How well an electrode or sensor seats on skin directly sets signal-to-noise ratio

What it is

The first limit on a wearable physiological channel is often not the algorithm but whether the part is actually on the skin. An air gap, a loose strap, a dry electrode that stands off, all decay the target by a square or exponential, and noise (ambient light, mains, cable micromotion) grows relatively. Signal-to-noise ratio (SNR) is nailed at contact; later filters only work inside an already dirty observation.

Why it happens

Optical heart rate modulates return light by pulsatile blood volume under the skin. A few tenths of a millimetre off, and light reaches the photodiode without passing through tissue; the AC term collapses and DC jumps. For biopotentials (EMG, ECG, EDA) contact is electrical impedance: too little pressure shrinks effective electrode area and raises impedance, amplifying thermal noise and motion artifact. Too tight alters local perfusion, flattening the optical signal, and over hours produces pressure injury. Fit therefore lives in a narrow window: coupled, not ischemic. Textile electrodes and hard windows differ in how they sit in that window; both obey the same contact physics.

Studying it

On one person and one algorithm, step strap holes or electrode pressure and measure AC/DC ratio, heart-rate error against ECG, EMG SNR. Factors: contact force, air gap present or not, skin tone or hair (optical). Outcomes: SNR, detection rate, artifact fraction. A pressure or capacitive “worn” detector should be reported with quality, so “on the body” is not billed as “seated.” Laboratory SNR with tape does not represent a daily watch.

Where it stops holding

Ear-canal and chest-strap contact is more stable than the wrist; findings do not move across sites. Green PPG AC is already weak on dark skin, so poor contact hurts more; infrared or longer wavelengths change the curve without cancelling contact dependence. Implanted or adhesive clinical electrodes, once stuck, turn fit into adhesive failure—still contact, on a different timescale.

Applying it

• Make adjustable fit (strap holes, elastic section) the first unboxing step, not the last page of a manual.
• Use wear detection on optics: emit no heart rate when unseated, so ambient light cannot invent a pulse.
• Accept at both too-loose and too-tight; perfusion drop from overtightening counts as a quality fail.
• Verify by comparing SNR when users don as they habitually do versus when an experimenter reseats to spec; if habitual wear is clearly worse, the product has not solved fit.

Related

• Same group: C9.09.2 Sweat, hair, and changing skin alter contact impedance during wear · C9.09.3 Loose or misplaced wear systematically shifts the measurement baseline · C9.09.4 Users can rarely judge fit themselves; the system needs to show signal quality
• Adjacent: C9.01 EMG Input · C9.03 Heart Rate and Electrodermal Activity
• Search: signal-to-noise ratio · optical coupling · wearable contact pressure