Operational amplifier positive feedback occurs when a portion of the output signal is routed back to the non-inverting (+) input, reinforcing the input difference and driving the op-amp rapidly into saturation. Instead of stabilizing a circuit into a linear amplifier—which is the role of negative feedback—positive feedback fundamentally changes the circuit by turning the op-amp into a non-linear switching device, introducing hysteresis to reject noise. Beginners often confuse this concept with acoustic 'howl' in PA systems or assume all feedback must be negative for a circuit to function, but in bench electronics, deliberate positive feedback is a critical design tool for creating clean, bounce-free signal transitions.
The Mechanics of Driving an Op-Amp to the Rails
To understand why this happens, you have to look at the open-loop gain ($A_{OL}$) of a typical op-amp. A standard TL072 has an open-loop gain of around 100,000 (100 dB). In a linear circuit, we use negative feedback to tame this massive gain down to a usable, predictable number like 10 or 100.
When you route a fraction of the output back to the non-inverting input, you create a runaway condition. If the voltage at the non-inverting pin is even a microvolt higher than the inverting pin, the output begins to swing positive. Because that positive output is fed back to the non-inverting pin, the differential voltage increases, which drives the output further positive, which feeds back even more. This is a compounding loop. Think of it like a snowball rolling down a steep, snowy hill: it gathers mass and speed until it violently hits the bottom. In an op-amp, the 'bottom' is the positive or negative saturation rail, limited only by the internal transistor voltage drops and your power supply.
This rapid transition is exactly what we want when building comparators. However, a bare comparator without hysteresis will chatter wildly if the input signal has noise near the threshold voltage. Positive feedback solves this by shifting the threshold voltage depending on the current output state, creating a 'dead band' or hysteresis window where noise cannot trigger a false switch.
Worked Numeric Example: Designing a Schmitt Trigger
Let's design a non-inverting Schmitt trigger using a standard TL072 op-amp to clean up a noisy 5V sensor signal. We will power the TL072 with a dual supply of +12V and -12V.
First, we must account for the op-amp's output saturation limits. A TL072 cannot swing perfectly to the supply rails; under typical loads, it saturates about 1V shy of the rails. Therefore, our output high ($V_{sat+}$) will be +11V, and our output low ($V_{sat-}$) will be -11V.
We set up a voltage divider between the output and ground, connected to the non-inverting (+) input:
- Feedback Resistor ($R_f$): 100 kΩ (from output to non-inverting input)
- Input Resistor ($R_i$): 10 kΩ (from non-inverting input to ground)
- Signal Input: Applied directly to the inverting (-) input
Now, we calculate the Upper Threshold Point (UTP) and Lower Threshold Point (LTP). The threshold is simply the voltage at the non-inverting pin, determined by the output state and the resistor divider ratio: $R_i / (R_f + R_i)$.
When the output is HIGH (+11V):
The threshold voltage (UTP) is $11V imes 0.0909 = +1.0V$. The input signal must rise above +1.0V to force the output to switch low.
When the output is LOW (-11V):
The threshold voltage (LTP) is $-11V imes 0.0909 = -1.0V$. Once the output has switched low, the input signal must drop all the way down to -1.0V to force the output to switch high again.
The total hysteresis band is the difference between the UTP and LTP: $1.0V - (-1.0V) = 2.0V$. Any noise on the input signal with a peak-to-peak amplitude of less than 2.0V will be completely ignored, resulting in a perfectly clean square wave at the output. For a deeper look at the math behind these thresholds, reference the Electronics Tutorials guide on op-amp comparators.
Where You Meet This in Practice
You will rarely see intentional positive feedback in audio amplifiers or linear sensor conditioning, but it is everywhere in control systems and power electronics.
| Application | Why Positive Feedback is Used | Typical Components |
|---|---|---|
| Thermostats & HVAC Controls | Prevents the compressor relay from rapidly clicking on and off (chattering) when the room temperature hovers exactly at the setpoint. | LM393 comparator, NTC thermistor |
| Battery Management Systems (BMS) | Creates a Low-Voltage Disconnect (LVD) with a recovery gap. A 12V LiFePO4 pack might cut off at 10.0V, but positive feedback ensures it won't reconnect until the charger pushes it back to 11.5V. | TL431 reference, LM358 op-amp |
| Switch Debouncing | Mechanical contacts bounce for milliseconds when pressed. A Schmitt trigger with a wide hysteresis band translates that messy analog bounce into a single, clean digital logic edge. | 74HC14 (internal Schmitt), or discrete op-amps |
| Relaxation Oscillators | Combining positive feedback (for the switching thresholds) with negative feedback via an RC network (for timing) creates a simple, cheap square-wave clock generator. | LM741, passive RC network |
When designing these circuits on a bench, always remember that standard op-amps are not optimized for rapid saturation recovery. If your application requires high-speed switching (above 100 kHz), you should use a dedicated comparator IC like the LM393 or LM311, which are specifically architected to handle the charge storage issues that plague standard op-amps when driven to the rails. The Texas Instruments comparator portfolio provides excellent application notes on the architectural differences between the two.
Frequently Asked Questions
Why does operational amplifier positive feedback cause saturation?
It causes saturation because it creates a regenerative loop. The op-amp's massive open-loop gain amplifies the initial voltage difference between the inputs. By feeding a portion of that amplified output back to the non-inverting input, you increase the input difference, which causes more amplification, which feeds back even more. This exponential growth continues in nanoseconds until the internal output transistors run out of voltage headroom and hit the power supply rails (saturation).
How do I calculate hysteresis with operational amplifier positive feedback?
Hysteresis is calculated by finding the difference between the Upper Threshold Point (UTP) and Lower Threshold Point (LTP). For a basic non-inverting Schmitt trigger where the input signal goes to the inverting pin and the feedback network goes to the non-inverting pin, the formula is: $V_{hysteresis} = (V_{sat+} - V_{sat-}) imes [R_i / (R_f + R_i)]$. You must use the actual saturated output voltages of your specific op-amp, not the raw power supply rails, to get accurate bench results.
Can I use positive and negative feedback on the same op-amp?
Yes, and this is actually how many oscillator circuits work. In a Wien bridge oscillator or a relaxation oscillator, negative feedback (often through an RC network or a thermistor) sets the overall gain and timing characteristics, while positive feedback (through a resistive divider) sets the switching thresholds and ensures the circuit breaks out of its linear region to oscillate. The dominant feedback path at any given frequency or DC state determines the circuit's behavior.
What is the difference between an op-amp and a comparator when using positive feedback?
While both can be wired with positive feedback to create a Schmitt trigger, their internal architectures differ drastically. Standard op-amps (like the TL072 or LM741) have internal frequency compensation capacitors to prevent oscillation in linear negative-feedback circuits. When you drive an op-amp into saturation via positive feedback, that capacitor must charge and discharge, causing a 'recovery delay' that limits switching speed. Dedicated comparators (like the LM393) omit this compensation capacitor for high-speed switching and typically feature open-collector or open-drain outputs, requiring a pull-up resistor but allowing you to interface directly with different logic voltage levels.
