System feedback is the process of routing a portion of a circuit's output signal back to its input to automatically regulate performance, stabilize gain, or reduce distortion. If you have ever built an amplifier that screamed with oscillation or a power supply that drifted under load, you have felt the absence of properly designed feedback. In practical electronics, we rarely use components in their raw, 'open-loop' state because manufacturing tolerances and temperature shifts make their baseline behavior entirely unpredictable. By feeding a sample of the output back to the input, we force the circuit to correct its own errors in real-time.
Think of a water tank with a float valve: as the water level (output) rises, the float pushes a lever that throttles the inlet pipe (input), preventing an overflow. In electronics, we use resistors, capacitors, and optocouplers instead of floats and levers, but the physics of error-correction remain identical.
Open-Loop vs. Closed-Loop: How Feedback Changes the Circuit
To understand what system feedback actually changes in a real installation, we need to look at the raw numbers. When a component operates without feedback, it is in an open-loop state. When we apply feedback, it enters a closed-loop state. Negative feedback—the most common type used in linear circuits—trades raw, unusable gain for precision, bandwidth, and stability.
Below is a spec-sheet comparison using the ubiquitous TL072 JFET-input operational amplifier. This table demonstrates exactly what negative feedback does to the silicon's native characteristics.
| Parameter | Open-Loop (No Feedback) | Closed-Loop (Negative Feedback) | What This Means on the Bench |
|---|---|---|---|
| Voltage Gain | 200,000 V/V (106 dB) | 10 V/V (20 dB) | Raw gain is too high to control; feedback dials it to a precise, usable number. |
| Bandwidth (-3dB) | ~15 Hz | ~300 kHz | Feedback dramatically widens the usable frequency range (Gain-Bandwidth Product = 3MHz). |
| Output Impedance | ~200 Ω | < 1 Ω | The circuit can now drive heavier loads without the output voltage sagging. |
| Total Harmonic Distortion (THD) | High (Clips immediately) | < 0.01% | Feedback actively cancels out non-linearities introduced by the internal transistors. |
| Sensitivity to Temperature | Extreme (Drifts mV/°C) | Negligible | The feedback loop automatically compensates for thermal drift in the silicon. |
Worked Numeric Example: Taming an Op-Amp
Let's calculate a real circuit to see how the math dictates the physical component choices. We will build a non-inverting amplifier using an LM358 (a standard dual op-amp) to amplify a 0.5V DC sensor signal up to a level readable by a 5V Arduino ADC.
The formula for the closed-loop gain ($A_{cl}$) of a non-inverting amplifier is:
$A_{cl} = 1 + (R_f / R_{in})$
We need a gain of 10 to turn 0.5V into 5.0V. Therefore:
- $10 = 1 + (R_f / R_{in})$
- $9 = R_f / R_{in}$
If we select a standard 1 kΩ resistor for $R_{in}$, we need a 9 kΩ resistor for $R_f$. Since 9 kΩ isn't a standard E24 value, we use a 10 kΩ potentiometer dialed to exactly 9 kΩ, or combine a 8.2 kΩ and 820 Ω resistor in series.
What happens without feedback?
The LM358 has an open-loop gain of roughly 100,000 V/V. If you applied 0.5V directly to the non-inverting pin without the feedback network, the op-amp would attempt to output $0.5V imes 100,000 = 50,000V$. Since it is powered by a 5V rail, it would instantly slam into the positive supply rail and saturate, giving you a useless, clipped 5V DC block regardless of minor sensor changes.
What happens with feedback?
The 9 kΩ and 1 kΩ resistor divider routes exactly 10% of the output voltage back to the inverting input. If the output tries to creep above 5.0V, the voltage at the inverting pin rises above the non-inverting pin. The op-amp's internal differential pair detects this error and instantly throttles the output back down. The system reaches equilibrium at exactly 5.0V.
Where You Meet System Feedback in Practice
Feedback isn't just for textbook op-amp circuits; it is the backbone of modern power and control systems.
Switch-Mode Power Supplies (SMPS)
In a flyback converter (like your laptop charger), the primary side switches high-voltage DC at 65 kHz. But how does the controller know when to stop switching? A precision shunt regulator (usually a TL431) monitors the low-voltage DC output. When the voltage hits the target, the TL431 turns on an optocoupler (like the PC817), which sends a feedback signal across the isolation barrier to the PWM controller on the primary side, telling it to skip a switching cycle.
Audio Amplifiers (Class D and Linear)
Class D amplifiers use pulse-width modulation (PWM) to drive speakers efficiently. However, the LC output filter and the speaker's own impedance curve introduce phase shifts and distortion. Modern Class D amps (like those using the TPA3116D2 chip) employ a post-filter feedback loop. They sample the actual voltage at the speaker terminals and feed it back to the modulator, correcting dead-time distortion and flattening the frequency response regardless of the speaker's impedance.
3D Printer Hotends (PID Control)
If you have tuned a 3D printer hotend, you have tuned a feedback loop. The Marlin firmware uses a Proportional-Integral-Derivative (PID) algorithm. The thermistor provides the feedback signal. If the temperature undershoots the 200°C setpoint, the integral term accumulates the error over time and increases the PWM duty cycle to the heater cartridge until equilibrium is reached.
Common Confusions: What Feedback Is (and Isn't)
When diagnosing circuits or reading schematics, hobbyists frequently mix up feedback with other concepts. Here is how to keep them straight.
Confusion 1: Feedback vs. DC Biasing
Biasing sets the static, DC operating point (quiescent point) of a transistor so it stays in its linear region. It is a one-way setup. Feedback is a dynamic, continuous loop that corrects AC and DC errors as they happen. A voltage divider on the base of a BJT is biasing; a resistor connecting the collector back to the emitter is feedback.
Confusion 2: Assuming All Feedback is Negative
Negative feedback opposes changes, creating stability. Positive feedback reinforces changes, driving the circuit to an extreme state. We intentionally use positive feedback to build oscillators (like a 555 timer in astable mode) and Schmitt triggers. In a Schmitt trigger, positive feedback creates 'hysteresis'—two different threshold voltages that prevent a noisy input signal from causing the output to chatter rapidly.
Confusion 3: Feedforward vs. Feedback
Feedforward anticipates an error before it happens by measuring the input or a disturbance, whereas feedback reacts only after the output has already deviated. Advanced motor controllers use both: feedforward calculates the baseline voltage needed for a target speed, while the feedback loop trims the voltage based on the actual back-EMF measured from the motor.
Frequently Asked Questions
Why does my op-amp circuit oscillate when I add long feedback wires?
Long wires introduce parasitic capacitance and inductance. In a high-speed feedback loop, this parasitic reactance introduces a phase shift. If the phase shift reaches 180 degrees at a frequency where the loop gain is still greater than 1, your negative feedback turns into positive feedback, and the circuit becomes an oscillator. Keep feedback traces as short and direct as physically possible, and use a small compensation capacitor (e.g., 10pF - 100pF) in parallel with the feedback resistor to roll off high-frequency gain.
Can feedback fix a poorly designed power supply?
No. Feedback can only correct errors within the bandwidth and power limits of the system. If your power transformer is undersized, your bulk capacitors are too small, or your switching MOSFETs are overheating, the feedback loop will simply hit its duty-cycle limit trying to compensate. The output will still sag or ripple. Feedback regulates; it does not create missing power capacity.
How does optocoupler degradation affect SMPS feedback?
Optocouplers suffer from Current Transfer Ratio (CTR) degradation over years of operation, especially at high temperatures. As the CTR drops, the feedback loop has to push more current through the optocoupler's LED to achieve the same control effect on the primary side. Eventually, the PWM controller maxes out its compensation range, leading to poor transient response, audible transformer whine, or complete loss of regulation. This is why high-reliability power supplies specify optocouplers with high initial CTR and long operational lifespans.
