Experiment 2

Transient Response of a Second-Order System

ECEN 2260

1. Introduction L
In this experiment, you will measure + +
the unit step response of a second-
order system. You will observe the vin(t) C R vout(t)
1 µF 470 Ω
effect of damping factor on
– –
overshoot and rise time, and you will
modify the circuit to obtain critical Fig. 1. Schematic of R–L–C network board.

damping. Your measurements will be

performed on the R-L-C circuit illustrated in Fig. 1.

2. Transient response of the general second-order system

Consider a circuit having the following second-order transfer function H(s):
vout(s) H0
= H(s) =
vin(s) 1 + 2ζ ωs + ωs
2

0 0
(1)
where H0, ζ, and ω0 are constants that depend on the circuit element values R, L, C, etc. In
the usual case of a passive circuit containing real positive inductor, capacitor, and resistor
values, the parameters ζ and ω0 are positive real numbers. The constants H0, ζ, and ω0 are
found by comparing Eq. (1) with the actual transfer function of the circuit. It is common
practice to measure the transient response of the circuit using a unit step function u(t) as an
input test signal:
vin(t) = (1 V) u(t) (2)
The initial conditions in the circuit are set to zero, and the output voltage waveform is
measured.
This test approximates the conditions of transients often encountered in actual
operation. It is usually desired that the output voltage waveform be an accurate
reproduction of the input (i.e., also a step function). However, the observed output voltage
waveform of the second order system deviates from a step function because it exhibits
ringing, overshoot, and nonzero rise time. Hence, we might try to select the component
values such that the ringing, overshoot, and rise time are minimized.

RWE rev 2 1/20/99 1

 ECEN 2260 Circuits Laboratory 2

The output voltage waveform vout(t) can be found using the Laplace transform. The
transform of the input voltage is

vin(s) = 1s (3)
The Laplace transform of the output voltage is equal to the input vin(s) multiplied by the
transfer function H(s):
H0
vout(s) = H(s) vin(s) = 1s
1 + 2ζ ωs + ωs
2

0 0
(4)
The inverse transform is found via partial fraction expansion.
The roots of the denominator of vout(s) occur at s = 0 and (by use of the quadratic
formula) at

s 1, s 2 = – ζω0 ± ω0 ζ 2 – 1 (5)
Three cases occur:
• ζ > 1. The roots s1 and s2 are real. This is called the overdamped case.
• ζ = 1. The roots s1 and s2 are real and repeated: s1 = s2 = –ζω0. This case is
called critically damped.
• ζ < 1. The roots s1 and s2 are complex, and can be written

s 1, s 2 = – ζω0 ± jω0 1 – ζ 2 (6)

This is called the underdamped case.
Figure 2 illustrates how the positions of the roots, or poles, vary with ζ. For ζ = ∞, there
are real poles at s = 0 and at s = –∞. As ζ decreases from ∞ to 1, these real poles move
towards each other until, at ζ = 1,
Im (s)
they both occur at s = –ω0. Further
decreasing ζ causes the poles to ζ=0
jω 0
become complex conjugates as
given by Eq. (6). Figure 3
illustrates how the poles then move
ζ=∞ ζ=1 ζ=∞ Re (s)
around a circle of radius ω0 until, at –∞ – ω0
ζ = 0, the poles have zero real parts
and lie on the imaginary axis.
Figure 2 is called a root locus
– jω0
diagram, because it illustrates how ζ=0
RHP
the roots of the denominator LHP
polynomial of H(s) move in the Fig. 2. Location of the two poles of H(s) vs. ζ, as
described by Eqs. (5) and (6).

2
 ECEN 2260 Circuits Laboratory 2

complex plane as the parameter ζ is Im (s)

varied between 0 and ∞.
Several other cases can be defined jω0

that are normally not useful in practical jω0 1 – ζ 2

engineering systems. When ζ = 0, the ω
0

roots have zero real parts. This is called – ζω0 θ Re (s)

the undamped case, and the output – ω0

voltage waveform is sinusoidal. The

transient excited by the step input does – jω0 1 – ζ 2

not decay for large t. When ζ < 0, the – jω0

roots have positive real parts and lie in the
right half of the complex plane. The
Fig. 3. For 0 ≤ ζ < 1, the complex conjugate
output voltage response in this case is poles lie on a circle of radius ω0.
unstable, because the expression for
vout(t) contains exponentially growing terms that increase without bound for large t.
Partial fraction expansion is used below to derive the output voltage waveforms for
the cases that are have useful engineering applications, e.g. the overdamped, critically
damped, and underdamped cases.

2.1. Overdamped case, ζ > 1

Partial fraction expansion of Eq. (4) leads to
K K K
vout(s) = s 1 + s – 2s + s – 3s
1 2 (7)
Here, s1 and s2 are given by Eq. (5), and the residues K1, K2 and K3 are given by

H0
K 1 = s 1s
1 + 2ζ ωs + ωs
2

0 0
s=0

1 H0
K 2 = s – s1 s
1 + 2ζ ωs + ωs
2

0 0
s = s1

1 H0
K 3 = s – s2 s
1 + 2ζ ωs + ωs
2

0 0
s = s2 (8)
Evaluation of these expressions leads to

3
 ECEN 2260 Circuits Laboratory 2

K1 = H 0

1 H0 s2
K 2 = s – s1 s =– H
1 – s 1 – ss
s s2 – s1 0
1 2
s = s1

1 H0 s1
K 3 = s – s2 s =– H
1 – s 1 – ss
s s1 – s2 0
1 2
s = s2
(9)
The inverse transform is therefore
s s
vout(t) = H 0 u(t) 1 – s –2 s e s 1t – s –1 s e s 2t
2 1 1 2
(10)
In the overdamped case, the output voltage response contains decaying exponential terms,
and the rise time depends on the magnitudes of the roots s1 and s2. The root having the
smallest magnitude dominates Eq. (10): for | s1 | << | s2 |, Eq. (10) is approximately equal
to

vout(t) ≈ H 0 u(t) 1 – e s 1t
(11)
This is indeed what happens when ζ >>1. Equation (11) can be expressed in terms of ω0
and ζ as
ω 0t
vout(t) ≈ H 0 u(t) 1 – e – 2ζ
(12)
When ζ >> 1, the time constant 2ζ/ω0 is large and the response becomes quite slow.

2.2. Critically damped case, ζ = 1

In this case, Eq. (4) reduces to
H0
vout(s) = 2
s 1 + ωs
0
(13)
The partial fraction expansion of this equation is of the form
K1 K2 K3
vout(s) = + +
s s + ω0 s + ω0
2

(14)
with the residues given by

4
 ECEN 2260 Circuits Laboratory 2

K1 = H 0
H0
K 2 = s + ω0 = – ω0 H 0
2
2
s 1 + ωs
0
s = – ω0

H0
K 3 = d s + ω0
2
2 = – H0
ds s 1 + ωs
0

s = – ω0 (15)
The inverse transform is therefore

vout(t) = H 0 u(t) 1 – 1 + ω0 t e – ω 0t
(16)
In the critically damped case, the time constant 1/ω0 is smaller than the slower time
constant 2ζ/ω0 of the overdamped case. In consequence, the response is faster. This is the
fastest response that contains no overshoot and ringing.

2.3. Underdamped case, ζ < 1

The roots in this case are complex, as given by Eq. (6). The partial fraction expansion of
Eq. (4) is of the form
K1 K2 K *2
vout(s) = + +
s s + ζω0 – jω0 1 – ζ 2 s + ζω0 + jω0 1 – ζ 2 (17)
The residues are computed as follows:

K1 = H 0

H 0ω 20
K 2 = s + ζω0 – jω0 1 – ζ 2
s s + ζω0 – jω0 1 – ζ 2 s + ζω0 + jω0 1 – ζ 2
s = – ζω 0 +
jω 0 1 – ζ 2

(18)
The expression for K2 can be simplified as follows:

5
 ECEN 2260 Circuits Laboratory 2

H 0ω 20
K2 =
– ζω0 + jω0 1 – ζ 2 2 jω0 1 – ζ 2
H0
=
– ζ + j 1 – ζ2 2 j 1 – ζ2
H0
=–
2 1 – ζ + j2ζ 1 – ζ 2
2

(19)
The magnitude of K2 is
H0
K2 =
2 1 – ζ2
(20)
and the phase of K2 is

2ζ 1 – ζ 2
∠K 2 = tan – 1
2 1 – ζ2
ζ
= tan – 1
1 – ζ2
(21)
The inverse transform is therefore

1 ζ
vout(t) = H 0u(t) 1 – e – ζω 0t cos 1 – ζ 2 ω0t + tan – 1
1 – ζ2 1 – ζ2
(22)
In the underdamped case, the output voltage rises from zero to H 0 faster than in the
critically-damped and overdamped cases. Unfortunately, the output voltage then overshoots
this value, and may ring for many cycles before settling down to the final steady-state
value.
In some applications, a moderate amount of ringing and overshoot may be
acceptable. In other applications, overshoot and ringing is completely unacceptable, and
may result in destruction of some elements in the system. The engineer must use his or her
judgment in deciding on the best value of ζ.

3. Step response waveforms

Equations (10), (16), and (19) were employed to plot the step response waveforms of Fig.
4. Underdamped, critically damped, and overdamped responses are shown. It can be
deduced from Fig. 4 that the parameter ω0 scales the horizontal (time) axis, while H 0 scales

6
 ECEN 2260 Circuits Laboratory 2

the vertical (output voltage) axis. The damping factor ζ determines the shape of the
waveform.

2
vout(t) ζ = 0.01
H0 ζ = 0.05

ζ = 0.125
1.5
ζ = 0.25

ζ = 0.5
1
ζ = 0.67
ζ=1
ζ = 1.67
ζ = 2.5
0.5
ζ=5

ζ = 10
ζ = 50
0
0 5 10 15
ω0t, radians
Fig. 4. Second-order system step response, for various values of damping factor ζ.

Three figures-of-merit for judging the step response are the rise time, the percent
overshoot, and the settling time. Percent overshoot is zero for the overdamped and
critically damped cases. For the underdamped case, percent overshoot is defined as

peak vout – vout(∞)

percent overshoot = 100%
vout(∞)
(20)
One can set the derivative of Eq. (19) to zero, to find the maximum value of vout(t). One can
then plug the result into Eq. (20), to evaluate the percent overshoot. Note that the final
(steady-state) value of the output vout(∞) is H 0. The following equation for the percent
overshoot results:

percent overshoot = e – πζ / 1 – ζ2
100%
(21)
Again, this equation is valid only in the underdamped case, i.e., for 0 < ζ < 1. It can be seen
from Fig. 4 that decreasing the damping factor ζ results in increased overshoot. The
overshoot is 0% for ζ = 1. In the limit of ζ = 0 (the undamped case), the overshoot
approaches 100%.
As illustrated in Fig. 5, the rise time is defined as the time required for the output
voltage to rise from 10% to 90% of its final steady-state value. When the system is

7
 ECEN 2260 Circuits Laboratory 2

v (t)
underdamped, the out
output waveform
may pass through
90% of its final overshoot
105%
value several 95% final
value
times; the first 90% v(∞)

pass is used in settling time

computation of the
rise time. It can be rise
10%
seen from Fig. 4 v(∞) of time

that the rise time t

Fig. 5. Salient features of step response, second order system.
increases
monotonically with increasing ζ.
The settling time is the time required by an underdamped system for its output
voltage response to approach steady state and stay within some specified percentage (for
example, 5%) of the final steady-state value. As can be seen from Fig. 4, systems having
very small values of ζ have short rise times but long settling times.

4. Experimental measurement of step response.

The difficulty in measuring a transient vin(t)
response is that it happens only once 1 V
—if you blink, you will miss it! This
problem can be alleviated by causing
the step input to be periodic: apply a
0V
square wave (Fig. 6) to the circuit input. 0 T/2 T t

The duration T/2 of the positive portion Fig. 6. Use of a square wave input, with
sufficiently long period T, allows the output
of the square wave is chosen to be transient to be observed on any oscilloscope.
much longer than the settling time of the output response, so that the circuit is in steady-
state just before each step of the input waveform occurs. In consequence, the output voltage
waveform is identical to the waveform observed when a single step input is applied, except
that the output transient occurs repetitively. The output transient waveform can now be
easily observed on an oscilloscope, and can be studied in detail.

8
 ECEN 2260 Circuits Laboratory 2

BIBLIOGRAPHY
[1] BENJAMIN C. KUO, Automatic Control Systems, New Jersey: Prentice-Hall.

[2] J. D’AZZO and C. HOUPIS, Linear Control System Analysis and Design: Conventional and
Modern, New York: McGraw-Hill, 1995.

Prelab assignment
1. Show that the angle θ identified in Fig. 3 is equal to cos–1 ζ.
2. For the circuit of Fig. 7:
(a) Derive an analytical Rs L
expression for the +
50 Ω
transfer function H(s).
vs(t) + C R vout(t)
(b) Write H(s) in the – 1 µF 470 Ω

standard form of Eq. (1), –

and derive analytical Frequency synthesizer,

Thevenin-equivalent
expressions for H0, ω0, circuit model

and ζ in terms of R, Rs, Fig. 7. Circuit to be tested in Exp. 4.

L, and C.
(c) For the values
R = 470 Ω Rs = 50 Ω
C = 1 µF L = 15 mH
find the numerical values of H0, ω0, and ζ. How large is the percent
overshoot? Estimate the rise time.
Note: retain a copy of your answers for part 2; you will need these in the lab.

Laboratory procedure: Part A

3. Observe the output of the function generator using an oscilloscope. Make sure that
the function generator is loaded only by the oscilloscope. Adjust the function
generator so that it produces a 50 Hz square wave, having a maximum value of one
volt and a minimum value of zero volts as in Fig. 6. Note that the function
generator front panel readouts are calibrated for operation with a 50 Ω load; since
the oscilloscope has a high-impedance input, the observed waveform voltages will
not correspond to the front panel readouts of the function generator.
4. Obtain an R–L–C network board from your TA. A schematic of this board is given
in Fig. 1. Connect the input terminals of the board to the function generator. Do

9
 ECEN 2260 Circuits Laboratory 2

not change the settings of the function generator. Observe the step response of the
voltage at the output of your R-L-C board, and measure the overshoot and rise time.
Adjust the oscilloscope time / division and volts / division so that the transient
approximately fills the screen, and then obtain a hard copy.
5. Add resistance in series with the inductor (effectively increasing the value of Rs),
such that the circuit is critically damped. You should use the same oscilloscope
scale settings that you used in part (a). Measure the rise time, and verify that there
is no overshoot. Obtain a hard copy.
6. Add more resistance in series with the inductor, such that the circuit is overdamped
with ζ ≥ 2. Use the same oscilloscope scale settings that you used in part (a).
Measure the rise time, and verify that there is no overshoot. Obtain a hard copy.

Laboratory procedure: Part B

7. Obtain the second R–L–C network board from your TA. This board is similar to
the original R–L–C board, but has different element values. It can be distinguished
by its physically smaller inductor. Repeat steps 3 and 4 using this board. You may
find it useful to increase the frequency of the square wave.
8. Use your measurements of rise time and overshoot to deduce the values of ω0 and
ζ for this network.
9. Read the resistor color code to determine the resistor value. Using your values of
ω0 and ζ from step 8 and your prelab analysis, calculate the values of L and C.
10. Simulate the circuit using your element values calculated in step 9. You may use
either PSPICE or Electronic Workbench. Compare your simulated step response
with your measured waveform. If the responses differ significantly, then refine
your circuit model and element values as necessary such that the simulated and
measured step responses agree.
11. Increase the frequency of the function generator square wave, to the values f0
(where f0 = ω0/2π), 5f0, and f0/5. Obtain a hard copy of the output voltage waveform
for each case. When the square wave frequency approaches or exceeds the network
natural frequency f0, does the output voltage waveform still resemble a square
wave? What happens?

10
 ECEN 2260 Circuits Laboratory 2

Post-lab report
12. Attach copies of all of your data.
13. Summarize the results of your prelab analysis of part 2: state the equations for the
transfer function H(s) and the parameters ω0 and ζ.
14. For the measurements of parts 4, 5, and 6, list the values of Rs, and of the ω0 and ζ
predicted by your theoretical equations. Compare the measured oscilloscope waveforms,
rise times, and overshoot to the theoretical curves of Fig. 4.
15. Report your values computed in steps 8-10, and document your computations of
steps 8 and 9. Include printouts of your EWB or PSPICE input files and step response
plot. Write a few sentences comparing your measured and simulated step responses.
16. Report your results of step 11, and explain physically what happened and why.
17. An equivalent circuit model of a high-speed computer data bus is illustrated in Fig.
8 (one bit is shown). This bus transmits digital data from one integrated circuit (IC) to
another. One IC actively drives the bus with pulses, and is modeled using a Thevenin
equivalent circuit as shown. The voltage source generates rectangular pulses at a frequency
equal to the bus clock frequency. The other IC receives these pulses, and can be modeled
by a capacitance in parallel with a resistor. Each IC connection consists of a bonding wire
and lead, that can be electrically modeled as an inductor. A copper trace on a printed circuit
board connects the IC pins; this trace is modeled here as a series resistance and inductance
(the copper trace may also exhibit significant capacitance to ground, but this capacitance is
neglected in this problem). Values of each model element are given in Fig. 8. It can be seen
that this network is similar to the networks studied in this lab assignment.
(a) Compute the values of ω0 and ζ for the network of Fig. 8.
(b) What is a reasonable upper limit for the clock frequency of this bus?

Bus driver Model of printed circuit

Thevenin equivalent board connection
{

model Model of receiver

200 Ω 4 nH 15 nH 0.5 Ω 4 nH
Inductance of

Inductance of
bonding wire

bonding wire
and lead

and lead

+ 0.5 10
– pF MΩ

11
 ECEN 2260 Circuits Laboratory 2

Experiment 2
Transient Response of a Second-Order System

Lab scoresheet

Instructions: Get your TA to initial each section as you complete it in class. Turn in this
scoresheet with your post-lab report.

Name

CUID#

points initials

Prelab assignment (15 points) turn in your prelab

assignment within 5 minutes of the beginning of
your first lab session. One prelab is required
from each student.

Laboratory procedure

Part 4. (10 points) Measure underdamped

transient response

Part 5. (10 points) Measure critically damped

transient response

Part 6. (10 points) Measure overdamped

transient response

Part 7. (10 points) Measure transient response

Part 10. (10 points) Simulation matches

measurement

Part 11. (10 points) Waveforms with square wave

frequencies of f0/5, f0, and 5f0

Post-lab report (25 points) One report is required from

each team.

12