Feedback and Control Systems

Feedback is the act of measuring what a circuit actually does, comparing it to what it should do, and feeding the difference back to drive the output toward the target. This single idea, formalized for electronics by Harold Black at Bell Telephone Laboratories in 1927, underlies automatic regulation across the discipline. A feedback system measures its output, subtracts a reference, and acts on the resulting error so that the output holds steady despite disturbances, load changes, supply variation, component tolerance, and aging. From a bimetallic thermostat to a precision servo to a switching regulator's voltage loop, the same loop of measurement, comparison, and correction recurs.

This category treats feedback as a control discipline: the theory that explains why a closed loop is more accurate and more linear than the parts it is built from, the analysis that decides whether that loop is stable or will break into oscillation, and the circuits that turn the theory into working regulators. The central trade is that feedback exchanges raw open-loop gain for predictability, and the central hazard is that the same loop, with a half-turn of extra phase shift, becomes a positive-feedback oscillator. Holding the benefit while avoiding the hazard is what stability margins and compensation are for. A companion page, Feedback and Stability, surveys much of this ground as a single article; the subcategories below develop the control-systems view in greater depth.

The four subcategories move from principle to edge case to practice. The first establishes the theory of negative feedback, where the loop is built to oppose change and so desensitize, linearize, and broaden a circuit. The second turns the loop the other way, examining positive feedback, where reinforcing the output deliberately produces thresholds, memory, and regeneration. The third confronts the question every loop must answer, stability, and the compensation that secures it. The fourth assembles these ideas into automatic control circuits that regulate real quantities such as speed, temperature, light, and position. The discussion that follows draws out the principles they share.

Articles in This Category

Negative Feedback Theory - Return a portion of the output in opposition to the input, and an imperfect amplifier becomes a precise one. This subcategory develops the feedback equation, in which the closed-loop gain A/(1 + Aβ) approaches 1/β once the loop gain Aβ is large, so that performance is set by a passive feedback network rather than by the active device. Coverage spans the four feedback topologies, series-shunt, shunt-series, series-series, and shunt-shunt, and how each shapes input and output impedance to build voltage, current, transconductance, or transresistance amplifiers; the desensitization of gain to device variation by the factor 1 + Aβ; and the reduction of distortion and the extension of bandwidth at constant gain-bandwidth product. It also frames the stability question that negative feedback inevitably raises, since too much phase lag turns opposition into reinforcement.
Positive Feedback Applications - Feed the output back in phase rather than against it, and the loop reinforces change instead of resisting it, a regenerative action that is a fault in a linear amplifier but the whole point of many circuits. This subcategory covers the deliberate use of positive feedback: the Schmitt trigger, invented by Otto Schmitt in 1934, whose dual thresholds give hysteresis that rejects noise and squares up slow or dirty signals; cross-coupled latches that store a bit because the loop holds whichever state it last reached; regenerative and dynamic comparators that snap to a decision in picoseconds; negative-resistance and tunnel-diode circuits; and Q-multiplication and super-regenerative receivers that operate a stage at the very edge of oscillation. The recurring theme is controlled instability, harnessed for switching, memory, and gain.
Control System Modeling and Block Diagrams - Model electrical, mechanical, and electromechanical systems as transfer functions, then combine them with block-diagram algebra, signal-flow graphs, and Mason's gain formula, up to cascaded current, speed, and position loops.
Bode and Nyquist Plots - Read and draw frequency-response plots: asymptotic Bode construction, gain and phase crossover, Nyquist plots and the Nyquist stability criterion, conditionally stable loops, and loop-gain measurement by injection.
Root Locus Analysis - Trace how closed-loop poles move as gain changes: the angle and magnitude conditions, the construction rules, reading damping from the locus, and lead, lag, PI, and PD compensator design on the locus.
Stability Analysis and Compensation - Determine whether a loop will settle or oscillate, and reshape it when it will not. The Barkhausen condition for sustained oscillation, loop gain of unity magnitude with 180 degrees of phase shift, defines the boundary every feedback design must avoid, and phase margin and gain margin measure the distance from it; practical designs target roughly 45 to 60 degrees of phase margin and 10 dB or more of gain margin. This subcategory presents the analysis tools, Bode plots, the Nyquist criterion, and root locus, and the compensation that buys margin: dominant-pole and lag compensation that roll gain off early, lead compensation that adds phase at crossover, and the pole-splitting Miller compensation that gives an internally compensated op-amp its single-pole response. It closes with nested-loop bandwidth separation, parasitic poles, and the link between phase margin and transient overshoot.
Time-Domain Response of Control Systems - Specify and measure how a control loop responds in time, from the rise time, overshoot, and settling time of the second-order prototype to dominant poles, steady-state error, system type, and the error constants.
State-Space Analysis and Control - Describe systems by state equations and design them in the time domain: controllability and observability, pole placement by state feedback, observers and the separation principle, integral action, LQR, and discrete-time models.
Digital Control Systems - Close control loops in software: sampled-data loops and their delays, discretization by Euler, Tustin, matched pole-zero, and zero-order-hold methods, digital PID with anti-windup, z-plane stability, quantization and limit cycles, and implementation.
Automatic Control Circuits - Assemble feedback into closed-loop regulators that hold a physical quantity at its setpoint. This subcategory is the applied counterpart to the theory: the block diagram of reference, summing junction, controller, plant, sensor, and feedback path, and above all the PID controller, whose proportional term sets responsiveness, integral term drives steady-state error to zero, and derivative term anticipates change to tame overshoot, together with the anti-windup and derivative filtering that practical analog implementations require. Coverage includes representative applications, DC motor speed control with inner current and outer speed loops, temperature control with its long lags and dead time, light-level and automatic-gain regulation, servo and multi-axis position control with feedforward, and adaptive schemes such as gain scheduling, alongside the tuning, safety, and fail-safe practices that field-worthy controllers demand.

Themes Across Feedback and Control Systems

The four subcategories treat feedback from different angles, yet a handful of ideas run through all of them.

Loop gain is the master variable. Nearly every benefit of negative feedback, desensitization, distortion reduction, impedance control, and disturbance rejection, scales with the factor 1 + Aβ, and the closed-loop gain settles to 1/β only where that loop gain is large. The same loop gain governs stability, because the system oscillates precisely where its magnitude is one and its phase reaches 180 degrees. Reading a design as a loop gain versus frequency, rather than as a collection of stages, is the habit that unifies the field.

The line between negative and positive feedback is a half-turn of phase. A loop built to oppose change becomes a loop that reinforces it once accumulated phase lag reaches 180 degrees at a frequency where gain remains. This is why an amplifier with too many poles oscillates and why a Schmitt trigger or comparator is simply that same instability put to work. Negative-feedback design spends its effort keeping the loop short of that boundary; positive-feedback design deliberately crosses it.

Stability is bought with margin, and margin costs bandwidth. Adequate phase and gain margins are what separate a smooth, well-damped response from one that rings or oscillates, and they are rarely free. Compensation almost always trades closed-loop bandwidth, and therefore speed, for the phase margin that guarantees a clean response. The art lies in buying just enough margin for worst-case tolerances and conditions without slowing the system more than the application allows.

Real loops are limited by what the ideal model omits. Parasitic capacitance and inductance, package and layout effects, sensor lag, actuator saturation, and finite slew rate add poles, delay, and nonlinearity that the textbook loop ignores. A design that is stable on paper can ring or saturate on the bench, so worst-case analysis, large-signal behavior such as integral windup, and measurement of the actual loop gain are part of every serious feedback design.

Nesting and hierarchy organize complex control. High-performance regulation rarely uses one loop. Cascaded structures, an inner current loop inside a velocity loop inside a position loop, let each loop be fast, simple, and independently tuned, provided their bandwidths are separated by roughly a factor of five to ten. Feedforward then supplies the predictable part of the command so that feedback need correct only the unpredictable remainder, a division of labor that recurs from motor drives to switching regulators.

Conclusion

Feedback and control systems turn the single idea of acting on measured error into the foundation of analog regulation. Negative feedback theory explains why a closed loop is more accurate and linear than its parts, positive feedback applications show how the same mechanism reversed yields thresholds, memory, and regeneration, stability analysis and compensation secure the loop against the oscillation that always threatens it, and automatic control circuits assemble these ideas into regulators of speed, temperature, light, and position. Across all four, loop gain is the variable to watch, margin is the price of stability, and the gap between the ideal loop and the real one is where designs succeed or fail. The subcategories above develop each in detail, and the related topics place feedback within the wider practice of analog design.