Instrumentation Amplifier Basics: Measure a 20 mV Sensor Signal Without the Noise

Author: Abloom Instruction Date: November 20, 2026

A strain gauge on a bridge, a thermocouple in a furnace, an ECG electrode on a chest — all of them give you a tiny difference between two wires, sitting on top of a much larger voltage that both wires share. A plain differential amplifier (covered in our earlier article) can subtract the two wires, but it loads the sensor with only 10 kΩ and its rejection of the shared hum depends on four resistors matching perfectly. An instrumentation amplifier fixes both problems: it puts a voltage follower in front of each input, so the sensor sees almost no load, and it sets its gain with a single resistor. This article works out a complete example: a 20 mV sensor difference becomes a clean 2.02 V output.

Schematic of a three-op-amp instrumentation amplifier with two preamp op-amps A1 and A2, a 200 ohm gain resistor Rg between them, 10 kilo-ohm feedback resistors, and a subtractor op-amp A3 giving 2.02 volts out, beside a plot of output voltage against input difference for three gains

Figure 1: A three-op-amp instrumentation amplifier (a) — the two preamps A1 and A2 amplify the difference, and the subtractor A3 removes what is common. The plot (b) shows the output against the input difference for three values of Rg. (Schematic, not to scale.)

How it works — two preamps and a subtractor

The circuit has three op-amps. A1 and A2 are non-inverting amplifiers whose inverting inputs are joined by a single resistor, Rg. The outputs of A1 and A2 go into a differential amplifier, A3, built from four equal resistors, so A3 has a gain of 1 and simply subtracts. The total gain is the product of the two stages:

G = (1 + 2R / Rg) × (RF / R1)

With RF = R1 the subtractor gain is 1, so the gain is set by Rg alone: G = 1 + 2R / Rg. Change one resistor and the gain changes — no resistor matching to worry about. And because A1 and A2 are non-inverting, their inputs draw only the op-amp's tiny bias current, so the input resistance is in the gigaohm range for a FET-input op-amp. A sensor with a high output resistance is hardly loaded at all, unlike the 10 kΩ the plain differential amplifier presents.

Worked example — 20 mV in, 2.02 V out

Take R = 10 kΩ and Rg = 200 Ω, so the gain is:

G = 1 + 2 × 10,000 / 200 = 1 + 100 = 101

Now feed a sensor signal: V1 = 5.00 V and V2 = 5.02 V, so the difference is only 20 mV on top of a 5 V common level. First the current through Rg: the two inverting inputs sit at V1 and V2, so the voltage across Rg is V2 − V1 = 0.02 V, and:

I = 0.02 / 200 = 0.1 mA

That same 0.1 mA flows through each 10 kΩ feedback resistor (the op-amp inputs draw no current), dropping 0.1 mA × 10 kΩ = 1.0 V across each one. A1's output is one drop below its input, A2's output one drop above its input:

A1 output = 5.00 − 1.0 = 4.00 V    A2 output = 5.02 + 1.0 = 6.02 V

The subtractor takes the difference: 6.02 − 4.00 = 2.02 V. Exactly G × (V2 − V1) = 101 × 0.02 = 2.02 V, as the formula promised. The 5 V the two wires share has vanished from the output.

The common mode — 5 V in, almost nothing out

Now feed the same voltage to both inputs: V1 = V2 = 5 V. No voltage appears across Rg, so no current flows through it and neither feedback resistor drops any voltage. Both preamps simply pass 5.00 V through, and the subtractor gives:

Vout = 5.00 − 5.00 = 0 V

A real amplifier is not perfect, and its common-mode rejection ratio (CMRR) says how much of the shared signal leaks through. A typical instrumentation amplifier IC offers 100 dB or more of CMRR, which means a common-mode signal is divided by 100,000: a 5 V common-mode input leaves only 5 / 100,000 = 50 µV at the output. Compare that with the plain differential amplifier from our earlier article, where 1% resistors alone limit the CMRR to about 34 dB — the instrumentation amplifier is roughly 2,000 times better at ignoring what both wires share, which is why it is the standard input stage for weighing scales, strain-gauge bridges and ECG monitors.

Designing for a gain of 1000

Keep R = 10 kΩ and solve for Rg at a gain of 1000:

Rg = 2R / (G − 1) = 20,000 / 999 = 20.02 Ω

Use the standard 20 Ω resistor, giving G = 1 + 20,000 / 20 = 1001. A 5 mV sensor signal then gives 1001 × 0.005 = 5.005 V out. Watch the supply limit: with ±12 V supplies the output can swing to about ±10 V, so at a gain of 1001 the input difference must stay within 10 / 1001 ≈ 10 mV or the output clips. For a smaller gain, Rg = 1 kΩ gives G = 1 + 20,000 / 1,000 = 21, so the same 20 mV input gives only 21 × 0.02 = 0.42 V. One resistor, any gain from 1 to 1000 and beyond — that is the whole point of the circuit.

Key points to remember

  • An instrumentation amplifier is two non-inverting preamps joined by one gain resistor Rg, followed by a subtractor: G = (1 + 2R / Rg) × (RF / R1).
  • Worked example: R = 10 kΩ, Rg = 200 Ω → G = 101; V1 = 5.00 V, V2 = 5.02 V → 0.1 mA through Rg, preamp outputs 4.00 V and 6.02 V, Vout = 2.02 V.
  • Common mode cancels: V1 = V2 = 5 V gives 0 V out ideally; a typical IC with 100 dB CMRR leaves only 50 µV from a 5 V common-mode signal.
  • The inputs see a gigaohm-scale input resistance, so a high-resistance sensor is hardly loaded — unlike the 10 kΩ a plain differential amplifier presents.
  • Design rule: for a gain of 1000 use Rg = 20 Ω (G = 1001); 5 mV in gives 5.005 V out, and the input difference must stay within about 10 mV on ±12 V supplies.