ECG AFE: ADS1291, MAX30003, AD8232
Compare ECG AFE chips ADS1291, MAX30003, and AD8232 on RLD, power, noise, and leads-off detection to pick the right front-end.

Contents & prerequisites
ECG AFE Architecture Fundamentals
An ECG analog front-end (AFE) takes a microvolt-level biopotential sitting on top of volts of common-mode noise and turns it into a clean digital waveform, and the three chips compared here — the Texas Instruments ADS1291, the Maxim (now Analog Devices) MAX30003, and the Analog Devices AD8232 — solve that problem with very different amounts of integration. Understanding why each architecture looks the way it does requires walking through the signal chain and the noise budget before comparing part numbers.
Signal chain overview: instrumentation amplifier → programmable gain → filter → ADC
Every ECG front-end follows the same functional order: an instrumentation amplifier (IA) pulls the differential signal off two or three electrodes while rejecting the common-mode voltage, a programmable-gain amplifier (PGA) boosts the millivolt-scale IA output into the ADC's input range, an anti-alias filter removes energy above the Nyquist frequency, and an analog-to-digital converter (ADC) — typically a delta-sigma converter for biopotential work — digitizes the result. The instrumentation amplifier itself is usually built from op-amp stages; the same differential-input, common-source output topology covered in Two-Stage Op-Amp: Differential Input + Common-Source Output underlies the front-end stage in all three chips, even though none of the datasheets expose that internal schematic. Where the three parts diverge is how much of this chain is on-die versus left to the board designer.
Common-mode rejection requirements for biopotential signals
Common-mode rejection ratio (CMMR) of at least 80 dB at 50/60 Hz is the practical floor for ECG work, because mains interference coupling into the body routinely exceeds the ECG signal by 60 dB or more. The electrode-skin interface is not symmetric — contact impedance mismatch between the two sensing electrodes converts common-mode noise into a differential-mode error that no amount of downstream CMRR can remove. This is why right-leg drive (RLD) circuits exist: they actively drive the body to a reference potential, lowering the effective common-mode voltage before it ever reaches the IA. Modeling that electrode impedance interacting with input bias networks is a direct application of the Voltage Divider and Current Divider Rules, since the skin-electrode impedance and the AFE's input impedance form exactly that kind of divider.
Input-referred noise specifications and SNR budgeting
Input-referred noise for a usable ECG front-end should sit below 1 µVpp to 10 µVpp in the diagnostic bandwidth (0.05 Hz to 150 Hz), since the P-wave and ST-segment features of interest are themselves only tens to hundreds of microvolts. Check each manufacturer's datasheet for the noise figure measured at the specific gain and bandwidth setting you intend to use — noise scales with gain and sample rate in delta-sigma converters, so a number quoted at one configuration doesn't transfer to another. Budgeting the full chain — electrode, IA, PGA, ADC quantization noise — against a target SNR is the same exercise described in more depth in Analog Front-End Power Budget for Wearable, where power and noise trade against each other directly.
Integration level: monolithic vs. discrete component trade-offs
Monolithic integration (ADC and PGA on the same die as the IA) cuts board area and simplifies layout, but it locks the designer into the vendor's filter corners, gain steps, and interface choice. A discrete approach — IA only, external ADC — costs more board space and more BOM lines but lets the designer pick the digitization resolution, sample rate, and interface independently. The ADS1291 and MAX30003 sit at the integrated end; the AD8232 sits at the discrete end.
ADS1291: TI's Single-Channel Ultra-Low-Power Approach
The ADS1291 is a fully integrated single-channel (expandable via ADS1291/1292/1298 family members to multi-channel) biopotential front-end built for portable diagnostic-grade ECG rather than continuous wearable use.
Architecture: integrated PGA, delta-sigma converter, on-chip leads-off detection
The signal path includes a low-noise IA, a programmable-gain stage, and a 24-bit delta-sigma ADC, all behind an SPI register map — check the ADS1291 datasheet for the exact gain steps and oversampling ratios available.
Power consumption profile and battery life implications
The ADS1291 trades some power headroom for resolution and flexibility; its quiescent current is higher than the MAX30003's but still low enough for battery-powered handheld recorders. Confirm the exact current-per-mode figures (high-resolution vs. low-power mode) in the datasheet before sizing a battery, since the gap between modes is large enough to change a product's runtime by multiples.
Sampling rates, resolution, and data output interface (SPI)
Output resolution is 24 bits with multiple selectable data rates, read out over SPI, which gives a host MCU full access to raw samples for custom filtering or arrhythmia algorithms rather than a pre-digested heart-rate number.
Lead detection circuit and electrode impedance measurement
The ADS1291 includes on-chip leads-off detection using either DC current injection or AC subthreshold excitation, letting firmware flag a disconnected or high-impedance electrode without external comparator circuitry.
Typical application circuit and component selection
A typical application circuit needs an AC-coupled or DC-coupled electrode interface, a reference electrode drive output, and decoupling per the datasheet's layout guidance — component count is modest because the PGA and ADC are internal, but external protection diodes and defibrillation-recovery components are still required for any design contacting skin.
MAX30003: Maxim's Wearable-Focused Single-Channel Design
The MAX30003 was built from the ground up for continuous, battery-sipping wearable monitoring rather than clinical bench use.
Integrated right-leg drive circuit for noise cancellation
The MAX30003 includes an on-chip RLD amplifier, so the common-mode-rejecting drive electrode doesn't need an external op-amp stage — this is the single biggest footprint and power saving relative to a discrete RLD implementation, and it directly answers why integrated RLD matters: it suppresses mains and motion-induced common-mode noise before the IA, at the cost of locking the designer into Maxim's RLD gain and bandwidth.
ECG-specific features: QRS detection, heart-rate output
Beyond raw waveform capture, the MAX30003 includes on-chip QRS complex detection and heart-rate output registers, offloading beat-detection math from the host MCU — useful in a wearable where the MCU is asleep most of the time.
Ultra-low power consumption (<1 mW) for continuous monitoring
Datasheet-rated operating current keeps total power under roughly 1 mW in continuous ECG-monitoring mode (verify the exact figure and supply voltage condition in the MAX30003 datasheet), which is the key enabler for multi-day battery life in a smartwatch form factor.
Small footprint and I2C interface for wearable integration
A small WLP/QFN package footprint and SPI/I2C-style digital interface (confirm exact interface in datasheet) suit tightly packed wearable PCBs where board area, not raw channel count, is the constraint.
Built-in test signal generation for self-diagnostics
An internal test signal generator lets firmware verify the full analog chain at power-up without external stimulus — valuable in a sealed wearable where there's no access to inject a test ECG signal post-assembly.
AD8232: Analog Devices' Flexible, Minimal-Integration Solution
The AD8232 is an instrumentation-amplifier-centric front-end that deliberately stops short of digitization, aimed at designs that want full control downstream.
Single Op-Amp-based instrumentation amplifier with external gain setting
Gain is set with external resistors around the AD8232's internal op-amp stages, following the same instrumentation-amplifier principles as Two-Stage Op-Amp: Differential Input + Common-Source Output, giving the designer continuous gain control rather than fixed PGA steps.
No integrated ADC: signal conditioning only, leaving digitization open
Because there's no on-chip ADC, the AD8232 outputs a conditioned analog waveform that the designer digitizes with any external ADC — a 10-bit MCU-internal ADC for a coarse heart-rate monitor, or a high-resolution delta-sigma part for diagnostic work. Modeling how the AD8232's output stage loads into that external ADC's input impedance is a two-port problem, the kind formalized by Two-Port Network Parameters: Z, Y, H, ABCD.
Lower power than integrated solutions; higher component count
Supply current for the AD8232 core itself is low (check the datasheet for the exact microamp figure at the specified supply), but the total system power and component count depend heavily on whatever external ADC and filter stages are added — so a fair comparison against the ADS1291 or MAX30003 has to include that external circuitry, not just the AD8232 alone.
Suitable for designs requiring analog output buffering
Because the output is a buffered analog signal, the AD8232 fits designs that need to feed an oscilloscope, a legacy analog recorder, or a mixed-signal board where the ADC is shared with other sensors.
Lead-off detection via AC excitation and envelope detection methods
Lead-off detection is implemented externally using AC excitation injected onto the electrodes and envelope detection of the resulting amplitude — this requires extra passive components and possibly a comparator, compared to the single-register check available on the ADS1291 or MAX30003.
Comparative Selection Matrix
| Dimension | ADS1291 | MAX30003 | AD8232 |
|---|---|---|---|
| Integrated ADC | Yes, 24-bit Σ-Δ | Yes | No |
| Integrated RLD | External op-amp typical | Yes, on-chip | External |
| Leads-off detection | On-chip | On-chip | External AC/envelope |
| Interface | SPI | SPI/I2C (verify datasheet) | Analog output |
| Power class | Low-power, diagnostic-grade | Ultra-low, <1 mW class | Low, excludes external ADC |
| Component count | Moderate | Lowest | Highest (external ADC + lead-off) |
| Design flexibility | Moderate | Low | High |
Power budget: wearable vs. clinical-grade monitoring
A wearable optimizing for multi-day battery life should budget sub-milliwatt continuous current, which points toward the MAX30003; a handheld or bedside diagnostic recorder with a larger battery or mains power can afford the ADS1291's higher but still modest draw in exchange for resolution and multi-channel scaling.
Integration density vs. design flexibility trade-off
Higher integration (MAX30003, ADS1291) means less board area and faster bring-up but fixed internal topology; lower integration (AD8232) means more board area and bring-up time but full control over gain, filtering, and digitization choices — relevant if a later product revision needs a different ADC resolution without a full respin.
Cost per channel and bill-of-materials impact
Per-channel silicon cost is lowest for single-purpose wearable parts produced at high volume, but total BOM cost for the AD8232 path must include the external ADC, voltage reference, and lead-off components, which can close or reverse the apparent price gap — get quotes for the full signal chain, not just the AFE die.
Feature set: RLD, leads-off detection, self-test capabilities
Only the MAX30003 bundles RLD, leads-off, QRS detection, and self-test all on-chip; the ADS1291 covers leads-off and PGA/ADC integration but typically needs an external RLD amplifier; the AD8232 provides none of these on-chip, pushing all three into the discrete design.
Data interface requirements (SPI, I2C, analog output)
SPI (ADS1291) suits hosts already running other SPI peripherals at moderate clock rates; I2C-style (MAX30003, confirm exact interface in datasheet) suits low-pin-count wearable MCUs; analog output (AD8232) suits boards where the ADC is a shared system resource or where legacy analog instrumentation is required.
Practical Design Considerations
PCB layout for low-noise differential routing
Route the differential electrode pair as a tightly coupled pair away from switching regulators and digital clock lines, and keep the RLD drive trace short — any of the three AFEs can be defeated by a noisy layout no matter how good their on-chip CMRR spec is.
Power supply decoupling and noise isolation strategies
Separate analog and digital supply domains with ferrite beads or dedicated LDOs, and decouple each supply pin per the datasheet's recommended capacitor values; digital switching noise coupling into the analog supply is a common cause of 50/60 Hz artifacts that look like a CMRR failure but are actually a power-integrity problem.
Electrode interface impedance matching and AC bias networks
Bias resistor networks that set the DC operating point of the electrodes interact with electrode-skin impedance exactly as described by the Voltage Divider and Current Divider Rules; undersized bias resistors load the signal, oversized ones slow settling after motion artifact.
Clock and reference voltage stability for ADC accuracy
A jittery ADC clock or an unstable voltage reference degrades effective resolution independent of the front-end's analog noise floor — use the vendor's recommended crystal or internal oscillator configuration and a low-drift reference where the chip doesn't provide one internally.
Firmware integration: data acquisition and heart-rate algorithms
Firmware scope differs sharply: the AD8232 requires the host to implement filtering, QRS detection, and lead-off logic entirely in software; the ADS1291 offloads digitization and leads-off but leaves beat detection to the host; the MAX30003 offloads QRS detection and heart-rate output as well, reducing host MCU duty cycle.
When to Choose Each AFE Chip
ADS1291: multi-channel portable ECG recorders, diagnostic devices
Choose the ADS1291 (or its multi-channel siblings) when the product needs diagnostic-grade resolution, multiple simultaneous leads, and SPI access to raw samples — portable Holter monitors and bedside recorders are the natural fit.
MAX30003: wearable smartwatches, fitness trackers, continuous monitoring
Choose the MAX30003 when battery life and board area dominate the design brief and a single-lead heart-rate or rhythm trace is sufficient — smartwatches and chest-strap fitness monitors are the natural fit.
AD8232: research platforms, custom analog signal conditioning, cost-sensitive designs
Choose the AD8232 when the project needs a fully custom digitization chain, shares an ADC across multiple sensors, or is a research platform where algorithm flexibility matters more than time-to-market.
System-level trade-offs: performance, power, size, cost, time-to-market
No single chip wins on every axis: the ADS1291 wins on resolution and channel scalability, the MAX30003 wins on power and footprint, and the AD8232 wins on flexibility and lowest per-die cost at the expense of total system component count — the right choice follows from which of those constraints the product spec treats as fixed.
Learning
Sign in to track your progress.
Evidence
Public projects engineers linked to ECG AFE: ADS1291, MAX30003, AD8232.
No engineer has linked a project to this topic yet. Built something that proves it? Add the project and tag it with biomedical-med-ecg-afe-ads1291-max30003-ad8232 — it then shows here and on your public profile.
