Steady state error analysis with

unit step and ramp input signal

Final Exam Report,

This report is submitted to the electrical department,
College of Engineering, Kirkuk University.

Student Name: mohammad abu bakr osman

Subject: steady state error analysis with unit step and

ramp input signal

Stage: fourth

Student Signature:
Content
1 Introduction
2 Error signals 2.1 Steady state error for a step input
2.2 Steady state error for a ramp input
2.3 Steady state error for a parabolic input
2.4 Summary
2.5 Example

3 The steady state error of non-unity feedback systems

1
1 Introduction

It has been observed that much information about a control system can be obtained from the analysis

of its response to test inputs.

Recall from the last semester, if we have a first order system

𝑌(𝑠) 𝐵
= ,
𝑅(𝑠) 𝜏𝑠 + 1

its unit step response is found as the inverse Laplace transform of

𝐵/𝜏
𝑌(𝑠) =
𝑠(𝑠 + 1/𝜏)

and is equal to

𝑦(𝑡) = 𝐵 − 𝐵𝑒 −𝑡/𝜏

The parameter 𝐵 represents the 𝐷𝐶 gain of the system. It establishes the final value the output

approaches in response to a unit step. The parameter 𝜏 is the time constant of the system. As a rule
of

thumb it usually takes time of about 5𝜏 seconds for the response to reach its final value 𝐵.

Thus, given a first order system, applied the unit step input, we obtain the output whose final

value is 𝐵 ≠ 1. Often, we use control to make sure that the system obeys given commands. For

example, controlling the altitude of a fighter, we wish to make sure that the fighter will position at the

required altitude. However, the above example shows that there may be an error between the desired

and resulted final values.

One of the major advantages of feedback control is the ability

2
1. to track reference inputs precisely, and

2. to maintain this precision in the face of disturbances.

In this note, we consider how

such precision can be achieved.

2 Error signals

Consider the feedback system

in Figure 1.

In this system 𝑅(𝑠) is the

Laplace transform of a reference
input, and 𝐷(𝑠) is the Laplace trans‐

form of a disturbance. The transfer function 𝐺(𝑠) is called loop gain transfer function. Note that it

may be a series combination of the plant and a compensator. The output y(𝑠) is required to track the

reference input 𝑅(𝑠) . The input into plant 𝐺(𝑠) is the tracking error

𝐸(𝑠) = 𝑅(𝑠) − y(𝑠)

Let 𝑒(𝑡) denote its inverse Laplace transform. The limit

𝑒ss = lim 𝑒 (𝑡)

𝑡→∞

is referred to the steady state error. The value of 𝑒i⊃S characterises the final value of error as a
difference

between the final value of the input 𝑟(𝑡) and the final value of the output 𝑦(𝑡) . In other words, 𝑒SS is

3
the value of error after transients have died out.

To calculate 𝑒i⊃S , we use the Final Value Theorem for Laplace transforms. According to this the‐

orem, as long as 𝐸(𝑠) does not have any poles in the right half of the complex plane, except maybe,

𝑠 = 0, then

𝑒SS = lim 𝑠 𝐸(𝑠) .

𝑠→0

We now find 𝐸(𝑠) given 𝑅(𝑠) and 𝐷(𝑠) . Since

𝑌(𝑠) = 𝐺(𝑠)𝐸(𝑠) + 𝐷(𝑠) , 𝐸(𝑠) = 𝑅(𝑠) − 𝑌(𝑠) ,

then

𝑅(𝑠)−𝐷(𝑠) 1 1
𝐸(𝑠) = = 1+𝐺(𝑠) 𝑅(𝑠) − 1+𝐺(𝑠) 𝐷(𝑠) .
1+𝐺(𝑠)

Note that a change in the disturbance input has the same effect on the error as a change in the
reference

input, except the sign is opposite. This observation allows us to drop 𝐷(𝑠) and focus on the reference

input 𝑅(𝑠) . Conclusions drawn about 𝑅(𝑠) can then be applied to 𝐷(𝑠) .

2.1 Steady state error for a step input

Let 𝐷(𝑠) = 0. Consider 𝑅(𝑠) = 𝐴/𝑠, the step input of the height 𝐴. In this case, 𝐸(𝑠) simplifies to

1
𝐸(𝑠) = 𝑅(𝑠) .
1+𝐺(𝑠)

It can be shown that, except for 𝑠 = 0, 𝐸(𝑠) and the closed loop transfer function 𝐺/(1 + 𝐺)

have the same poles. Therefore, the condition of the Final Value Theorem for 𝐸(𝑠) to have all its

poles in the left half of the complex plane or at the origin, reduces to the condition that the closed loop

system must be stable or maybe have poles at 𝑠 = 0. If this condition is met, then we can continue

the calculation:

𝑅(𝑠) 𝐴/𝑠 𝐴
𝑒ss = lim 𝑠 = lim 𝑠 = .
𝑠→0 1 + 𝐺(𝑠) 𝑠→0 1 + 𝐺(𝑠) 1 + lim 𝐺 (𝑠)

Note that in this case, the steady state error is determined by the DC gain of 𝐺(𝑠) . The larger is DC

4
gain, the smaller is the steady state error. Furthermore, if 𝐺(𝑠) has one or more poles at 𝑠 = 0, then

lim 𝐺 (𝑠) = ∞. In this case, 𝑒i⊃S = 0. A pole at 𝑠 = 0 occurs if 𝐺(𝑠) is proportional 1/𝑠, i.e. if

𝐺(𝑠) includes integrator. This effect is called the integral action.

Thus, as long as 𝐺(𝑠) has one or more poles at 𝑠 = 0 and the closed loop system is stable, the

steady state error remains zero despite large modelling errors or changes in the plant. Note that this
is

true for a constant reference input or constant disturbance input.

The number of poles at the origin of the loop gain transfer function (i.e. the number of

integrators) defines the system’s type number.

Type zero systems do not include an integrator and therefore have a finite DC gain 𝐺(0) . A

constant called the position error constant, 𝐾𝑝 , is used to indicate the size of DC gain 𝐺(0) ,

𝐾𝑝 = lim 𝐺 (𝑠) .
𝑠→0

Note that the constant 𝐾𝑝 is relevant only in releation to step inputs. A step input has the physical

interpretation as a change in reference position for a positioning system and thus the name “position

error constant For a type zero system the steady‐state error in response to a step input of height 𝐴 is

𝐴
𝑒SS =
1 + 𝐾𝑝

Note that a large position error constant corresponds to a small steady‐state error.

2.2 Steady state error for a ramp input

Now consider the steady‐state error in response to a ramp input 𝑟(𝑡) = 𝐴𝑡. The Laplace transform of

the ramp input is 𝑅(𝑠) = 𝐴/𝑠 2 and

𝐴/𝑠 2 𝐴
𝑒ss = lim 𝑠 = .
𝑠→0 1 + 𝐺(𝑠) lim 𝑠 𝐺(𝑠)

Since ramp input corresponds to a change in velocity in a position control system, we call the error

constant associated with ramp inputs the velocity error constant, 𝐾𝑣 :

5
 𝐾𝑣 = lim 𝑠 𝐺(𝑠) .
𝑠→0

If 𝐺(𝑠) is a type zero system with no poles at the origin, then 𝐾𝑣 = 0 and the steady‐state error in

response to the ramp input is infinite, that is, the error grows with time. If you have a type zero system

and you wish to be able to follow a ramp within some degree of accuracy, you must add at least one

integrator to your controller.

If 𝐺(𝑠) is a type two or higher system, 𝐾𝑣 is infinite and the steady‐state error for a ramp input or

disturbance is zero.

A type one system will have a finite, non zero 𝐾𝑣 and the steady‐state error is given by

𝑒ss = 𝐾𝑣

and is constant. This situation is demonstrated in Fig. 2. Notice again that a large 𝐾𝑣 indicates a small

4𝑠+1 4𝑠+1
Figure 2: Ramp responses of type 1 (𝐺1 (𝑠) = 𝑠(𝑠2+4𝑠+6)) and type 2 systems (𝐺2 (𝑠) = 𝑠2 (𝑠2 +4𝑠+6))

steady‐state error. Also, notice that 𝐾𝑣 is defined only with respect to a ramp input.

2.3 Steady state error for a parabolic input

For a parabolic input 𝑟(𝑡) = 𝐴𝑡 2 /2, we take 𝑅(𝑠) = 𝐴/𝑠 3 . Since parabolic input corresponds to a

6
change in acceleration in a position control system, the associated error constant, 𝐾𝑎 , is called the

acceler
ation
error
constan
t in
referenc
e to a
positioni
ng
system:

𝐾𝑎 = lim 𝑠 2 𝐺(𝑠)
𝑠→0

Then, we obtain

𝐴/𝑠 3 𝐴 𝐴
𝑒ss = lim 𝑠 = = .
𝑠→0 1 + 𝐺(𝑠) lim 𝑠 2 𝐺(𝑠) 𝐾𝑎

Type zero and type one systems have 𝐾𝑎 = 0 and an infinitely growing error to a parabolic input. A

type three or higher system has an infinite 𝐾𝑎 and zero steady‐state error. A type two system has a

finite 𝐾𝑎 and a finite steady‐state error with a parabolic point.

2.4 Summary

The table summarizes the results of this section for steady‐state errors.

7
2.5 Example

2(𝑠+1) 100(𝑠+1)
Consider the control system where 𝐺𝑝 (𝑠) = (𝑠+2)(𝑠+3) , 𝐻(𝑠) = .
𝑠+4

Figure 3:

1. Find the steady state error when 𝑟(𝑡) is a unit step.

200(𝑠+1)2
First, find the loop gain 𝐺(𝑠) = 𝐻(𝑠)𝐺𝑝 (𝑠) = (𝑠+2)(𝑠+3)(𝑠+4) Second, find the position

error constant, 𝐾𝑝 = 𝐺(0) = 200/24 = 8.3333. Then,

1 1
𝑒ss = = = 0.1071.
1 + 𝐾𝑝 9.3333

2. Find the steady state error when 𝑟(𝑡) is a unit ramp.

Find the velosity error constant,

𝐾𝑣 = lim 𝑠 𝐺(𝑠) = 0 (note that this is a type 0 system). Then,

1
𝑒ss = = ∞.
𝐾𝑣
100(𝑠+1)
3. Now introduce integral action by modifying 𝐻(𝑠) to 𝐻(𝑠) = 𝑠(𝑠+4)
. With the modified

200(𝑠+1)2
compensator, 𝐺(𝑠) = 𝐻(𝑠)𝐺𝑝 (𝑠) = 𝑠(𝑠+2)(𝑠+3)(.𝑠+4). For the unit step input, the position

8
error constant 𝐾𝑝 = ∞, and 𝑒ss = 0. For the unit ramp input, the velosity error constant
200
𝐾𝑣 = lim 𝑠 𝐺(𝑠) = = 8.3333
𝑠→0 2⋅3⋅4
and 𝑒i⊃S = 1/8.3333 = 0.12.

4. With 𝐻(𝑠) as in 3, set 𝑟(𝑡) = 0 and take a look at the steady state error due to a constant

disturbance. That is, we now consider the system in Fig. 4

Figure 4:
2(𝑠+1) 100(𝑠+1)
where 𝐺𝑝 (𝑠) = (𝑠+2)(𝑠+3) , 𝐻(𝑠) = 𝑠(𝑠+4)
.

We can either note that the result will be the same as a step input result but with opposite sign (that

is 𝑒SS = 0 since we have integral action in the compensator), or derive the result “from the scratch”

Toward this end, we redraw the block diagram

Figure 5:

and find the transfer function from the disturbance input to the error:
𝐸(𝑠) −1
= .
𝐷(𝑠) 1 + 𝐺(𝑠)
𝐴
Set 𝐷(𝑠) = . Then from the Final Value Theorem
𝑆

−1
𝑒ss = lim 𝑠 𝐸(𝑠) = lim 𝑠 ⋅
𝑠→0 𝑠→0 1 + 𝐺(𝑠)

But 𝐾𝑝 = ∞ in this example, hence 𝑒SS = 0.

9
 𝐴 −𝐴 −𝐴
= = .
𝑠 1 + lim 𝐺 (𝑠) 1 + 𝐾𝑝

3 The steady state error of non‐unity feedback systems

The approach to direct computing the steady state error of non‐unity feedback systems has been

demonstrated in the previous example. That is, consider a non‐unity feedback system

Figure 6:

For this system,

y(𝑠) 𝐺𝑝 (𝑠)
= ,
𝑅(𝑠) 1 + 𝐺𝑝 (𝑠)𝐻(𝑠)

𝐺𝑝 (𝑠)
𝐸(𝑠) = 𝑅(𝑠) − y(𝑠) = (1 − ) 𝑅(𝑠)
1 + 𝐺𝑝 (𝑠)𝐻(𝑠)

Let 𝑟(𝑡) be a unit step, and 𝑅(𝑠) = 1/𝑠. Then from the Final Value Theorem,

𝐺𝑝 (𝑠) 1
𝑒i⊃S = lim 𝑠 (1 − )
𝑠→0 1 + 𝐺𝑝 (𝑠)𝐻(𝑠) 𝑠

𝐺𝑝 (0)
= 1− .
1 + 𝐺𝑝 (0)𝐻(0)

Thus, to achieve a zero steady state error, we require

𝐺𝑝 (0)
= 1, or 𝐺𝑝 (0) = 1 + 𝐺𝑝 (0)𝐻(0) .
1+𝐺𝑝 (0)𝐻(0)

Refrences
Basic control systems engineering Paul H. Lewis & Chang Yang

Control system engineering U.A. Bacshi & S.C. Goyal

Nagoor Kani - Control System Engineering

Control Systems by Srivastava

10