LIBRARYBiopotential Analog Front-Ends.
The amplifier stage that makes microvolt EEG measurable: what an instrumentation amplifier does, why biopotentials demand one, and what a modern integrated AFE folds in.
A biopotential analog front-end (AFE) is the circuitry between the electrodes and the analog-to-digital converter. Its job sounds simple, make a microvolt signal big and clean enough to digitize, but biopotentials make it hard: an EEG is tens of microvolts riding on a much larger, ever-present common-mode interference, picked up by electrodes with high and unequal contact impedance. The AFE is where you win or lose the whole recording.
The instrumentation amplifier
At its heart is the instrumentation amplifier, a differential amplifier built for exactly this: very high input impedance (so it doesn't load the high-impedance electrode), high gain for the tiny differential signal, and very high common-mode rejection to throw away the interference that's common to both inputs. High input impedance matters more than people expect: if the amplifier draws appreciable current, the electrode impedance turns that into an error, and impedance mismatch between electrodes turns rejected common-mode back into apparent signal.
| AFE property | Why a biopotential needs it |
|---|---|
| High input impedance | Electrodes are high-impedance; the amp must not load them or mismatch becomes error |
| High CMRR (~100–110 dB in good EEG amps) | Mains hum is common-mode and 100s–1000s× the signal |
| Low input-referred noise | The signal itself is µV-scale; amp noise must be sub-µV |
| Programmable gain | Scale µV inputs to the ADC's full range without clipping |
| A bias/reference path | Center the inputs and host the driven-bias (RLD) loop |
Discrete vs integrated front-ends
Historically you built this from discrete parts: an instrumentation-amp IC per channel, gain-setting resistors, filtering, a separate ADC, and a driven-bias circuit, and then you fought noise, channel-to-channel matching, and board complexity by hand. The modern path is an integrated AFE that folds the whole chain (per-channel low-noise PGA, a 24-bit ADC per channel, reference, and the bias-drive amplifier) into one chip, such as the TI ADS1299. For EEG, the bias loop and lead-off detection being on-chip is most of the value.
▸Deep dive· What an integrated AFE actually saves you
A discrete front-end means owning four hard problems at once, on every channel: a sub-microvolt noise budget, common-mode rejection that survives real electrode mismatch, channel-to-channel gain and offset matching, and a stable bias loop, all on a board you also have to keep electrically quiet. An integrated AFE collapses those into one characterized part with guaranteed datasheet specs: you trade hands-on control for numbers you can trust. For most projects that's the right call; building it discrete once is still the best way to understand what the chip is doing for you.
Checkpoint
Quick check
References
Keep going
Designing a real biopotential front-end, the instrumentation, gain, the bias loop, and getting it quiet, is the build in the OTD Academy EEG front-end project.
One Thousand Drones Academy · reviewed June 2026
Coming soon
8-Channel EEG Front-End on ESP32 →Design the analog board that reads real brainwaves: the BCI.
Want the whole EEG & BCI cluster as one book?