GATE Electronics (EC): Control Systems (block diagrams, time response, Routh and root locus, Bode, Nyquist, compensators, state space), previous-year questions solved, part 2
Part 2 of 3 of the Control Systems hub: 9 more official GATE EC questions on stability: Routh-Hurwitz and root locus, each solved step by step and checked against the official answer key. Part 1 covers block diagrams, signal flow graphs and transfer functions, time response and steady-state error, frequency response: Bode, Nyquist and compensators and state-space models.
Questions from official GATE EC papers (IIT Kharagpur 2022, IIT Kanpur 2023, IISc Bengaluru 2024, IIT Roorkee 2025, IIT Guwahati 2026); solutions by Skill-Lync. · Updated 17 Nov 2026 · 9 of 25 questions on this page
Root locus with five asymptotes at 36, 108 and 180 degrees, centre at -1: which G(s)?
Select the transfer function that results in the root-locus plot of the closed-loop system as shown in the figure.
- (A)
- (B)
- (C)
- (D)
Answer: (A) Official key: A
- The figure shows five branches that run off to infinity along straight lines. The rays make angles , , , and with the real axis, apart, and they all start from the point . So the number of asymptotes is and the centroid is .
- Asymptote angles for : . For these are , which matches the figure.
- Centroid: . Test each option.
- (A) , , five poles at : . Matches.
- (B) , , poles are the fifth roots of , which add up to : . Same angles, wrong centre.
- (C) , but , and the zero at is in the right half plane. Wrong centre.
- (D) , . Wrong centre.
- Only (A) has five asymptotes centred at . Answer: , option (A).
Trap: option (B) has exactly the same five asymptote angles, so a reader who checks only the angles cannot tell (A) from (B). The code took each option, formed the closed-loop polynomial at with np.roots, and read off the five largest roots. The centres came out as (A) , (B) , (C) and (D) , all agreeing with the centroid formula, and the angles for (A) are -108, -36, 36, 108, 180 degrees. Only (A) gives centre with the angles in the figure.
Asymptote angles and centre tell a control engineer at a glance how a loop behaves at high gain: five or more excess poles means some branches always cross into the right half plane, so gain has to be limited. That is the first thing checked when a high-order servo or filter loop starts to ring as gain is raised.
Even polynomial s^4 + 5s^2 + 4 + K: range of K for which all roots lie on the imaginary axis
,
where is an unknown real parameter. The complete range of for which has all its roots on the imaginary axis is ________.
- (A)
- (B)
- (C)
- (D)
Answer: (A) Official key: A
- has only even powers, so its roots come in pairs . Put : .
- A root is on the imaginary axis exactly when is real and (then , with giving ). So both roots of the quadratic must be real and not positive.
- Real roots: discriminant , so and .
- Not positive: the sum of the roots is , which is negative, so the only way to get a positive root is a negative product. Need product , so .
- Both conditions: , option (A). At the roots are . At they are the double pair , both on the axis.
Trap: stopping at the discriminant, , and forgetting the lower limit. For the product is negative, so one root is positive and is a real pair off the axis. The code swept from to in steps of and tested the roots from np.roots: the set where every root has real part below in magnitude runs from to with no gaps. Just outside, gives a real pair and gives complex roots with real part . , which lies inside options (C) and (D), gives a real pair , and , inside (B), gives roots with real part . Only (A) matches the swept set exactly.
Roots that sit exactly on the imaginary axis mark the edge of stability, and finding the gain at which that happens is how the limit of a loop gain or a PID setting is located, for example the ultimate gain used in Ziegler-Nichols tuning of a temperature or motor loop. An even auxiliary polynomial in a Routh array is the same object.
Unity feedback, G = K/((s+1)(s+2)(s+3)): for which K does the impulse response decay faster than e^(-t)?
- (A)
- (B)
- (C)
- (D)
Answer: (A) Official key: A
- Closed-loop characteristic equation: .
- The impulse response is a sum of terms over the closed-loop poles . It decays faster than when every pole has .
- Shift the axis by substituting (so becomes ). Then , and the equation becomes .
- Routh array in : row : . Row : . Row : . Row : .
- No sign change needs and , so is the exact range.
- Of the four options only (A), , lies inside . Options (C) and (D) have negative , and (B) has . So option (A).
Trap: option (B). Plain stability of the same loop (poles left of the axis, with no shift) holds for , and lies inside that range, so a student who only tests ordinary stability is tempted by (B). The code found with np.roots that at the largest real part of a pole is -0.957, stable but to the right of , so the response decays slower than . Scanning finely gave the exact range 0.00 to 6.00 for poles left of (and -6.0 to 60.0 for plain stability). It also checked each option over points of its interval: only (A) keeps every pole left of , and at the largest real part is -1.048.
Asking for poles to the left of a line, not just the imaginary axis, is how a settling-time or relative-stability spec is turned into a gain limit. A drive or process-control engineer uses it to make sure a loop returns to its setpoint with a decay faster than a chosen rate.
Plant 1/s^2 with C(s) = K(s+a)/(s+4): value of a for closed-loop poles at -1 +/- j sqrt(3)
- (A)
- (B)
- (C)
- (D)
Answer: (B) Official key: B
- Characteristic equation: , that is .
- It is a cubic, so it has three roots. Two are , which multiply to . Call the third root . Then .
- Match : , so (the sum of the three roots is fixed at , so the third root is ).
- Match : .
- Match : , so , option (B).
- Check with the angle condition at : the angle of is and of is . The root locus needs , so . That means is purely imaginary, so and . The magnitude condition gives , the same .
Trap: option (C), , is the value of the third pole's distance from the origin, , which is easy to confuse with the zero position . Pole and zero are different things here: the pole is at and the zero is at . The code solved the three coefficient equations with sympy and got , , . It then ran np.roots on for each option's over a fine grid of from to and measured how close any root comes to : it reaches it (distance below ) only for . For and the roots are -3.09+0.00j, -0.46+2.23j, -0.46-2.23j, nowhere near the target. The angle condition and magnitude condition gave and , matching the algebra.
Placing closed-loop poles with a lead-type compensator is the basic recipe for attitude and gimbal control, where a double integrator () plant needs the extra phase from the compensator zero to be stable at all. Satellite reaction-wheel loops and camera-gimbal loops are tuned this way.
Root locus of unity feedback loop with K G(s): for what K is there a pole at -1 + j1?
For what value(s) of will the system in Figure (i) have a pole at ?
- (A)
- (B)
- (C)For no positive value of
- (D)For all positive values of
Answer: (C) Official key: C
- Read from Figure (ii): open-loop zeros at and , poles at and . So , and the closed-loop poles solve . (Check: the real-axis locus lies on and , where an odd number of poles and zeros lie to the right, as in the figure.)
- A point is a closed-loop pole for some real only if is a negative real number (angle ), because .
- At : .
- has angle , not . It is not a negative real number, so has no real solution: is imaginary.
- Equivalently, the complex part of the locus is the circle with centre and radius , and the point lies at distance from the centre, outside it.
- So the loop has a pole at for no positive value of , option (C).
Trap: option (B), , uses only the magnitude condition with , and forgets the angle condition. Option (A), , is the same slip inverted. The code swept over values from to with np.roots of : the closest any root comes to is , never zero. It also solved the characteristic equation at (roots and ) and (roots and ), and neither is . All 15252 complex roots found lie on the circle of the figure, and with angle .
Engineers use the angle and magnitude conditions of the root locus to check a pole-placement target before choosing a gain: if the angle condition fails at the desired pole, no amount of gain from a plain proportional controller will put a pole there, and a lead or lag network is needed. This shows up in servo and position-loop tuning.
G(s) = 1/(10 s^2) in unity feedback: can M = KI/s or M = KP + s KD give a stable loop?
Consider the following statements.
(i) There exists no controller of the form , where is a positive real number, such that the closed loop system is stable.
(ii) There exists at least one controller of the form , where and are positive real numbers, such that the closed loop system is stable.
Which one of the following options is correct?
- (A)(i) is TRUE and (ii) is FALSE
- (B)(i) is FALSE and (ii) is TRUE
- (C)Both (i) and (ii) are FALSE
- (D)Both (i) and (ii) are TRUE
Answer: (D) Official key: D
- The loop is unity negative feedback, so the closed-loop poles solve .
- Statement (i): gives , that is . The and coefficients are both zero. A polynomial with a missing power (and nonzero coefficients elsewhere) cannot have all roots in the left half-plane, since the Routh-Hurwitz necessary condition needs every coefficient present and of the same sign. Directly: has three roots of equal magnitude at , and on the circle, so two have positive real part. For every the loop is unstable, so (i) is TRUE.
- Statement (ii): gives , that is . A second-order polynomial is stable exactly when all three coefficients are positive, which holds for every positive and . For example gives roots . So (ii) is TRUE.
- Both statements are true: option (D).
Trap: option (B), (i) FALSE, comes from thinking integral action can cure a double integrator. The code computed the roots of for values of from to : the rightmost root always has a positive real part (the smallest value found is , so none is stable); for they are , and . Option (A) or (C) comes from thinking a plant with two poles at the origin cannot be stabilised by a PD controller. The code checked pairs on a log grid from to and every closed loop was stable (the rightmost real part never exceeded ). The derivative term supplies the damping that the plant lacks.
A mass, a satellite or a motor shaft with no friction behaves as a double integrator, and the standard fix is exactly a PD (or PID) loop: the proportional term sets stiffness and the derivative term adds damping. Pure integral control on such a plant makes it oscillate and then diverge, which is why integral action is added only on top of a PD.
p(s) = s^5 + 7s^4 + 3s^3 - 33s^2 + 2s - 40: roots with negative, zero and positive real parts
is the number of roots of with negative real parts.
is the number of roots of that are purely imaginary.
is the number of roots of with positive real parts.
Which one of the following options is correct?
- (A) and
- (B) and
- (C) and
- (D) and
Answer: (A) Official key: A
- Build the Routh array of . Rows and hold the coefficients. Each next entry is from the two rows above.
- row: and .
- row: and .
- row: and the next entry is too. A whole row of zeros means the polynomial has roots in symmetric pairs (equal and opposite). Use the auxiliary polynomial from the row above, , that is , roots . These are purely imaginary, so .
- Replace the zero row with the derivative of the auxiliary polynomial, . The entry becomes , and the entry is .
- First column: . There is exactly one sign change (from to ), so root in the right half-plane.
- Total degree is , so . Check by factoring: , roots .
- : option (A).
| Row | Entries |
|---|---|
| (whole row), replaced by | |
Trap: option (B), , comes from seeing the pair and assuming the remaining three roots are stable. The code computed the roots with np.roots: , , and , and the factorisation with sympy. The root at is in the right half-plane, so , not . It also built the Routh array in exact fractions and got the same first column and the same count of 1 sign change, with the zero row replaced through the auxiliary polynomial.
Routh-Hurwitz is the quick pen-and-paper stability test used before simulating a loop, for example to find the largest gain of a power-converter or flight-control loop before it starts to oscillate. A row of zeros is the signature of a marginal pair on the imaginary axis, which is the sustained oscillation you see when a loop is pushed to its stability limit.
G(s) = 1/((s+1)(s+2)) with C(s) = K(s+alpha): which (K, alpha) puts poles at -3 +- j root 5?
- (A)2, 3
- (B)3, 4
- (C)2, 4
- (D)3, 3
Answer: (B) Official key: B
- Unity negative feedback around , so the closed-loop poles are the roots of : .
- Expand: .
- The wanted poles give .
- Match coefficients. Of : , so . Constant: , so .
- , option (B). Check: has roots .
Trap: option (D), . It has the right but the constant term becomes , so the poles move to , that is : the same real part, a different damped frequency. Options (A) and (C) have , which gives -coefficient and poles with real part (the code found for (A) and for (C)). The code ran np.roots on the closed-loop polynomial of each option, and only (B) gives . It also built for , in sympy and found the same poles.
Placing closed-loop poles by choosing a gain and a compensator zero is how a PD-type term is tuned in a servo or gimbal loop: the zero adds damping, and the gain sets how fast the pair of poles moves left. A real part of against a frequency of is a damping ratio of about , a typical target for a well-damped position loop.
Unity feedback with K/(s(s+7)(s+11)): gain K > 0 at marginal stability (integer)
Answer: Official key: 1386 to 1386
- Characteristic equation: , that is .
- Routh array: : . : . : . : .
- Marginal stability is when the row becomes zero: , so .
- Check with : real part and imaginary part . So () and .
Trap: using , the product of the two lag poles. The code showed that is well inside the stable range (all roots of have real part below ; the rightmost is ), so it is not marginal. With np.roots, gave all roots in the left half plane, put two roots on the axis at (), and put a pair in the right half plane. A sweep of integer gains found as the first with a root on or right of the axis.
The gain at which a loop starts to oscillate, the ultimate gain, and the oscillation frequency are exactly what the Ziegler-Nichols tuning method measures on a real process to set PID gains. A drive or temperature controller with a lag chain like this has a hard upper limit on proportional gain, and designers keep well below it to leave a gain margin.