Electronics Design AU
Op-AmpsSensors

What Is an Instrumentation Amplifier and How Does It Work?

Last updated 27 June 2026 · 7 min read

Direct Answer

An instrumentation amplifier (in-amp) is a precision differential amplifier built from three op-amps: two unity-gain input buffers that present high impedance to both signal sources, followed by a difference amplifier output stage. Gain is set by a single external resistor Rg: G = 1 + 2R/Rg, where R is the matched internal resistor pair. The symmetrical structure gives very high common-mode rejection ratio (CMRR) — typically 80–100 dB — meaning a signal present on both inputs (such as mains interference) is cancelled while a small differential signal is accurately amplified. In-amps are the standard front-end for Wheatstone bridge sensors, thermocouples, load cells, current shunts, ECG electrodes, and any application where a small differential voltage sits on top of a large common-mode voltage.

Detailed Explanation

An instrumentation amplifier solves a fundamental problem in precision measurement: amplifying a small signal that is superimposed on a much larger common-mode voltage, without that common-mode voltage causing errors in the result.

A classic example is a Wheatstone bridge connected to a load cell. The bridge may produce only 1–10 mV of differential signal, but both sides of the bridge sit at roughly Vsupply/2 — a large common-mode voltage that a simple single-ended amplifier cannot handle. An in-amp amplifies only the difference (the 1–10 mV bridge output) while rejecting the common-mode (the Vsupply/2 sitting on both inputs).

Three-Op-Amp Topology

The classic in-amp topology uses three op-amps:

Stage 1 (two input buffers): Each input connects to the non-inverting terminal of its own op-amp. Both op-amps share a feedback network connected through the external gain resistor Rg. This shared feedback means the two buffers' outputs carry the differential signal amplified by 1 + 2R/Rg, while any common-mode signal passes through at unity gain. Because both inputs see only the op-amp's non-inverting terminal, input impedance is extremely high (>10 MΩ for BJT types, >1 GΩ for CMOS types).

Stage 2 (difference amplifier): The outputs of the two buffers feed a standard difference amplifier with a gain of 1 (matched resistors). This stage rejects the common-mode component that passed through the input stage. Its output equals the differential signal amplified by G = 1 + 2R/Rg.

The gain-setting resistor Rg appears between the inverting inputs of the two input op-amps. Decreasing Rg increases differential gain. Removing Rg entirely (open circuit) typically gives unity gain, though some INA devices require a minimum Rg.

Gain Formula and Resistor Selection

For the standard three-op-amp in-amp: G = 1 + 2R/Rg

For the INA128/INA129 (R = 25 kΩ): G = 1 + 50,000/Rg (Ω)

Desired GainRg (INA128)Notes
1Open circuitUnity gain; no gain resistor needed
250.0 kΩE96 value
512.4 kΩE96: 12.4 kΩ → G = 5.03
105.49 kΩ
100499 ΩStandard E96 value
100050.0 Ω

Use 1% (or better) tolerance resistors for Rg when accuracy matters. The internal R resistors are typically laser-trimmed to 0.01% match on precision INA ICs, so the external Rg tolerance dominates the gain accuracy.

Common-Mode Rejection Ratio (CMRR)

CMRR is the ratio of differential gain to common-mode gain, expressed in dB. A CMRR of 100 dB means that a 1 V common-mode input signal produces only 10 µV of error at the output (referred to the input, RTI). High CMRR is critical in:

  • Thermocouple amplifiers (the thermocouple sits in an industrial environment with significant electrical interference on both wires)
  • ECG and biomedical circuits (mains interference couples to the patient's body as common mode)
  • Current shunt monitoring on a high-side switch (both shunt terminals sit at the supply voltage — 12 V, 24 V, or 48 V — as common mode)

CMRR degrades with frequency (the symmetry of the input stage is less perfect at high frequencies) and with lower gain settings. At G = 1, the in-amp is essentially a difference amplifier and CMRR depends mainly on resistor matching in Stage 2.

Monolithic INA Devices

Most designs use a monolithic IC rather than three discrete op-amps. The internal resistors are precisely laser-trimmed, giving 80–100 dB CMRR and gain accuracy better than ±0.25% at standard gains.

Common INA ICs:

PartSupplyGain rangeInput noiseFeatures
INA128/INA129±2.25–18 V1–10,0008 nV/√HzGeneral purpose, audio
INA3331.8–5.5 V1–100050 nV/√HzLow power, single-supply, micropower
INA8261.35–36 V1–10007 nV/√HzLow power, rail-to-rail output
AD8221±2.3–18 V1–10007 nV/√HzVery low offset, precision
INA2193–5.5 VFixedI2C current/power monitor (specialised)

Practical Applications

Wheatstone bridge signal conditioning: A bridge sensor (strain gauge, pressure sensor, load cell) produces a differential output of a few mV/V of excitation, sitting on a common-mode of Vsupply/2. An in-amp at gain 100–500 amplifies the differential output to a useful range for an ADC. The voltage divider article covers the Thevenin model of a resistive bridge.

High-side current sensing: A shunt resistor in series with the load (high side) develops a differential voltage proportional to current. Both shunt terminals are at or near the supply rail. The in-amp amplifies the shunt voltage while rejecting the common-mode supply voltage. Dedicated current monitor ICs (INA219, INA226) use this principle.

Thermocouple amplification: Thermocouples generate ~8–50 µV/°C — a signal that needs significant gain and strong common-mode rejection to be usable. Dedicated thermocouple INA ICs (MAX31855, AD8495) include cold-junction compensation in addition to the differential amplifier.

Biomedical signals: ECG requires amplifying signals in the 0.05–100 Hz range at 0.1–5 mV amplitude. Mains interference at 50/60 Hz couples as common-mode to both electrodes and must be rejected. CMRR > 80 dB across 50 Hz is a standard biomedical requirement.

For the complete path from in-amp gain stage through active filtering to ADC conversion, see sensor signal conditioning basics.

For precision sensor front-end design — from signal chain architecture through PCB layout for low noise — Zeus Design's electronics engineering team can support development from concept to production.

Design Considerations

  • REF pin (reference pin): Most monolithic INAs have a REF pin that shifts the output by a fixed voltage. For single-supply operation, connect REF to Vsupply/2 (via a resistor divider or voltage reference) so the amplifier output sits at mid-supply and can swing in both directions with a bipolar input signal.
  • Input protection: In-amps have limited ESD and overvoltage protection. If the inputs can go beyond the supply rails (common in industrial or automotive environments), add series resistors (100 Ω–1 kΩ) and clamping diodes (Schottky or TVS) to protect the input stage.
  • Bandwidth vs gain trade-off: Like op-amps, in-amps have a gain-bandwidth product. The INA128's bandwidth is 1.3 MHz at G = 1, falling to 200 kHz at G = 10, 28 kHz at G = 100. Design the signal chain bandwidth around the required measurement bandwidth, not the data-sheet maximum.
  • Input bias current and source impedance: Even high-impedance in-amps have non-zero input bias current (INA128: 5 nA typical). For high-impedance sources (e.g. pH electrodes, piezo sensors), this current flowing through the source impedance creates an offset voltage. Minimise source impedance or use a CMOS in-amp with picoamp bias currents.

Common Mistakes

  • Forgetting the REF pin on single-supply designs: Leaving REF tied to ground on a single-supply in-amp limits the output swing to one polarity. For a bipolar differential signal (bridge output that can be positive or negative), the output will clip on the negative swing. Always connect REF to a mid-supply reference when operating from a single supply.
  • Applying gain before filtering for high-frequency noise: High gain amplifies all noise within the in-amp's bandwidth, including any high-frequency interference at the inputs. Add a differential input filter (two matched series resistors and a capacitor from each resistor junction to ground) to roll off high-frequency common-mode and differential noise before it reaches the input stage.
  • Mismatching the gain resistor footprint: Gain accuracy depends on Rg. An 0603 resistor with 5% tolerance gives ±5% gain error. Use 1% 0402 or 0603 resistors for Rg to get ±1% gain accuracy — or use the INA's internal gain settings if the device offers them.

Frequently Asked Questions

What is the difference between an op-amp and an instrumentation amplifier?
A standard op-amp is a single differential input stage that requires an external resistor network (inverting and non-inverting configurations) to set gain. The resistor matching in these networks limits CMRR to 60–80 dB in practice. An instrumentation amplifier uses two buffered input stages that present very high, equal impedance to both inputs, followed by a difference amplifier — the buffering and symmetry give 80–100 dB CMRR without requiring matched external resistors. An op-amp is a general-purpose building block; an in-amp is a finished differential gain stage optimised for precision measurement.
How do you calculate the gain of an instrumentation amplifier?
For the classic three-op-amp topology (and most monolithic INA ICs that follow it): G = 1 + 2R/Rg, where R is the matched internal resistor and Rg is the external gain-setting resistor. For the INA128, R = 25 kΩ on each side, giving G = 1 + 50,000/Rg. For unity gain, leave Rg open or very large. For G = 100, Rg = 50,000/99 ≈ 505 Ω. The gain formula is printed in every INA datasheet along with a table of standard Rg values for common gains. For gains below 10, the in-amp's CMRR typically degrades — use gain ≥ 10 for most applications.
When should you use an instrumentation amplifier instead of a difference amplifier?
Use an instrumentation amplifier when the signal source impedance is high or unequal between the two inputs. A simple difference amplifier (one op-amp with four resistors) has input impedance equal to its input resistors — typically 1–10 kΩ — which loads the source and degrades CMRR if source impedances are unequal. An in-amp's input impedance is typically >1 GΩ (CMOS) or >10 MΩ (BJT-input), so it doesn't load the source. For low-impedance, well-matched sources (like a 4-wire bridge where the input to the amplifier is a known low impedance), a difference amplifier is simpler and cheaper.

References

Related Questions

Related Forum Discussions