the instrumentation amplifier
/ in-amp /
The instrumentation amplifier is the precision tool for reading a tiny difference that rides on top of a big shared signal — like picking out a whisper of a few millivolts from a strain-gauge bridge while ignoring volts of interference that hit both wires equally. It is a difference amplifier built to extremely high standards: enormous input impedance so it never loads the sensor, and a powerful ability to reject whatever the two inputs have in common.
The classic design uses three op-amps. Two of them buffer the inputs directly, presenting near-infinite input impedance so the source is never loaded, and a single gain-setting resistor Rg between them sets the gain (typically gain = 1 + 2R/Rg) without ever touching the precisely matched resistors of the third, difference-taking stage. Because the gain is set by one resistor and the matching is done at the factory by laser trimming, the common-mode rejection stays superb. Example: a load cell outputs 2 mV of difference sitting on a 2.5 V common-mode level; an in-amp with a gain of 500 turns the 2 mV into a clean 1 V output while the 2.5 V common-mode is rejected.
Why this matters: in-amps are the front end of nearly every precision sensor system — strain gauges and load cells, thermocouples, ECG and other biopotentials, and 4-20 mA current loops, especially when the sensor sits far away on a long, noisy cable. The honest advice: buy an integrated, laser-trimmed in-amp (a single chip with one external gain resistor) rather than building a difference amplifier from loose op-amps and resistors, because a hand-built version's rejection is limited by how well you can match its resistors.
A Wheatstone bridge weighing scale outputs only 10 mV at full load, sitting on a 2.5 V common-mode. An in-amp set to a gain of 100 (one external resistor) lifts that to 1 V for the ADC while rejecting the common-mode and the mains hum picked up on the cable.
One gain resistor, factory-matched difference stage — the high-CMRR sensor front end.
An in-amp still needs a path for its input bias current to flow to ground. Float both inputs (for example through a transformer or an isolated sensor) with no bias return and the output drifts to a rail — always provide bias-current return resistors.