NLS, UNIT-V, KARTHIK THIRUMALA JULY 2022

Networks and Linear Systems

Unit – V
Introduction to discrete time system – difference equations –
z-transforms – inverse z-transforms for typical signals – pulse transfer
function – solution of difference equation – stability analysis

Dr. Karthik Thirumala

Email: [email protected]

Reference textbook:
1) I. J. Nagrath, M. Gopal, ‘Control Systems Engineering’, New Age International Publisher, 4th
edition, 2008. Chapter 11 – Digital Control Systems.
2) D. Roy Choudhury, ‘Networks and Systems’, New Age International Publications, 1st Edition,
2013.

1
 NLS, UNIT-V, KARTHIK THIRUMALA JULY 2022

Lecture - 1
Introduction
In the analysis dealt so far, the signal in the system is a continuous function of time. In particular, the
controller elements are such that the controller produces continuous time signals from continuous time
signal input signals. Such a controller is referred as an analog controller. As the complexity of a control
system increases there arises the demands of flexibility, adaptability and optimality. In fact constructing
a complex control function may even become technically infeasible, if one is restricted to use only
analog elements. The use of digital computer, microprocessor, embedded microcontrollers, as a
compensator or controller device has grown during the past two decades. A digital controller has the
versatility that its control function can be easily modified by changing a few program instructions or
even the entire program. Digital controllers used in the digital control systems have the inherent
characteristics that they accept the data as a short duration pulses (i.e., sampled or discrete data) and
produce a similar kind of output as a control signal. The figure shows a simple control scheme
employing a digital controller.

2
 NLS, UNIT-V, KARTHIK THIRUMALA JULY 2022

Discrete time signals are arise when the continuous time signals are sample. According to the Shanon’s
sampling theorem, the information contained in a signal is fully preserved in the sample version as long
as the sampling frequency is at least twice the maximum frequency contained in the signal. In this
lecture, we shall restrict ourselves to uniform periodic sampling. The mathematical model of a sampled-
data control system is essentially in the form of difference equations. The analysis and design of
sampled-data systems with linear elements may be carried out by use of the z-transform, which was
evolved from the Laplace transform as a special form and was later established on its own merit.

Sample and hold circuit

The sample and hold circuit, as its name implies, samples an input signal and holds to its last sampled
value until the input is sampled again. Figure shows a sample-and-hold circuit using op-amp with an E-
MOSFET. In this circuit, the MOSFET works as a switch that is controlled by the control voltage vs.
The capacitor serves as a storage element. The analog signal vin to be sampled is applied to the drain,
and the vs is applied to the gate of the MOSFET. During the pulse, the MOSFET conducts and acts as
closed switch. This allows the input voltage to charge capacitor. So, the input voltage appears across
the capacitor and in turn at the output as shown in the waveform. On the other hand, when vs is zero,
the MOSFET is off, it acts as an open switch. The only discharge path for C is through the op-amp.
However, the input resistance of the op-amp voltage follower is very high, hence the voltage across the
C is retained. The time period of the sample-and-hold control voltage vs during which the voltage across
the C is equal to the input voltage are called sample periods Ts. The time periods of vs during which the
voltage across C is constant are called hold periods Th. The output of the op-amp is usually
processed/observed during the hold periods. To obtain close approximation of the input waveform, the
frequency of the sample-and-hold control voltage must be significantly higher than that of the input.

3
 NLS, UNIT-V, KARTHIK THIRUMALA JULY 2022

Signal Reconstruction

Lecture – 2
Introduction to difference equation
Difference equations can be viewed either as a discrete analogue of differential equations, or
independently. They are used for approximation of differential operators, for solving mathematical
problems with recurrences, for building various discrete models, etc. Computers can only work with
discrete data, so continuous equations must be discretized before they can be solved numerically. To
illustrate a connection between difference equations and differential equations, let us begin with a
homogeneous, first order, constant-coefficient ordinary differential equation

A simple discretization method is to use finite differences and approximate the derivative by the
expression

This is a forward difference equation. As an illustration we will convert the differential equation, to a
difference equation by using a forward-difference approximation as

4
 NLS, UNIT-V, KARTHIK THIRUMALA JULY 2022

To simplify the notation let y[n] = y(nΔt) where the square brackets [ ], distinguish a function of discrete
time from a function of continuous time which is indicated by using parentheses, ( ). In this notation,
time is not explicitly indicated but, since the time between consecutive discrete-time values of the
function, y is always Δt, we do not need to explicitly indicate time. Some authors use the notation, yn
= y(nΔt) for the nth value of y. This is an exact equivalent of y[n].
Using simplified notations, the differential equation becomes difference equation

Note carefully that the term n2 in our fourth example does not imply non-linearity since linearity is
determined by the yn terms.

Example:

5
 NLS, UNIT-V, KARTHIK THIRUMALA JULY 2022

Linear Constant Coefficent Difference Equation

• General Advance Operator form

o The order of the system is M, N. Usually N ≥ M for causality. Hence, the order is N.

• General Delay Operator form

o If a non-unity coefficient of y[k] can be divide to force unity coefficient.

o Requires initial conditions up to the order to perform computations
o Initial Conditions:

• Shorthand notation form

Example:

Example:

6
 NLS, UNIT-V, KARTHIK THIRUMALA JULY 2022

Example:

In, general the shorthand notation for advanced operator form of difference equation is

The straight-forward method of modelling of sampled-data systems gets very messy for higher-order
systems. In this, we apply z-transforms to obtain the solution of certain types of difference equation.
We shall see that this is done by turning the difference equation into an ordinary algebraic equation.
We investigate both first and second order difference equations. A key aspect in this process is the
inversion of the z-transform.

7
 NLS, UNIT-V, KARTHIK THIRUMALA JULY 2022

Lecture – 3
Z - Transform
The z-transform is an important tool in the analysis and design of discrete-time systems. It simplifies
the solution of discrete-time problems by converting LTI difference equations to algebraic equations
and convolution to multiplication. Thus, it plays a role similar to that served by Laplace transforms in
continuous-time problems.

One-sided or unilateral z-transform of a sequence

Two-sided or bilateral z-transform of a sequence

Basic expansions

Region of Convergence:
The Fourier transform does not converge for all sequences i.e., the infinite sum may not always be
finite. Similarly, the z-transform does not converge for all sequences or for all values of z. The region
of convergence or the ROC is simply range of a value of Z for which the z transform converges. ROC
allows us to determine the inverse z–transform uniquely. The ROC cannot contain any poles.

Convergence requires that

which is only the case if |az−1| < 1, or equivalently |z| > |a|. In the ROC, the series converges to

Since it is just a geometric series, the z-transform has a region of convergence for any finite value of a. When
the signal has many causal terms, the ROC is outside the circle having the largest radius. From the example
for the right-handed exponential sequence, the first term in this sum converges for |z| > 0.5, and the second
for |z| > 0.3. The combined transform X(z) therefore converges in the intersection of these regions, namely
when |z| > 0.5.

8
 NLS, UNIT-V, KARTHIK THIRUMALA JULY 2022

Left sided-sequence:
Now consider the sequence x[n] = −anu[−n − 1]. This sequence is left-sided because it is nonzero only
for n ≤ −1. The z-transform is

Note that the expression for the z-transform (and the pole zero plot) is the same as for the right-handed
exponential sequence, only the region of convergence is different. Whereas, when the signal has many
anti-causal terms, the ROC is inside the circle having the smallest radius.

➢ If x[n] is a right-sided sequence then the ROC extends outward from the outermost finite pole to
infinity.
➢ If x[n] is left-sided then the ROC extends inward from the innermost nonzero pole to z = 0.
➢ A two-sided sequence has a ROC consisting of a ring in the z-plane, bounded on the interior and
exterior by a pole (and not containing any poles).

Unit impulse function

Unit step function

This z-transform is valid only for |z| > 1, in the region of convergence in the z-plane. The variable z-1
can be regarded as a time delay operator.

9
 NLS, UNIT-V, KARTHIK THIRUMALA JULY 2022

Example:
Obtain the z-transform of the sequence
Solution:

Decayed Exponential Function

The region of convergence is e−aT < |z|. If a > 0, the system is stable and the pole at z = e−aT < 1 is inside
the unit circle. If a < 0, the system is unstable and the pole at z = e−aT > 1 is outside the unit circle.

Set of n and 1/n

Sine and Cosine functions

|Z| > |ejθ|,|e-jθ| → |Z| > 1

10
 NLS, UNIT-V, KARTHIK THIRUMALA JULY 2022

Damped Sine or cosine wave

11
 NLS, UNIT-V, KARTHIK THIRUMALA JULY 2022

Properties of Z- transform
Linearity

2Z
Example: 2 + 4δ(n) +4
Z−1

Time advance property

= ∑∞𝑛=0 𝑥(𝑛 + 1)𝑧

−𝑛

= 𝑧 ∑𝑛=0 𝑥(𝑛 + 1)𝑧 −(𝑛+1)

∞

Let m =n+1 = 𝑧 ∑∞
𝑚=1 𝑥(𝑚)𝑧
−𝑚

= 𝑧[∑∞𝑚=0 𝑥(𝑚)𝑧
−𝑚
− 𝑥(0)]
= 𝑧[𝑋(𝑧) − 𝑥(0)]

(iv) Z[x(n+k)] = zkX(z) – zkx(0)- zk-1x(1) - …. – zx(k-1)

12
 NLS, UNIT-V, KARTHIK THIRUMALA JULY 2022

Example:

One step delay with u(k)

m step delay with u(k)

Multiplication by n or Differentiation in Z-domain

13
 NLS, UNIT-V, KARTHIK THIRUMALA JULY 2022

Example

Hence,

Change of Scale

Similarly, Z[a-nx(n)]=X(Za)

Time Reversal

If then

Frequency shifting

14
 NLS, UNIT-V, KARTHIK THIRUMALA JULY 2022

When an input sequence x(k) is applied to a system having an impulse response g(k), the response y(k)
of the system is given the convolution sum.

The equivalent expression in the z-transform is Y(z)=G(z)X(z)

Example:

Solution

15
 NLS, UNIT-V, KARTHIK THIRUMALA JULY 2022

Lecture – 4
Pulse Transfer Function
Consider a linear time-invariant discrete system that produces an output c(k) for an input sequence r(k).
Such a system can be characterized by its response h(k) to unit discrete impulse. The system is assumed
to be causal so that the output appears only after application of the input.

The operation to obtain the output sequence is indicated by the block diagram shown below

16
 NLS, UNIT-V, KARTHIK THIRUMALA JULY 2022

The S-domain and Z-domain relationship

17
 NLS, UNIT-V, KARTHIK THIRUMALA JULY 2022

A causal system is said to be stable when all its poles lie inside the unit circle, the zeros are irrelevant.
Mathematically the condition is
∣pmax∣<1
where ∣pmax∣ is the magnitude of the largest pole (poles) of the system . Likewise, an anticausal system
is stable when all its poles lie outside the unit circle, again the zeros are irrelevant.
∣pmin∣>1
where ∣pmin∣ is the magnitude of the smallest pole (poles).

Inverse Z-transform
Methods to solve Inverse Z-Transform are

Expansion or long-division
We first use long division to obtain as many terms as desired of the z-transform expansion; then we use
the coefficients of the expansion to write the time sequence. The following two steps give the inverse
z-transform of a function F(z)

18
 NLS, UNIT-V, KARTHIK THIRUMALA JULY 2022

The number of terms i obtained by long division is selected to yield a sufficient number of points in the
time sequence.

Example: Obtain the inverse z-transform of the function

Solution:

Inverse Transformation gives

Partial Fraction Expansion

This method is almost identical to that used in inverting Laplace transforms. However, because most z-
functions have the term z in their numerator, it is often convenient to expand F(z)/z rather than F(z).
The procedure for inverse z-transformation is

19
 NLS, UNIT-V, KARTHIK THIRUMALA JULY 2022

Example:

Solution: It is instructive to solve this problem using two different methods. First, we divide by z; then we obtain
the partial fraction expansion. Second, directly obtain the partial fraction for F(z).

Although it is clearly easier to obtain the partial fraction expansion without dividing by z, inverse
transforming requires some experience. There are situations where division by z may actually simplify
the calculations

Example:

Practice problems:

20
 NLS, UNIT-V, KARTHIK THIRUMALA JULY 2022

Lecture - 5
Solution of difference equation using Z-transform
By a process analogous to Laplace transform solution of differential equations, one can easily solve
linear difference equations. The equations are first transformed to the z-domain (i.e., both the right-
and left-hand sides of the equation are z-transformed). Then the variable of interest is solved for and
inverse z-transformed. To transform the difference equation, we typically use the time delay or the
time advance property.

Consider a general difference equation

Where, y(k) is output sequence, and u(k) is input sequence. In control systems, b0 is often 0, as input
u(k) does not immediately affect output y(k). Take the Z-transform of the difference equation
considering u(k) = 0 and y(k) = 0 for k<0.

21
 NLS, UNIT-V, KARTHIK THIRUMALA JULY 2022

Example:

>

Example:

22
 NLS, UNIT-V, KARTHIK THIRUMALA JULY 2022

Example

Example:

Solution:

23
 NLS, UNIT-V, KARTHIK THIRUMALA JULY 2022

This solution is of course readily obtained by direct recursive solution of difference equation without
need for z-transforms. In this case the general term (a + nd) is readily seen from the form of the recursive
solution.

Example: Second order equation

Practice:

Answers:

24
 NLS, UNIT-V, KARTHIK THIRUMALA JULY 2022

Lecture – 6
Analysis of Sampler and Zero-Order Hold
Sampled-data control systems are in fact hybrid systems as they have both discrete-time and
continuous-time signals. Special techniques are therefore required for analysis of these systems. As
ZOH holds the input signal value for a period T, it means that for a short-duration (Δ) input pulse,
it produces an output pulse of duration T, the sampling period.

The ZOH output pulse appearing at kT instant can be expressed as

25
 NLS, UNIT-V, KARTHIK THIRUMALA JULY 2022

We can give a new input-output interpretation to the above equation. Taking inverse Laplace transform
of the infinite series, we get

Thus the output o(t) of pulse sampler and ZOH can be produced by impulse sampled i(t) when passed
through a transfer function

This is illustrated in the figure, which offers ease of analysis.

Consider now the basic sampled-data control system whose block diagram is depicted in figure .

In terms of impulse sampling the diagram can be redrawn as

In z-transform form, we can write the transfer function as

The above block diagram is modified by the inclusion of a digital computer, which modifies the
strength of the pulses from the sampler in accordance with a set of equations. The action of the
digital computer in the z-domain is therefore that of a z-transfer function, D(z). The block diagram
with digital computer in forward path of basic sampled data system is

26
 NLS, UNIT-V, KARTHIK THIRUMALA JULY 2022

The transfer function with digital computer is

Stability Criterion

As the poles of the closed loop transfer function are the same as the roots of the system’s
characteristic equation, the system stability is determined by the roots of the characteristic equation.
For a simple system shows above, the equation is

For a system with digital computer, the equation is

27
NLS, UNIT-V, KARTHIK THIRUMALA JULY 2022

28
 NLS, UNIT-V, KARTHIK THIRUMALA JULY 2022

Example:

29