Frequency Response Interview Questions

Q1. What is frequency response and why is it important in control systems?

Frequency response is the steady-state response of a linear system to a sinusoidal input at various frequencies, characterized by how the gain (magnitude ratio) and phase (angle between input and output) change with frequency. A PI speed controller for a 5 kW induction motor must maintain flat gain and less than 45° phase lag up to 50 Hz to achieve good bandwidth and avoid oscillation in the speed loop. Frequency response analysis reveals stability margins, bandwidth limitations, and resonance peaks without solving differential equations directly.

Follow-up: What is the frequency response of a pure integrator 1/s?

Q2. What is a Bode plot and what are its two components?

A Bode plot consists of two graphs plotted against logarithmic frequency: the magnitude plot showing gain in decibels (20 log|G(jω)|) and the phase plot showing the phase angle ∠G(jω) in degrees, both versus ω on a log scale. For G(s) = 100/(s+10), the magnitude asymptote starts at 20 dB at low frequencies, breaks downward at ω = 10 rad/s (the pole frequency), and falls at −20 dB/decade; the phase shifts from 0° to −90°. Bode plots are drawn on semi-log graph paper and can be constructed by superimposing contributions from each pole, zero, and gain constant separately.

Follow-up: What is a decade in the context of a Bode magnitude plot?

Q3. How does a first-order pole at s = −a affect the Bode magnitude and phase plots?

A first-order pole factor 1/(1 + s/a) contributes 0 dB magnitude and 0° phase for ω << a, transitions at the corner frequency ω = a (where exact gain is −3 dB and phase is −45°), and asymptotically contributes −20 dB/decade slope and −90° phase for ω >> a. For a low-pass RC filter with R = 1 kΩ and C = 1 μF, the pole is at a = 1/RC = 1000 rad/s, so the filter attenuates signals above 1000 rad/s (≈159 Hz) at 20 dB/decade. The actual magnitude at the corner frequency is −3.01 dB, not 0 dB as the asymptote suggests.

Follow-up: How does a second-order pole pair differ from two separate first-order poles in the Bode plot?

Q4. What is gain margin and how do you read it from a Bode plot?

Gain margin (GM) is the additional gain the open-loop system can tolerate before becoming unstable, measured in dB as the negative of the open-loop magnitude in dB at the phase crossover frequency (where phase = −180°). A Type-1 position control loop with GM = 12 dB means the loop gain can be increased by a factor of 4 (20 log 4 ≈ 12 dB) before the closed-loop system becomes unstable. GM is read from the Bode plot by finding the frequency where the phase plot crosses −180° and reading the corresponding gain from the magnitude plot; if gain < 0 dB at that frequency, GM is positive and the system is stable.

Follow-up: What does a negative gain margin indicate about a system's stability?

Q5. What is phase margin and how does it relate to closed-loop damping?

Phase margin (PM) is the additional phase lag the open-loop system can tolerate before instability, measured as 180° plus the open-loop phase at the gain crossover frequency (where magnitude = 0 dB or 1). A PM of 45° for a DC servo motor position loop corresponds to a closed-loop damping ratio ζ ≈ 0.45, producing about 20% overshoot in step response, which is acceptable for most positioning applications. Higher PM gives more damping but reduces bandwidth; a PM of 70° gives near-critically damped response but is conservative for fast-response requirements.

Follow-up: What is the approximate relationship between phase margin and damping ratio for a second-order system?

Q6. What is the gain crossover frequency and how is it related to system bandwidth?

The gain crossover frequency ωgc is where the open-loop magnitude equals unity (0 dB), and it approximately equals the closed-loop bandwidth ωBW for systems with adequate phase margin. An ABB servo drive with gain crossover at 200 rad/s has a closed-loop bandwidth of approximately 200 rad/s, meaning the drive can track sinusoidal position references up to about 32 Hz without significant attenuation. Increasing loop gain shifts ωgc to a higher frequency, increasing bandwidth but also potentially reducing phase margin and causing oscillation.

Follow-up: What is the relationship between closed-loop bandwidth and step response rise time?

Q7. What is a Nyquist plot and how does it relate to Bode plots?

A Nyquist plot is a polar plot of the open-loop transfer function G(jω)H(jω) as ω varies from 0 to ∞, plotting real versus imaginary parts, while a Bode plot shows the same information in separate magnitude and phase graphs versus log frequency. For a stable system, the Nyquist curve must not encircle the −1+j0 point, which is the graphical representation of the Nyquist stability criterion. The gain margin corresponds to how far the Nyquist curve crosses the negative real axis from the −1 point, and phase margin corresponds to the angle between the curve's unit-circle crossing and the negative real axis.

Follow-up: What is the Nyquist stability criterion and how does it handle unstable open-loop systems?

Q8. How do you sketch the Bode plot for G(s) = K·(s+z)/[s(s+p)]?

Factorize into standard form: K·(z/p)·(1+s/z)/[s/ω × (1+s/p)], compute the DC gain asymptote as 20log(Kz/p) dB at ω = 1, draw −20 dB/decade slope from ω = 0 due to the integrator, add +20 dB/decade break at ω = z (zero), net 0 dB/decade between z and p, then −20 dB/decade again above ω = p; phase starts at −90° (integrator), rises +45°/decade near ω = z, and falls −45°/decade near ω = p. For a PI controller with K = 10, z = 1, p = 100, the Bode plot shows the controller adding phase lead near 1–10 rad/s while maintaining flat gain above 100 rad/s. Sketching this in under 2 minutes on a whiteboard demonstrates strong frequency domain fluency.

Follow-up: What is the purpose of the zero in a PI or lead compensator in terms of Bode plot?

Q9. What is resonant peak in the Bode magnitude plot of a second-order system?

The resonant peak Mr is the maximum value of the closed-loop magnitude frequency response above its low-frequency value, occurring at the resonant frequency ωr = ωn√(1−2ζ²) for 0 < ζ < 0.707. A second-order system with ωn = 100 rad/s and ζ = 0.2 has Mr = 1/(2×0.2×√(1−0.04)) ≈ 2.55 (8.1 dB), indicating that sinusoidal inputs near 98 rad/s will be amplified 2.55× by the closed loop. Resonant peaks above 3 dB (Mr > 1.4) indicate damping ratio below 0.35 and typically produce unacceptable overshoot in step response.

Follow-up: What is the relationship between resonant peak Mr and damping ratio ζ in a standard second-order system?

Q10. What is a phase lead compensator and how does it improve frequency response?

A phase lead compensator C(s) = K(s+z)/(s+p) with p > z adds positive phase in the frequency range between z and p, increasing the phase margin at the gain crossover frequency without significantly reducing gain. Adding a lead compensator with z = 10 rad/s and p = 100 rad/s to a marginally stable motor drive loop increases PM from 10° to 50°, improving damping from near-zero to ζ ≈ 0.5, eliminating the speed oscillation observed at no load. The maximum phase lead is arctan[(√(p/z) − √(z/p))/2] = arctan[(α−1)/(2√α)] where α = p/z, occurring at ωm = √(zp).

Follow-up: What is the trade-off of using a very large phase lead compensator (large p/z ratio)?

Q11. What is a lag compensator and what does it do to the Bode plot?

A lag compensator C(s) = K(s+z)/(s+p) with z > p attenuates the gain at high frequencies, shifting the gain crossover to a lower frequency where phase is less lagged, thereby increasing phase margin indirectly without adding phase directly. A lag compensator with z = 1, p = 0.1 applied to a Type-1 loop attenuates high-frequency gain by 20 dB (factor of 10), moving ωgc from 100 rad/s (where PM = −10°) down to 10 rad/s (where PM = 40°), achieving stability at the cost of reduced bandwidth. Lag compensation is used when steady-state accuracy must be maintained but the loop gain is too high for stable operation.

Follow-up: What is the effect of lag compensation on steady-state error compared to lead compensation?

Q12. What is the bandwidth of a control system and what limits it?

Bandwidth is the frequency range over which the closed-loop magnitude stays within −3 dB of its DC value, indicating the range of input frequencies the system can track faithfully, and it is limited by plant dynamics, actuator saturation, sensor noise amplification, and stability requirements. A pneumatic valve actuator with a mechanical resonance at 20 Hz limits the bandwidth of the pressure control loop to about 5 Hz (one-quarter of the resonance frequency) to maintain sufficient gain and phase margins. Extending bandwidth beyond the plant resonance requires notch filters or robust control, which is why industrial PLC PID loops are typically detuned to one-tenth of the resonance frequency.

Follow-up: Why does increasing loop gain eventually reduce stability even though it initially increases bandwidth?

Q13. What is the effect of a pure time delay e^(−τs) on the Bode plot?

A pure time delay e^(−τs) contributes zero magnitude change (|e^(−jωτ)| = 1 for all ω) but adds linearly increasing phase lag of −ωτ radians = −57.3ωτ degrees, which grows without bound as frequency increases. A SCADA-controlled industrial furnace with a 2-second measurement transport delay has −360° × (50/360) = −50° additional phase at 0.25 rad/s, severely reducing phase margin and limiting the maximum usable controller bandwidth to roughly 0.1 rad/s. Time delay is the most common reason industrial processes require detuned PID controllers and why Smith Predictor or internal model control is needed for processes with τ/T > 0.3.

Follow-up: What is the Smith Predictor and how does it compensate for pure time delay in a control loop?

Q14. How do you determine the type number of a system from its Bode plot?

The system type number equals the number of integrators (poles at the origin), identifiable from the Bode magnitude plot as the slope at very low frequencies: Type 0 has a flat 0 dB/decade low-frequency asymptote, Type 1 has −20 dB/decade, and Type 2 has −40 dB/decade as ω → 0. A paper machine speed control loop with −20 dB/decade low-frequency Bode magnitude slope is a Type 1 system, guaranteeing zero steady-state error to a step speed reference. The type number directly determines the steady-state error coefficients (position, velocity, acceleration constants) without solving the closed-loop transfer function.

Follow-up: What steady-state error does a Type 1 system produce in response to a ramp input?

Q15. What is a Nichols chart and when is it used instead of a Bode plot?

A Nichols chart is a graph of open-loop magnitude in dB versus open-loop phase in degrees, with overlaid contours of constant closed-loop magnitude (M-circles) and constant closed-loop phase, allowing closed-loop frequency response to be read directly from the open-loop data. A Nichols chart is used for systems where the closed-loop bandwidth, resonant peak, and phase margin must all be specified simultaneously, such as in aircraft flight control design where gain, phase, and closed-loop bandwidth are all contractual requirements. Modern control toolboxes in MATLAB plot Nichols charts as a single function call (nichols(G)), but interpreting them manually remains a differentiating skill in control-focused technical interviews.

Follow-up: What is the M = 1 (0 dB) contour on a Nichols chart and what is its significance for stability?