Home Blog Blog Details

Wearable Health Electronics: Tiny Boards, Big Vibes

September 01 2025
Ersa

Inquiry

Global electronic component supplier ERSAELECTRONICS: Rich inventory for one-stop shopping. Inquire easily, and receive fast, customized solutions and quotes.

QUICK RFQ
ADD TO RFQ LIST
Wearable health electronics are the **friendly neighborhood sidekicks** of modern medicine

Overview

Wearable health electronics are the **friendly neighborhood sidekicks** of modern medicine—more Spider-Man than Iron Man: agile, close to the body, and surprisingly clever for their size. A watch that spots a rhythm hiccup, a ring that learns your recovery curve, a patch that whispers about hydration—none of this works without a quiet orchestra of **sensors, low-noise AFEs, accurate ADCs, a power tree that minds its manners**, and code that knows when to speak up.

A great wearable lives by three rules. First, **measure honestly**—tiny signals, big discipline. Second, **spend power like a miser**—this is the way, Mando. Third, **be transparent**—explain uncertainty, sync with the phone, and log what matters. Do that, and your product feels more Wednesday Addams (calm, unflappable) than drama-queen gadgetry.

Wearable health electronics architecture—sensors (PPG, ECG, temp, IMU), AFE/ADC, MCU, PMIC/battery, BLE, app/cloud
From skin to signal to story: the wearable stack in one picture.

User Stories & Requirements

Real wrists and fingers don’t read spec sheets. They sweat (hello, Dune stillsuit fantasy), swing wildly on runs, and live under LED billboards. Translate that into engineering:

  • Accuracy vs. battery: Publish numbers that hold on Tuesdays. Duty-cycle wisely; log confidence, not bravado.
  • Comfort & fit: Mechanical preload for PPG and ECG, but avoid pressure marks. Textures matter more than marketing.
  • Latency & UX: Trending can be slow; alerts may not. Precompute on-device, summarize on phone.
  • Privacy: Sensitive data by default. Local first, opt-in cloud, clear consent text a human can read.

System Architecture

Think in rings: **Sensing** (PPG/ECG/temp/IMU) → **Readout** (AFE/ADC) → **Compute** (MCU/SoC) → **Connectivity** (BLE/NFC/Wi-Fi) → **Power** (PMIC/battery/fuel gauge) → **Safety & logging**. The board is tiny, so discipline beats brute force.

Wearable block diagram—sensor front ends, digitization, memory/MCU, radio and power domains
Quiet analog lanes, busy digital streets, and a radio that minds the neighbors.

Sensing Layer: PPG, ECG, Bioimpedance & Temperature

PPG (photoplethysmography) is your Loki variant detector for blood volume changes. Green + IR LEDs, a low-leakage photodiode, and an AFE with picoamp bias currents keep the baseline civilized. Use **multi-wavelength** if you care about SpO₂ or skin tone robustness. Modulate LEDs away from display PWM and power-rail ripple; park dark frames for ambient subtraction.

ECG is the grown-up in the room. Dry electrodes, high CMRR instrumentation amps, right-leg drive (or virtual ground), and **lead-off detection** make the difference between clean PQRST and “modern art.” Use shielded routes and guard rings around the inputs; keep the radio and haptics socially distant during sampling windows.

Bioimpedance (hydration, respiration proxy, BIA): inject a tiny AC current and measure the voltage vector. Choose a frequency plan that dodges your own clocks; true RMS or I/Q demodulation turns raw samples into meaningful magnitude/phase.

Skin & core-adjacent temperature: single or dual sensors (skin + internal reference). Thermal isolation from PMIC heat plus a tiny window in the enclosure avoids “your watch thinks you’re a toaster.”

PPG LEDs/photodiode, ECG electrodes, bioimpedance inject/sense, temp sensor placement and shielding
Choose the diplomats your wearer’s skin will actually respect.

Motion & Context: IMU, Altimeter & Fusion

An **IMU (accelerometer + gyro)** is the show’s narrator—capturing cadence, posture, sleep micro-moves, and “you bumped the watch while washing dishes.” Couple it with a **barometric altimeter** for stairs and outdoor context. Use sensor fusion to separate true physiology (heart rate changes) from motion artifacts. Spidey-sense is cool; **artifact rejection** is cooler.

IMU and altimeter aiding PPG/ECG artifact rejection and activity classification
Move smarter: fuse motion with physiology for fewer false alarms.

Low-Noise AFE/ADC & Signal Integrity

Wearables thrive on **microvolts**. Use chopper-stabilized IAs for ECG and PPG transimpedance for photodiodes. A 16–24-bit ΔΣ ADC with stable clocks keeps gain honest; timestamp channels so data lands where filters expect it. Ratiometric references, guard traces, and **matched thermals** rescue you from drift the datasheet didn’t warn about.

  • Clock plan: Keep LED drive, ADC, BLE and haptics from forming a boy band. Stagger edges; align conversions.
  • Leakage control: Clean board, sensible solder mask, and keep the coffee off your photodiode node.
  • Self-test hooks: MUX in references; measure offsets/gain at boot; version coefficients with CRC.
PPG TIA and ECG IA into delta-sigma ADC with ratiometric reference and synchronized sampling
Quiet analog is a choice. Make it early; enjoy it forever.

Algorithms & On-Device ML

The math is the translator: **PPG → heart rate/SpO₂**, **ECG → intervals/beat types**, **IMU → activity/posture**. Use adaptive filters, motion-aware windows, and **confidence scores**. For on-device ML, small models (TinyML) classify events without eating the battery—think Loki variants but only the helpful ones. Keep a journal: version models and features, log when the algorithm asks the user to tighten the strap, and explain an alert in one sentence.

Feature extraction, motion gating, model inference, confidence and on-device decisions
Not just “AI”—deterministic math with receipts.

Power, Battery & Thermal

Battery life is the plot twist that decides your reviews. **PMIC + fuel gauge** handle bucks/LDOs, charging and safety; the MCU runs a ruthless sleep schedule; LEDs and radios live on quotas. Thermal modeling keeps skin-contact surfaces comfy and sensors truthful—no one wants a toaster-watch cameo in the Barbieheimer universe.

  • Duty-cycling: Burst PPG, batch BLE, nap hard. Log “time-to-trust” so users know when a number is solid.
  • Heater avoidance: Isolate PMIC heat; place temp sensors where physics agrees, not where layout is easy.
  • Aging: Calibrate fuel-gauge models; track internal resistance; be honest about “% remaining.”
PMIC rails for analog/digital/radio, charger, fuel gauge and thermal considerations
Quiet rails for sensing, thrifty rails for radios, and a battery that tells the truth.

Connectivity, Apps & Security

**BLE** is your loyal courier; **NFC** handles quick pairing; **Wi-Fi** (if you must) moves logs on a charger. Minimize pairing pain, encrypt at rest and in flight, and embrace the offline life—wearables should keep helping even when phones ghost them.

  • BLE hygiene: Use bonded keys, rotate addresses, and prefer GATT profiles humans can pronounce.
  • App UX: Factors and trends over single-number doom. Explain context; teach better strap fit.
  • Updates: Signed firmware, rollback paths, crash-safe writes. Never brick on leg day.
BLE pairing/data flow, secure storage, app dashboards and privacy choices
Share just enough, to just the right places, with just the right words.

EMC, Layout & Enclosure

The best wearables don’t measure their own noise. Separate **loud** (LED edges, radios, haptics) from **quiet** (PPG/ECG inputs). Terminate shields 360° to metalwork; keep reference planes continuous; route short, honest returns; and give the photodiode node its personal space. Enclosure wise, IP ratings apply when mated; gaskets, meshes and IR windows need love.

  • Connectors: Gold where signals are tiny and cycles are many; tin where current is king.
  • Haptics: Schedule buzzes away from sampling windows; you can feel vibes without corrupting them.
  • Water & sweat: Vents for pressure equalization; coatings and seals that survive showers and august marathons.
Shielding, ground strategy, short returns, gasketed enclosure and water ingress controls
Discipline on the PCB; mercy in the enclosure; sanity in the lab.

Safety, Privacy & Compliance

Even if your wearable is “wellness,” build like a grown-up. Electrical safety, EMC, software lifecycle, usability and risk are boring by design. Keep requirement → test → evidence tidy. Document how you handle **alerts**, **false positives**, **data retention** and **consent**. Avoid medical claims unless you’ve truly earned them; accuracy statements must match your verified conditions.

Sample BOM (Component-Level)

  • Sensors: PPG module (multi-λ LEDs + photodiode), ECG electrodes & IA, skin/internal temp, 3-axis accel + gyro, baro altimeter, optional bioimpedance.
  • AFE/ADC: PPG TIA, ECG IA (high CMRR), mux paths, ΔΣ ADC (16–24-bit), precision/ratiometric references.
  • Compute: Low-power MCU/SoC with timers/DMA, crypto, and enough RAM/flash for on-device ML.
  • Connectivity: BLE 5.x (LE Audio optional), NFC for tap-pair, optional Wi-Fi on chargers.
  • Power: PMIC bucks/LDOs, charger, fuel gauge, protections, load switches; verified sequencing.
  • Haptics & UI: VCM/LRA driver, small display or LEDs, side buttons or crown, microphone (if needed).
  • Interconnect: Flex with shielding where needed, gold contacts for low-level signals, strain relief.
  • Enclosure: Gaskets, IR/optical windows, coatings; adhesives rated for sweat, soap and UV.
Sensors, AFE/ADC, MCU, radio, PMIC/battery, UI/haptics and interconnect blocks
Choose parts that play nice together under sweat, motion and Mondays.

Disclaimer: This page covers electronics design for wearable devices and does not provide medical advice or clinical claims.

Ersa

Archibald is an engineer, and a freelance technology technology and science writer. He is interested in some fields like artificial intelligence, high-performance computing, and new energy. Archibald is a passionate guy who belives can write some popular and original articles by using his professional knowledge.

FAQ

What signals do most wearables measure?

Common sets are PPG (heart rate/SpO₂ proxy), ECG (electrical rhythm), temperature, IMU (motion/posture), sometimes bioimpedance (hydration/body comp proxy).

How do I choose PPG wavelengths?

Green for HR robustness; IR/Red for SpO₂ and darker skin tones; multi-λ improves artifact rejection. Consider LED efficiency, skin coupling, and ambient light.

Why are AFEs critical in wearables?

Signals are small and close to noise. Chopper IAs for ECG, low-leakage TIAs for photodiodes, ΔΣ ADCs with stable clocks, and ratiometric references keep data honest.

How do I reduce motion artifacts in PPG/ECG?

Mechanical preload + IMU-gated filtering, windowed averaging, and confidence scoring. Duty-cycle LEDs to avoid rail ripple and display PWM collisions.

What sampling rates are typical?

PPG 100–400 SPS; ECG 250–500 SPS (higher for R-peak fidelity); IMU accel/gyro 100–400 Hz; tune to power budget and algorithm bandwidth.

How do I extend battery life without killing accuracy?

Burst sample sensors, batch BLE transfers, aggressive sleep states, and time-to-trust logic (publish when stable). Pick PMICs with low quiescent current.

Should I process on device or in the app?

Mix both. On-device (TinyML/deterministic filters) for latency and privacy; app/cloud for trends and heavy compute. Version models and log decisions.