A non-inverting op-amp is an operational amplifier configuration where the input signal is applied to the positive (non-inverting) terminal, producing an output that is scaled up but maintains the exact same phase and polarity as the input. In a real circuit, this topology changes a fragile, high-impedance signal into a robust, low-impedance output capable of driving an ADC or downstream stage without flipping the signal's polarity. Beginners commonly confuse it with the inverting op-amp (which flips the phase 180 degrees and suffers from low input impedance) or a simple voltage follower (which is just a non-inverting amplifier locked to a gain of exactly 1).
The Core Math: Gain Formula and Input Impedance
The closed-loop voltage gain ($A_v$) of a non-inverting op-amp is determined entirely by the external feedback network. The formula is:
$A_v = 1 + \frac{R_f}{R_i}$
Where $R_f$ is the feedback resistor (connecting the output to the inverting input) and $R_i$ is the ground resistor (connecting the inverting input to ground). The most common bench mistake is forgetting the 1 + at the beginning of the equation; the minimum possible gain for this configuration is 1, never 0.
Think of it like a rigid mechanical lever where the fulcrum is fixed at the ground, the input lifts one end, and the output is taken from a point further up the lever—it moves in the exact same direction, just amplified by the ratio of the lever arms.
| Parameter | Non-Inverting Op-Amp | Inverting Op-Amp |
|---|---|---|
| Phase Shift | 0° (In-phase) | 180° (Inverted) |
| Input Impedance | Extremely High (MΩ to TΩ) | Low (Equal to input resistor $R_{in}$) |
| Minimum Gain | 1 (Unity) | 0 (Theoretical) |
| Virtual Ground | No (Inputs track $V_{in}$) | Yes (Inverting input sits at 0V) |
Because the input signal is fed directly into the op-amp's high-impedance positive terminal, the circuit draws virtually zero current from the source. This makes it the undisputed champion for buffering sensors, as documented in Texas Instruments' op-amp design guides.
Worked Numeric Example: Designing a Sensor Amplifier
Let's design a front-end for a shunt current sensor. The shunt outputs a maximum of 150 mV at peak load. We need to feed this into a 3.3V microcontroller ADC, and we want the peak signal to hit exactly 1.65 V (mid-scale for maximum bipolar headroom).
- Calculate Required Gain: $A_v = \frac{V_{out}}{V_{in}} = \frac{1.65V}{0.150V} = 11$.
- Select Resistor Ratio: Using $A_v = 1 + \frac{R_f}{R_i}$, we need $11 = 1 + \frac{R_f}{R_i}$, which means $\frac{R_f}{R_i} = 10$.
- Choose Standard Values: We select $R_i = 10\text{ k}\Omega$ and $R_f = 100\text{ k}\Omega$. These are standard E24 values and keep the impedance high enough to avoid loading the op-amp's output, but low enough to minimize thermal (Johnson-Nyquist) noise.
If you use standard 5% carbon film resistors, your 100kΩ resistor could be 105kΩ and your 10kΩ could be 9.5kΩ. Your actual gain becomes $1 + (105/9.5) = 12.05$. Your 150mV signal will now hit 1.8V, ruining your ADC scaling. Always use 1% or 0.1% metal film resistors for the feedback network in precision DC applications.
Where You Meet This in Practice
You will rarely see an ideal, textbook op-amp on a PCB. Instead, you will encounter specific integrated circuits chosen for their non-ideal characteristics. Here is where the non-inverting topology shows up in the wild, and which ICs to grab from your parts bin:
- DC Sensor Buffering (LM358 / LM324): When amplifying slow-moving DC signals from thermocouples or current shunts, the LM358 is the industry workhorse. It runs on a single supply and its input common-mode range includes the negative rail (ground), meaning it can accurately amplify a 10mV signal sitting right at 0V.
- Audio Pre-Amplifiers (TL072 / NE5532): In audio mixing consoles and guitar pedals, the non-inverting amp is used for gain staging. The TL072 offers incredibly low noise (18 nV/√Hz) and high slew rate, but it requires a dual-polarity supply (e.g., ±12V) because its inputs cannot tolerate voltages near the negative rail.
- High-Speed Data Acquisition (THS3091 / OPA657): When buffering signals for a 100+ MSPS ADC, you need a current-feedback or ultra-high-bandwidth voltage-feedback op-amp. The non-inverting configuration is preferred here because its high input impedance prevents the op-amp from loading down the preceding 50Ω transmission line.
Real-World Gotchas: Common-Mode Range and Output Swing
The math says your output will be exactly $V_{in} \times A_v$. The silicon disagrees. When debugging a non-inverting op-amp circuit that 'isn't working', check these two datasheet parameters immediately:
1. Input Common-Mode Voltage Range ($V_{ICR}$)
This defines the voltage range that both inputs can safely sit within while the op-amp remains functional. If you power an LM358 with a single 5V supply, its $V_{ICR}$ is typically 0V to 3.5V (which is $V_{CC} - 1.5V$). If your non-inverting input receives a 4.2V signal from a sensor, the op-amp will experience phase reversal—the output will suddenly slam to the negative rail instead of amplifying the signal. If you need to measure signals near the positive rail on a single supply, you must buy a 'Rail-to-Rail Input' (RRI) op-amp like the MCP6002.
2. Output Voltage Swing ($V_{OH}$ / $V_{OL}$)
Even 'rail-to-rail output' (RRO) op-amps cannot perfectly touch the supply rails. Under a 10kΩ load, an RRO op-amp on a 5V supply might only swing from 0.05V to 4.85V. If your math demands a 4.95V output, the waveform will clip flat at the top. Always design your gain to leave at least 200mV of headroom from your supply rails, or use a charge pump to generate a slightly higher supply voltage for the op-amp.
Frequently Asked Questions
Can a non-inverting op-amp have a gain of less than 1?
No. The mathematical minimum gain for a standard non-inverting configuration is exactly 1 (unity gain), achieved when $R_f = 0\Omega$ and $R_i = \infty$ (open circuit). If you need to attenuate a signal (e.g., scale a 10V battery down to 3.3V for a microcontroller) while maintaining high input impedance, you must place a resistive voltage divider before the non-inverting input of a unity-gain buffer. Do not try to put the voltage divider inside the feedback loop to achieve fractional gain; the math does not support it and the circuit will become unstable.
Why does my non-inverting op-amp output clip at the positive rail?
Premature positive clipping is almost always caused by violating the Input Common-Mode Range ($V_{ICR}$) or exceeding the Output Voltage Swing limits. First, measure the voltage at the positive input pin with your multimeter. If it is higher than $V_{CC} - 1.5V$ on a standard op-amp like the LM358, the internal input differential pair has starved for current, and the op-amp has lost control of the feedback loop. Second, check your load resistance; if you are driving a 100Ω load, the op-amp's internal output transistors will drop significant voltage, severely limiting the maximum positive swing.
What is the difference between a non-inverting op-amp and a voltage follower?
A voltage follower (or unity-gain buffer) is simply a specific, simplified edge-case of the non-inverting op-amp. In a voltage follower, the output is tied directly to the inverting input ($R_f = 0$), and there is no resistor to ground ($R_i = \infty$). This forces the gain equation to $1 + (0 / \infty) = 1$. While a standard non-inverting amp scales the voltage up, a follower provides zero voltage gain; its sole purpose is impedance transformation—converting a high-impedance source into a low-impedance driver without altering the signal amplitude. For a deeper look into buffer stability, consult the All About Circuits semiconductor textbook.
Do I need a resistor on the non-inverting input for bias current compensation?
In older bipolar op-amps (like the LM741 or NE5532), input bias currents are relatively high and mismatched. Textbooks often recommend adding a 'compensation resistor' ($R_{comp} = R_f || R_i$) in series with the non-inverting input to balance the voltage drops caused by these bias currents. However, for modern CMOS or JFET-input op-amps (like the TL072, OPA2134, or MCP6002), the input bias current is in the picoamp range. Adding a large compensation resistor in series with the input will actually increase thermal noise and make the circuit more susceptible to EMI. Only use bias compensation if the datasheet explicitly recommends it for bipolar input stages.
