8. Classical Feedback Controller Design

Introduction and Context

Feedback control is the principal means by which mechanical systems are made to follow commands, reject disturbances, and operate safely despite uncertainty. In a feedback loop, a measured output is compared with a desired reference, and the resulting error is processed by a controller to determine the actuator command. Classical controller design refers primarily to methods based on transfer functions, frequency response, root-locus analysis, pole placement, and experimentally established tuning rules. Although modern state-space and optimization methods are powerful, classical techniques remain central in industrial motion systems, process machinery, engine control, robotics, vibration suppression, and electromechanical drives.

The central design problem is not simply to make the error small. A useful controller must also provide adequate speed, stability margins, disturbance rejection, robustness to model uncertainty, acceptable actuator effort, and tolerable sensitivity to measurement noise. These requirements frequently conflict. Increasing proportional gain may improve tracking but cause oscillation or actuator saturation. Integral action may eliminate steady-state error but create windup. Derivative action can improve damping, yet amplify high-frequency noise. The practical designer therefore treats controller synthesis as a constrained compromise rather than as the selection of a formula.

This chapter develops the principal classical controllers and their industrial combinations. It also considers feedforward and cascade structures, saturation and anti-windup, noise, tuning methods, and an applied servo-drive case.

Learning Objectives

After studying this chapter, the reader should be able to:

1. Explain the roles of proportional, integral, and derivative action in a closed-loop system.

2. Derive and interpret P, PI, PD, PID, lead, lag, and lead–lag compensators.

3. Select controller structures from tracking, disturbance-rejection, bandwidth, and robustness requirements.

4. Apply analytical, simulation-based, and industrial tuning methods.

5. Design feedforward and cascade control for multivariable or disturbance-prone mechanical systems.

6. Diagnose the effects of actuator saturation, integral windup, sensor noise, dead time, and model uncertainty.

7. Evaluate controller performance using time-domain and frequency-domain measures.

8. Apply professional best practices to controller commissioning and documentation.

The Closed-Loop Design Problem

For a single-input, single-output plant with transfer function \(G(s)\), controller \(C(s)\), and unity feedback, the loop transfer function is

\[

L(s)=C(s)G(s).

\]

The reference-to-output transfer function and the reference-to-error transfer function are respectively

\[

T(s)=\frac{L(s)}{1+L(s)}, \qquad S(s)=\frac{1}{1+L(s)}.

\]

Here, \(T(s)\) is the complementary sensitivity function and \(S(s)\) is the sensitivity function. A large loop gain at low frequency generally gives good tracking and disturbance rejection, because \(S(s)\) becomes small. However, high gain over a wide frequency range can reduce stability margins, increase actuator demand, and transmit measurement noise to the actuator. The identity

\[

S(s)+T(s)=1

\]

expresses an important limitation: sensitivity reduction in one frequency range is normally accompanied by increased sensitivity elsewhere.

For a plant subject to an additive output disturbance \(d\), the disturbance contribution is approximately \(S(s)d\). For measurement noise \(n\), the controller-side noise transmission is approximately \(-C(s)S(s)n\). Thus, disturbance rejection favors high low-frequency loop gain, whereas noise rejection favors limited high-frequency controller gain.

Important performance measures include rise time, settling time, overshoot, steady-state error, bandwidth, gain margin, phase margin, resonant peak, and control effort. A design should specify acceptable values before tuning begins. For a mechanical positioning system, for example, low steady-state error and adequate damping may be more important than the shortest possible rise time.

Proportional, Integral, and Derivative Action

The basic PID controller is written in ideal form as

\[

C(s)=K_p\left(1+\frac{1}{T_i s}+T_d s\right),

\]

where \(K_p\) is proportional gain, \(T_i\) is integral time, and \(T_d\) is derivative time. Equivalently,

\[

u(t)=K_p e(t)+K_i\int_0^t e(\tau)d\tau+K_d\frac{de(t)}{dt}.

\]

Proportional control

A proportional controller produces \(u=K_p e\). Increasing \(K_p\) usually reduces tracking error and increases loop bandwidth. It cannot, however, eliminate steady-state error for a plant without an integrator. Excessive gain can generate oscillation, sensitivity to unmodeled dynamics, and saturation. P control is often adequate where a finite offset is acceptable, where the plant already contains integral behavior, or where simplicity and predictable actuator behavior are important.

Integral control

Integral action accumulates error and introduces a pole at the origin. It increases the system type and can eliminate constant steady-state error for a stable closed-loop system. A PI controller is

\[

C(s)=K_p\left(1+\frac{1}{T_i s}\right)=K_p\frac{T_i s+1}{T_i s}.

\]

The integral term is valuable for load changes and constant disturbances. Its disadvantages are reduced phase margin, slower recovery when poorly tuned, and windup when the actuator saturates. Integral action should therefore be used deliberately, not merely because zero offset is desirable in theory.

Derivative control

Derivative action responds to the rate of change of error and contributes phase lead. It can improve damping and reduce overshoot, particularly in systems with significant inertia or lightly damped modes. An ideal derivative is physically unrealizable and amplifies high-frequency measurement noise. A practical derivative term uses a filter:

\[

C_D(s)=K_p\frac{T_d s}{1+T_d s/N},

\]

where \(N\), commonly between approximately 5 and 20, limits high-frequency gain. Derivative action is often applied to the measured output rather than the error so that a sudden reference step does not create a derivative kick.

PID and its practical forms

The parallel form is

\[

C(s)=K_p+\frac{K_i}{s}+\frac{K_d s}{1+s/\omega_f}.

\]

The series or interacting form represents the controller as cascaded first-order factors. These forms are not numerically identical for the same parameter values; the parameterization must therefore be identified before tuning or transferring settings between software platforms. A two-degree-of-freedom PID controller introduces separate weights for reference proportional and derivative actions. This allows tracking response to be shaped without weakening disturbance rejection.

Lead, Lag, and Lead–Lag Compensation

Classical compensators are often designed from frequency-response requirements. A lead compensator has the form

\[

C_{lead}(s)=K\frac{1+Ts}{1+\alpha Ts}, \qquad 0<\alpha<1.

\]

Its zero occurs below its pole, and it supplies positive phase near the crossover frequency. Lead compensation can increase phase margin, improve transient speed, and shift the root locus toward better-damped poles. The maximum phase contribution is

\[

\phi_{max}=\sin^{-1}\left(\frac{1-\alpha}{1+\alpha}\right),

\]

occurring at \(\omega_m=1/(T\sqrt{\alpha})\). In practice, the desired additional phase is selected with a safety allowance, and the compensator is positioned so that \(\omega_m\) lies near the intended crossover frequency.

A lag compensator is commonly written as

\[

C_{lag}(s)=K\frac{1+Ts}{1+\beta Ts}, \qquad \beta>1.

\]

Its pole is closer to the origin than its zero. Lag increases low-frequency gain relative to higher-frequency gain, improving steady-state accuracy while limiting the increase in bandwidth. It introduces a small negative phase contribution and may slow the response. Lag is appropriate when accuracy must improve but the existing transient response and robustness should be disturbed as little as possible.

A lead–lag compensator combines both effects. Lead is used to recover phase margin or increase speed; lag is used to increase low-frequency loop gain. The two sections should be separated sufficiently in frequency so that their purposes remain distinguishable. A lead–lag design is useful when a plant simultaneously requires faster transient response and improved steady-state accuracy.

Root-locus design provides a complementary interpretation. Adding a lead network can pull dominant closed-loop poles toward a desired damping ratio and natural frequency. Adding lag changes the magnitude condition more strongly at low frequency and can improve error constants without moving the dominant poles excessively. Frequency-response design instead specifies crossover frequency and phase margin, then shapes the loop to meet those targets.

Feedforward, Cascade, and Disturbance-Rejection Structures

Feedback corrects errors after they occur. Feedforward acts on a measured reference or disturbance before its effect appears at the output. If a disturbance enters through \(G_d(s)\), an ideal disturbance feedforward compensator is approximately

\[

F_d(s)=-\frac{G_d(s)}{G(s)},

\]

subject to realizability, sensor availability, and model accuracy. Feedforward is especially effective for measurable load changes, commanded acceleration, gravity compensation, and predictable friction. It cannot by itself correct unmeasured disturbances or model errors, so it is normally combined with feedback.

In a two-degree-of-freedom structure, the feedback controller determines robustness and disturbance rejection, while a prefilter or feedforward path shapes command tracking. This separation is valuable when a system must reject disturbances aggressively but respond smoothly to reference changes.

Cascade control uses an outer loop around an inner loop. The inner loop should be faster, well damped, and capable of rejecting disturbances that enter near the actuator or inner process variable. Examples include position–velocity–current control in electric drives, temperature–flow control in thermal machinery, and position–force control in robotics. The inner loop is tuned first. The outer loop is then designed using the closed inner loop as part of the effective plant. A common practical rule is that the inner-loop bandwidth should be several times higher than the outer-loop bandwidth, although the achievable ratio depends on sampling, resonance, noise, and actuator limits.

Analytical and Industrial Tuning Methods

Model-based tuning

A transfer-function model permits direct use of root locus, Bode plots, Nyquist plots, and pole-placement approximations. A typical frequency-domain process is:

1. Define performance requirements and actuator constraints.

2. Identify the relevant plant dynamics, including delay, resonance, and sensor filtering.

3. Select a controller structure.

4. Choose a target crossover frequency and phase margin.

5. Shape the loop using gain, lead, lag, or PID parameters.

6. Verify sensitivity, control effort, robustness, and nonlinear behavior through simulation.

For a desired phase margin, the crossover frequency is selected where the compensated plant has sufficient phase reserve. Gain is then adjusted so the magnitude is unity at that frequency. The design must also check gain margin and the behavior at frequencies beyond the nominal model range.

Ziegler–Nichols and related rules

Classical process rules use either a step response or an ultimate-gain experiment. In the ultimate-gain method, integral and derivative actions are disabled, proportional gain is increased until sustained oscillation occurs, and the ultimate gain and oscillation period are used to calculate P, PI, or PID settings. These rules are useful as initial estimates but can produce excessive overshoot and poor robustness, particularly in lightly damped mechanical systems. They should not be applied blindly to equipment that could be damaged by sustained oscillation.

Reaction-curve methods approximate a process by a first-order-plus-dead-time model and are useful for slow processes. Cohen–Coon and IMC-based rules provide alternatives. Internal Model Control tuning generally emphasizes a selected closed-loop time constant; increasing that time constant produces a slower but more robust response.

Relay and optimization-based tuning

Relay autotuning replaces the controller temporarily with a controlled switching element and infers critical frequency information from the resulting oscillation. It is useful when a reliable model is unavailable, but safety limits and operating-point dependence must be considered. Simulation-based tuning can use parameter sweeps, numerical optimization, or multiobjective criteria such as

\[

J=\int_0^\infty \left(q_e e^2(t)+q_u u^2(t)+q_{\dot u}\dot u^2(t)\right)dt.

\]

The weights express the relative importance of tracking error, actuator effort, and command smoothness. Optimization does not remove engineering judgment: an apparently optimal nominal controller may be fragile to uncertainty or unacceptable during saturation.

Saturation, Windup, Noise, and Actuator Limits

Real actuators have finite position, velocity, force, torque, voltage, current, and rate limits. Let the unsaturated controller output be \(u_c\), while the actual actuator command is

\[

u=sat(u_c).

\]

When saturation occurs, the plant no longer receives the control effort calculated by the controller. An integrator that continues accumulating error can become large, causing a long recovery delay and severe overshoot after the actuator leaves saturation. This is integral windup.

A widely used remedy is back-calculation. If \(u\) is the saturated output, the integrator state is corrected according to

\[

\dot{x}_I=K_i e+K_{aw}(u-u_c),

\]

where \(K_{aw}\) determines the unwinding speed. Conditional integration is another method: integration is suspended when the actuator is saturated and the error would drive it further into saturation. Clamping, tracking integrators, and reference governors are related approaches. Anti-windup design must be tested for both positive and negative saturation, rate limiting, mode changes, and sensor failure.

Noise places a practical limit on derivative action and bandwidth. Sensor filtering reduces noise but adds phase lag. A low-pass filter should therefore be selected together with the controller, not added after tuning without rechecking stability margins. Notch filters may suppress a known structural resonance, but excessive filtering can conceal instability or cause unacceptable delay.

Actuator limits should be included in simulation from the earliest design stage. A controller that performs well in a linear model may fail when friction, backlash, dead zones, quantization, compliance, or current limits are introduced. Bumpless transfer is necessary when switching between manual and automatic control or between operating modes. The internal controller state should be initialized so that the output does not jump at the moment of transfer.

A Practical Design Workflow

A robust workflow begins with the operating envelope rather than with a preferred controller formula. The designer first identifies reference ranges, disturbance magnitudes, actuator capabilities, sensor resolution, sampling period, safety constraints, and failure modes. The plant model is then developed from physical equations, experimental identification, or both.

The next step is to select the simplest structure that can satisfy the requirements. P or PI control is often sufficient for a nonresonant process. PID or lead compensation may be needed for damping. Feedforward should be added when an important disturbance is measurable. Cascade control is justified when an intermediate variable can be measured and its dynamics are substantially faster.

Tuning should proceed from inner loops to outer loops and from linear nominal analysis to nonlinear simulation. Step tests, disturbance tests, frequency sweeps, and Monte Carlo uncertainty studies reveal different weaknesses. Commissioning should use conservative gains, explicit output limits, alarm thresholds, and a rollback procedure. Each final setting should be documented with units, controller form, filter constants, sample time, saturation values, anti-windup parameters, and the operating conditions under which it was validated.

Illustrative Case Study: Position Control of an Industrial Rotary Axis

Background

Consider an illustrative rotary positioning axis driven by a permanent-magnet motor through a compliant transmission. The control system has a current-controlled drive, a velocity sensor, and an encoder measuring position. The axis must follow position commands while rejecting variable cutting torque. The motor current and velocity are limited, and the transmission has a resonance above the desired position-loop bandwidth.

Situation and problem

An initial proportional position controller gives acceptable motion at low speed but produces a steady position offset under cutting torque. Increasing proportional gain reduces the offset but excites the transmission resonance. A position PI controller removes the offset, yet large command steps cause current saturation and a delayed, overshooting recovery. Encoder quantization also produces visible high-frequency activity when derivative action is applied directly to the position error.

Analysis

The available velocity loop is faster than the position loop, making cascade control appropriate. The inner velocity loop can reject torque disturbances before they produce large position errors. The position loop should therefore be designed around the closed velocity-loop dynamics, not around the bare motor model. The resonance must remain outside the effective crossover region, with a notch filter considered only if necessary.

The steady offset under constant torque indicates inadequate low-frequency loop gain. However, integral action in the outer loop is not the only remedy: velocity feedforward based on the commanded trajectory can reduce tracking error during motion, while acceleration feedforward can reduce the torque demanded from feedback. Cutting torque remains a feedback-rejection problem because it is not perfectly predictable.

Approach and solution

The inner velocity controller is tuned first using a PI structure. Its bandwidth is selected well above the intended position-loop bandwidth, while current limits and rate limits are included in simulation. Anti-windup is enabled in the velocity integrator, and the velocity measurement receives a carefully selected low-pass filter.

The outer position loop begins with proportional control. Its gain is increased until the desired speed of response is reached without unacceptable resonance amplification. A small integral term is then introduced to remove residual static position error, with conditional integration and back-calculation activated during current saturation. The derivative term is omitted from the position error because encoder quantization would produce a noisy derivative signal. The measured velocity already supplies damping information through the cascade structure.

Velocity feedforward is added from the commanded position trajectory. A limited acceleration feedforward term is subsequently introduced and scaled experimentally. A reference prefilter smooths abrupt command changes without reducing the feedback loop's ability to reject cutting torque. Simulations include transmission compliance, quantization, saturation, friction, and a range of load inertias.

Outcome

The resulting cascade controller tracks commanded motion with substantially reduced following error, rejects constant cutting torque without persistent offset, and avoids the prolonged recovery associated with integral windup. The feedforward terms reduce feedback effort during predictable motion, while the feedback loops retain authority over unmodeled load changes. Performance is validated not only by nominal step response but also by disturbance tests, reversal tests, saturation tests, and uncertainty cases.

Learning points

The case illustrates that controller structure is often more important than parameter refinement. A fast inner loop, modest outer-loop integral action, feedforward based on known motion, and explicit anti-windup can outperform an aggressively tuned single-loop PID. It also demonstrates why derivative action should not be added automatically and why actuator limits must be part of the design model.

Common Design Errors and Professional Responsibilities

A frequent error is tuning only from a reference step. A system may track steps well but reject disturbances poorly, saturate during realistic commands, or amplify sensor noise. Another error is treating gain margin and phase margin as guarantees of acceptable performance; they are indicators of robustness, not substitutes for examining sensitivity and nonlinear behavior.

Other mistakes include ignoring computational delay, using a continuous-time design without accounting for sampling, placing filters after the controller has been tuned, copying tuning parameters between different PID forms, and increasing integral gain to compensate for an incorrectly calibrated sensor. Mechanical resonance, backlash, friction changes, and thermal variation should be treated as model uncertainty rather than as exceptional events.

Controller design also carries professional and ethical responsibilities. A poorly tuned industrial axis can damage equipment, produce unsafe motion, or compromise product quality. Commissioning must include independent safety functions, physical travel limits, emergency stopping, alarm management, access control, and documented change procedures. Cybersecurity is relevant when controllers are networked: unauthorized parameter changes can alter machine behavior even if the control algorithm itself is sound. Automatic tuning should be supervised, bounded, and reversible.

Digital Implementation and Emerging Practice

Modern industrial controllers implement classical algorithms digitally, often with discrete PID, parameter scheduling, diagnostics, and remote monitoring. Discretization must preserve the intended dynamics. The sampling frequency should be sufficiently higher than the desired closed-loop bandwidth, while computational and communication delays are included in the model. Derivative filtering, bumpless transfer, set-point weighting, and anti-windup are now standard features in many platforms.

Data-driven identification can estimate plant models from operating data, and digital twins can test controllers under varying loads and faults before deployment. Optimization and machine learning may assist gain scheduling, anomaly detection, or disturbance prediction. Nevertheless, learned components should not obscure the stability mechanism or bypass actuator and safety constraints. A classical feedback loop with interpretable limits often provides the dependable supervisory layer around more advanced predictive or adaptive functions.

Best-Practice Checklist

A professional design should satisfy the following principles:

• Define tracking, disturbance-rejection, robustness, and actuator requirements quantitatively.
• Use the simplest controller structure that meets those requirements.
• Tune cascade loops from the inside outward.
• Include saturation, rate limits, delays, filters, noise, and uncertainty in simulation.
• Use integral action with an explicit anti-windup strategy.
• Filter derivative action and avoid derivative kick where appropriate.
• Separate command shaping from disturbance rejection through feedforward or two-degree-of-freedom design.
• Validate with reference, disturbance, noise, reversal, and fault-oriented tests.
• Document controller form, units, limits, sample time, and commissioning conditions.
• Retain independent safety mechanisms and a controlled procedure for parameter changes.

Chapter Summary

Classical feedback controller design uses transfer functions, root locus, frequency response, and tuning rules to shape the behavior of closed-loop mechanical systems. Proportional action improves responsiveness, integral action removes persistent error, and derivative action adds damping but requires noise protection. PI and PID controllers are widely used, while lead, lag, and lead–lag compensators provide systematic control of phase, bandwidth, and steady-state accuracy.

Feedback should be complemented by feedforward when disturbances or commanded motion are measurable, and by cascade control when a fast intermediate variable can be controlled. Tuning may be model-based, experimental, rule-based, or optimization-based, but every method must be checked against uncertainty and nonlinear limits. Saturation creates integral windup, requiring anti-windup measures; noise and delay limit derivative action and achievable bandwidth. Effective industrial design combines analytical margins, nonlinear simulation, cautious commissioning, clear documentation, and independent safety provisions.

Key Terms

Sensitivity function: Transfer function describing how disturbances and plant uncertainty affect a closed-loop system.

Complementary sensitivity: Transfer function from reference to output, \(T=L/(1+L)\).

Proportional control: Control action proportional to instantaneous error.

Integral control: Control action proportional to the accumulated error.

Derivative control: Control action proportional to the rate of error change.

PID controller: Controller combining proportional, integral, and derivative actions.

Phase margin: Additional phase lag required at crossover to reach the onset of instability.

Gain margin: Multiplicative gain increase required to reach marginal stability.

Lead compensator: Network that adds positive phase and commonly increases response speed.

Lag compensator: Network that increases low-frequency gain with modest negative phase.

Lead–lag compensator: Combined network providing both phase advancement and low-frequency gain shaping.

Feedforward control: Control action based on a reference or measured disturbance before its effect appears.

Cascade control: Nested-loop architecture with a faster inner loop and slower outer loop.

Actuator saturation: Limitation that prevents the actuator from producing the demanded command.

Integral windup: Excessive accumulation of integral state during actuator saturation.

Anti-windup: Method for preventing or rapidly removing integrator accumulation during saturation.

Bandwidth: Frequency range over which a closed-loop system responds effectively.

Set-point weighting: Separate weighting of reference signals in proportional or derivative paths.

Bumpless transfer: Transition between control modes without an abrupt output change.

Robustness: Ability to maintain acceptable behavior despite uncertainty and disturbances.

Review Questions

Short Answer Questions

1. What are the principal effects of increasing proportional gain in a feedback loop?

2. Why can integral action eliminate constant steady-state error?

3. Distinguish between a lead compensator and a lag compensator.

4. What is integral windup, and under what condition does it occur?

5. Why is derivative action usually filtered in an industrial controller?

Long Answer Questions

1. Derive the sensitivity and complementary sensitivity functions and explain their significance for disturbance rejection and noise transmission.

2. Compare P, PI, PD, and PID controllers in terms of steady-state error, transient response, robustness, and practical limitations.

3. Explain a frequency-domain procedure for designing a lead compensator to meet a specified phase-margin requirement.

4. Describe the design and tuning sequence for a cascade position–velocity control system.

5. Discuss analytical, rule-based, experimental, and optimization-based controller tuning, including their assumptions and limitations.

Analytical / Application Questions

1. A mechanical plant has satisfactory transient response with proportional control but exhibits a constant position offset under load. Propose a controller structure and explain how you would prevent saturation-induced windup.

2. A measured disturbance enters a machine before a slow process, and the disturbance sensor is reliable. Explain how feedforward and feedback should be combined and what model limitations must be considered.

3. A PID-controlled actuator becomes noisy after derivative action is increased. Develop a diagnostic and redesign procedure that considers filtering, sensor location, bandwidth, sampling delay, and actuator limits.

22. Multiple Choice Questions with Answers

Question 1

Which controller property most directly eliminates steady-state error caused by a constant disturbance?

A. Increasing derivative filtering

B. Integral action

C. Reducing phase margin

D. Adding measurement noise

Correct Answer: B. Integral action

Question 2

For a unity-feedback system, which expression represents the sensitivity function?

A. L(s)

B. 1+L(s)

C. 1/[1+L(s)]

D. L(s)/[1-L(s)]

Correct Answer: C. 1/[1+L(s)]

Question 3

A lead compensator is primarily introduced to:

A. Add positive phase and improve transient robustness

B. Eliminate actuator saturation

C. Remove all measurement noise

D. Reduce low-frequency loop gain

Correct Answer: A. Add positive phase and improve transient robustness

Question 4

What is the principal cause of integral windup?

A. A sensor with infinite bandwidth

B. Actuator saturation while integration continues

C. A lead zero below its pole

D. Excessive phase margin

Correct Answer: B. Actuator saturation while integration continues

Question 5

In a properly arranged cascade system, which loop is normally tuned first?

A. The outer loop

B. The reference prefilter

C. The inner loop

D. The feedforward loop only

Correct Answer: C. The inner loop

Question 6

Why is feedforward unable to replace feedback completely in most applications?

A. It cannot act on a measurable disturbance

B. It cannot correct unmeasured disturbances and model errors

C. It always reduces low-frequency gain

D. It necessarily causes instability

Correct Answer: B. It cannot correct unmeasured disturbances and model errors

Question 7

A lag compensator is generally selected when the main requirement is to:

A. Increase low-frequency accuracy with limited bandwidth change

B. Maximize high-frequency noise amplification

C. Create derivative kick

D. Force the actuator to saturate

Correct Answer: A. Increase low-frequency accuracy with limited bandwidth change

Question 8

Which measure is most appropriate for preventing a reference step from causing derivative kick?

A. Apply derivative action only to the reference

B. Apply derivative action to the measured output or use set-point weighting

C. Remove all proportional action

D. Increase integral gain without limits

Correct Answer: B. Apply derivative action to the measured output or use set-point weighting

Question 9

In the illustrative rotary-axis case, why was a fast inner velocity loop useful?

A. It eliminated the need for position measurement

B. It rejected torque disturbances and provided damping before the outer loop acted

C. It guaranteed no actuator saturation

D. It removed transmission compliance physically

Correct Answer: B. It rejected torque disturbances and provided damping before the outer loop acted

Question 10

Which design practice is most reliable for a controller intended for industrial deployment?

A. Tune only from an ideal reference step

B. Ignore rate limits until commissioning

C. Validate reference, disturbance, noise, saturation, and uncertainty cases

D. Copy parameters without checking the PID form

Correct Answer: C. Validate reference, disturbance, noise, saturation, and uncertainty cases