PPG Sensor Wearable Accuracy: Green and Red LED Limits

19 Sep 20261063 wordsWingPhoenix AI factory team

Optical measurement on a wrist wearable rests on one physical principle: light enters skin, some of it scatters back, and blood volume changes modulate that returning signal. Green light penetrates shallowly and tracks the pulse wave well when the wrist is still; red and infrared light penetrate deeper and are used for the ratio that produces a blood oxygen estimate. Neither wavelength tells you anything about the wearer's health on its own. The device reports a waveform-derived number, and the quality of that number is decided by contact pressure, skin perfusion, ambient light and movement long before any algorithm runs. Buyers who understand this ask different questions of a supplier than buyers who treat ppg sensor wearable accuracy as a chipset specification.

What green light actually measures

Green PPG is a relative measurement. The optical heart rate sensor sees a pulsatile component superimposed on a large steady component, and the firmware extracts a rate from the pulsatile part. That works well at rest and during steady low-intensity movement. It degrades when the sensor lifts off the skin, when the strap is too loose to hold optical coupling, when the wearer has cold hands and peripheral vasoconstriction reduces the signal, and when skin tone or tattoos absorb the emitted light before it reaches the capillary bed. A supplier who quotes a single accuracy figure without describing the test condition is telling you very little. Ask instead how the reference was collected, on how many subjects, at what activity levels, and whether the comparison device was chest-strap ECG or something weaker.

Motion artefact is the dominant error source

During running, the wrist accelerates and the sensor-to-skin distance changes every stride. The resulting signal artefact sits in the same frequency band as the pulse, so filtering cannot separate them cleanly. This is why the same module can look excellent in a seated bench test and poor on a treadmill. Accelerometer-assisted tracking helps, but it helps by inferring cadence, not by recovering a clean optical signal. On a GPS sport watch the honest engineering answer is to state the conditions where the optical reading is reliable, and to support an external chest strap for interval work. Buyers should treat that as a feature statement, not a limitation to be hidden.

Red and infrared: the SpO2 boundary

Blood oxygen estimation uses the ratio of light absorption at two wavelengths and depends on assumptions about calibration that hold best at rest, in good contact, with adequate perfusion. Cold fingers, nail polish, movement and poor fit all shift the estimate. The regulatory boundary is therefore not a technical detail but the core of the product definition: SpO2 measurement on a consumer wearable is wellness guidance, never diagnosis, and no app screen, retail listing or packaging line may suggest otherwise. If a distributor asks for wording that implies screening, detection or clinical use, the correct answer is no. The same discipline applies across the Health & Wellness line, where every metric is presented as a trend the wearer observes, not a finding a clinician acts on.

Form factor changes the optical problem

Where the sensor sits determines what it can measure. A wrist device fights motion and fit; a finger-worn device has better perfusion and more stable contact but a smaller battery and different user expectations. A smart health ring can hold optical coupling more consistently than a loose watch, yet it still reports wellness trends rather than clinical values. A health smartwatch carries the opposite trade-off: more room for a larger optical window and more compute for artefact rejection, but more mass to stabilise against the wrist. Neither form factor removes the underlying physics.

Procurement sequence that avoids rework

1. Fix the claim wording first, then the hardware. Copy drives the compliance scope.
2. Confirm the MOQ, sampling and lead times against your forecast before committing to a sensor window size.
3. Book CE RED and the other test scopes per hardware revision with an accredited laboratory, not once for the product family.
4. Review the QC gates and test flow that apply to your build, including IQC, IPQC, FQC and OQC at the agreed AQL.
5. Decide whether the project needs OEM/ODM tooling or a configured standard model, because the schedule differs by months.

Where an importer should say no: to any supplier who offers a clinical-sounding SpO2 claim, who cannot describe the test condition behind an accuracy figure, or who treats certification as a one-off document rather than a scope booked per hardware revision. WEEE, BattG, DEEE/Triman and EORI/VAT obligations sit with you as the importer, so the paperwork conversation belongs in the first meeting, not the last. Start by sending your target markets, claim wording and forecast volume to the team through the product catalogue enquiry route, and ask for a written statement of what the optical channel can and cannot support.

Questions buyers ask

Can a consumer wearable claim to measure blood oxygen for medical purposes?

No. SpO2 output on a consumer device is wellness guidance, never diagnosis, and packaging or app copy must say so. Any medical claim moves the product into a different regulatory scope that a general consumer build does not cover.

Why does heart rate read badly during exercise but well at rest?

Green-light PPG depends on stable skin contact and low relative motion between sensor and wrist. Running, weight training and cold hands all break that contact, so the optical heart rate sensor output degrades even when the algorithm is sound.

Related: more on ai wearables · buyer guides · customisation tiers