Nivaar worked solution (AI-drafted; not reviewed by a licensed engineer)
Notes on this paper
Paper format. 07-Mec-A3 System Analysis and Control,
National Exams May 2015 — 3 hours, closed book (approved Casio or
Sharp calculator and semi-log graph paper only). Six questions; the paper instructs that
"any four questions constitute a complete paper" and that all questions are of equal
value, so each question is worth 25 marks. All six are solved here. A Laplace-transform table
is supplied with the paper (pages 5–6).
Reference texts. K. Ogata, Modern Control Engineering, 5th ed.
(Pearson) — Ch. 5 (transient response), Ch. 6 (root locus), Ch. 7 (frequency
response); N. S. Nise, Control Systems Engineering, 8th ed. (Wiley) — Ch. 4,
6, 8, 10; R. C. Dorf and R. H. Bishop, Modern Control Systems, 13th ed.
(Pearson); G. F. Franklin, J. D. Powell and A. Emami-Naeini, Feedback Control of
Dynamic Systems, 8th ed. (Pearson).
Given. Two open-loop transfer functions:
(a) $G_a(s) = \dfrac{4}{s+2}$, a single real pole at $s=-2$;
(b) $G_b(s) = \dfrac{4}{(0.4s+1)(s+1)}$, two real poles written already in
time-constant form with $\tau_1 = 0.4\ \text{s}$ and $\tau_2 = 1\ \text{s}$.
Find. The straight-line (asymptotic) log-magnitude characteristic of
each function in decibels against $\log_{10}\omega$, with the low-frequency level, every
corner frequency, and every asymptote slope identified.
Approach. Put each factor into the normalised Bode form
$(1+j\omega\tau)$, read the low-frequency asymptote from the resulting constant, then add
one $-20\ \text{dB/decade}$ break at each pole corner $\omega_c = 1/\tau$.
Normalise (a) to time-constant form. A Bode plot is drawn from
factors of the form $(1+j\omega\tau)$, so the pole binomial must be scaled so that its
constant term is unity:
$$G_a(s)=\frac{4}{s+2}=\frac{4}{2\left(\tfrac{s}{2}+1\right)}=\frac{2}{0.5s+1}$$
The Bode gain is therefore $K_B = 2$, not 4. This normalisation is the single most common
place to lose marks: reading the gain as 4 puts the whole plot 6 dB too high.
Low-frequency asymptote of (a). As $\omega \to 0$ the denominator
tends to unity, so
$$20\log_{10}|G_a| \to 20\log_{10}(2) = \boxed{6.02\ \text{dB}}$$
This horizontal line is the entire plot below the corner.
Corner frequency and high-frequency slope of (a). The break occurs
where the imaginary part of the normalised factor equals its real part, i.e. at
$\omega_c = 1/\tau = 1/0.5 = 2\ \text{rad/s}$. Above it the single pole contributes
$-20\ \text{dB/decade}$, so the magnitude falls linearly at that rate. Starting from
6.02 dB at the corner, the asymptote reaches $0\ \text{dB}$ after
$6.02/20 = 0.301$ decade, i.e. at $\omega = 2 \times 10^{0.301} = 4\ \text{rad/s}$
— one octave above the corner. The exact curve runs $3.01\ \text{dB}$ below the
asymptote at the corner, and it crosses unity gain slightly earlier than the asymptote
predicts: $|G_a(j\omega)|=1$ requires $|j\omega+2| = 4$, giving
$\omega = \sqrt{12} = 3.46\ \text{rad/s}$.
Question 1(a): asymptotic (solid) and exact (dashed) magnitude of $G_a(s)=4/(s+2)$. Corner at 2 rad/s; the exact curve is 3 dB below the break point.
Part (b) needs no algebraic normalisation because both binomials are already printed in time-constant form, so the constant 4 is the Bode gain.
Low-frequency asymptote of (b). With both factors tending to
unity,
$$20\log_{10}|G_b| \to 20\log_{10}(4) = \boxed{12.04\ \text{dB}}$$
Locate both corners and accumulate the slopes. The corner of a
$(1+j\omega\tau)$ factor is $\omega_c = 1/\tau$:
$$\omega_{c1}=\frac{1}{1}=1\ \text{rad/s},\qquad
\omega_{c2}=\frac{1}{0.4}=2.5\ \text{rad/s}$$
The slope is therefore $0\ \text{dB/dec}$ below 1 rad/s, $-20\ \text{dB/dec}$
between 1 and 2.5 rad/s, and $-40\ \text{dB/dec}$ above 2.5 rad/s.
Evaluate the asymptote at the second corner. Falling at
$-20\ \text{dB/dec}$ from 12.04 dB over the interval $1 \to 2.5\ \text{rad/s}$
(a factor 2.5, i.e. $\log_{10}2.5 = 0.398$ decade):
$$12.04 - 20(0.398) = 4.06\ \text{dB}$$
so the second break starts from $4.06\ \text{dB}$ and the $-40\ \text{dB/dec}$ tail runs
from there. Solving $|G_b(j\omega)| = 1$ numerically gives the exact unity-gain crossing
at $\omega = 2.59\ \text{rad/s}$; the asymptotic construction predicts
$2.79\ \text{rad/s}$, the usual few-per-cent optimism when two corners lie close
together.
Question 1(b): asymptotic (solid) and exact (dashed) magnitude of $G_b(s)=4/[(0.4s+1)(s+1)]$, showing the 0 / −20 / −40 dB-per-decade segments.