
 1
Ei (s )=I ( s )R1 + I ( s )R 2 + I (s )
Cs
1
E 0 ( s )=I (s )R 2 + I (s )
Cs
1
E0 (s ) R 2 + Cs
Transfer function , Glag (s )= =
Ei ( s ) 1
R 1 + R2 +
Cs
R2 Cs+1
⇒G lag ( s )=
( R 1 + R2 )Cs+ 1
On substituting
T =R2 C
( R 1 + R2 )
β=
R1

in the above equation (Where, T and β are respectively the time constant and DC gain), we have
E 0( s ) 1+Ts
Transfer function , G lag ( s )= =
Ei( s ) 1+ β Ts
The above network provides a high frequency gain of 1 / β. Let us draw the pole zero plot for the above
transfer function.

Clearly we have -1/T (which is a zero of the transfer function) is far to origin than the -1 / (βT) (which is
the pole of the transfer function). Thus we can say in the lag compensator pole is more dominating than
the zero and because of this lag network introduces negative phase angle to the system when connected

in series. Let us substitute s = jω in the above transfer function and also we have β < 1. On finding the
phase angle function for the transfer function we have:

θ( ω)=tan−1 (ωT )−tan−1 ( βωT )

We know that, the phase of the output sinusoidal signal is equal to the sum of the phase angles of input
sinusoidal signal and the transfer function.
So, in order to produce the phase lag at the output of this compensator, the phase angle of the transfer
function should be negative.
Now in order to find out the maximum phase lag occurs at a frequency let us differentiate this phase
function and equate it to zero. On solving the above equation we get:
1−sin θ m
β=
1+sin θm
Where, θm is the maximum phase lead angle. Remember β is generally chosen to be greater than 10.
Frequency response polar plot of a lag compensator with the transfer function:

Transfer function , Glag (s)=K c ( 1+Ts

)
1+β Ts is shown below.

Bode diagram of a lag compensator with K c =1 , β=10 is shown below.

Effect of Phase Lag Compensation

1. Gain crossover frequency increases.
2. Bandwidth decreases.
3. Phase margin will be increase.
4. Response will be slower before due to decreasing bandwidth, the rise time and the settling time
become larger.
Advantages of Phase Lag Compensation
Let us discuss some of the advantages of phase lag compensation –
1. Phase lag network allows low frequencies and high frequencies are attenuated.
2. Due to the presence of phase lag compensation the steady state accuracy increases.
Disadvantages of Phase Lag Compensation
Some of the disadvantages of the phase lag compensation –
 1. Due to the presence of phase lag compensation the speed of the system decreases.
Phase Lag Lead Compensation
With the single lag or lead compensation may not satisfy design specifications. For an unstable
uncompensated system, lead compensation provides fast response but does not provide enough phase
margin whereas lag compensation stabilizes the system but does not provide enough bandwidth. So we
need multiple compensators in cascade. Given below is the circuit diagram for the phase lag-lead
compensation network. It is an electrical network which produces phase lag at one frequency region and
phase lead at other frequency region. It is a combination of both the lag and the lead compensators.

Now let us determine transfer function for the given network and the transfer function can be determined
by finding the ratio of the output voltage to the input voltage. This circuit looks like both the
compensators are cascaded. So, the transfer function of this circuit will be the product of transfer
functions of the lead and the lag compensators.

Transfer function , GLead-lag (s )=

E ( s)
0
=
( s+
1
R C )(1
s+
R C )
1
1 2 2

(R C R C R C ) R R C C
E ( s)
i 21 1 1 1
s + + + s+
1 1 2 2 2 1 1 2 1 2

( R1 C 1 s +1 )( R2 C 2 s+1 )
⇒G Lead-lag ( s )=
R1 R2 C 1 C2 s 2 +( R1 C 1 + R 2 C2 + R 1 C2 )s +1

On substituting the αT1 = R1C1, R2C2 = βT2, R1R2C1C2 = αβT1T2 and

T 1 T 2 =R1 R2 C 1 C 2 in the above

equation (where T1, T2 and α, β are respectively the time constants and attenuation constants). We have
( 1+ αT 1 s )( 1+ βT 2 s )
Transfer function , GLead-lag ( s )=
( 1+T 1 s )( 1+T 2 s )
Let us draw the pole zero plot for the above transfer function.
Clearly we have -1/T (which is a zero of the transfer function) is far to the origin than the -1/(βT) (which
is the pole of the transfer function). Thus we can say in the lag-lead compensation pole is more
dominating than the zero and because of this lag-lead network may introduces positive phase angle to the
system when connected in series.
Frequency response on a Bode diagram of a lag-lead compensator given by:

And Polar plot of a lag-lead compensator given by

Advantages of Phase Lag Lead Compensation

Let us discuss some of the advantages of phase lag- lead compensation-
1. Due to the presence of phase lag-lead network the speed of the system increases because it shifts
gain crossover frequency to a higher value.
2. Due to the presence of phase lag-lead network accuracy is improved.
Comparison between lead and lag compensators
 Lead compensator Lag compensator
High pass Low pass
Approximates derivative plus proportional control Approximates integral plus proportional control
Contributes phase lead Attenuation at high frequencies
Increases the gain crossover frequency Moves the gain-crossover frequency lower
Increases bandwidth Reduces bandwidth