Fingerprint locks are sold on capacity, template count, and unlock time. The failures that generate support calls come from neither. They come from hands: a dry winter, a grandmother whose ridges have worn smooth, a child whose prints are still too small, a tradesman who has spent twenty years with abrasive material and solvent. The reader's technology decides which of those hands it can serve, and that decision is made at the sample stage, long before anyone argues about the app.
Three technologies with three different failure families
The optical reader takes a photograph of the ridges. It copes with very dry skin, which is its main advantage at a door, and it is the most vulnerable to what the last person left behind: a latent print on the platen is a real attack surface, and a scratched or greasy platen degrades every subsequent read. The capacitive semiconductor reader measures the live skin's capacitance across a grid. It is small, cheap, and good at rejecting a photograph, and it is the one that fails first in the cold, on cracked fingertips, and through a film of water. The ultrasonic reader builds a three-dimensional map below the skin surface; it handles wet and dry hands better than either, costs more, draws more, and is still rare on doors.
The error rate that hurts is the false reject
Two numbers describe a reader, and they pull against each other. The false accept rate is the chance that a stranger gets in; the false reject rate is the chance that the owner does not. Marketing tends to quote whichever number is more flattering, usually the false accept rate, because a small number sounds secure. At a door the false reject rate is what the household experiences, and small percentages are not small in use: a rate of one in a hundred across a family of four unlocking twenty times a day is a failure nearly every day. A reader is not secure by rejecting everyone and it is not useful by accepting everyone, and the trade-off has to be quoted for a named population, not for an ideal finger.
Hands are not a uniform input
Fingerprint quality falls with age, with manual work, with contact with detergents and solvents, and with anything that dries the skin. Children below school age often have ridges too fine to enrol reliably. Cold hands reduce blood flow and change the capacitance signal, which is why a lock that works all summer starts refusing people in an unheated porch. The practical mitigations are unglamorous: enrol the same finger several times at different angles, register two or three fingers per person, and keep a fallback that does not depend on the reader at all.
Where the template lives, and why it drifts
The fingerprint is stored as a template, not an image, and the template's home matters. A template held in the lock stays with the door and survives a change of platform; a template held in the cloud follows the account and can be lost with it. Many readers also update the stored template slightly on every successful match, which adapts to a finger that is slowly changing and, over years, can drift far enough from the enrolment that the original angle stops working. That behaviour is invisible until it fails, and the remedy is a periodic re-enrolment rather than a repair.
Anti-spoof is a separate purchase
Rejecting a photograph is not automatic on all technologies, and a latent print left on a platen is a different attack again. Liveness detection, whether it looks at skin properties, subsurface structure, or micro-movement, is a design choice that either exists in the product or does not. It belongs in the specification as a named method, because "fingerprint" alone says nothing about whether a lifted print works.
Wake behaviour is a battery decision
A reader that is always scanning draws current all day and is fast. A reader that wakes on touch is slower by a fraction of a second and lasts far longer on a cell. A reader that wakes on a capacitive ring around the sensor is a good compromise and a common source of false wakes in rain. None of these appear in the unlock-time figure, and all of them decide whether the lock is pleasant in the second year.
Six questions that predict the complaints
Ask for the technology type by name. Ask for the false reject and false accept rates together, with the population and the finger condition they were measured on. Ask for the working temperature range and how the reader behaves below it. Ask how many templates per user and whether one finger can be enrolled from several angles. Ask what liveness method rejects a lifted print. And ask what the fallback is when the reader refuses: a keypad code, a physical key, and an emergency power port are three separate answers, and a lock missing any of them has converted an inconvenience into a locksmith bill.
A reader that suits a warehouse door and a reader that suits a family home are different products, even when they carry the same sensor. At XingLiao's smart lock range the reader type, the enrolment guidance, and the fallback paths are stated per model, because those are the three things a household actually lives with. For private-label programmes, the lock manufacturing line treats the sensor choice as a market decision: the same mechanism with an optical head and with a capacitive head serves two different climates and two different sets of hands. Specifications and the fallback policy can be discussed through our factory page or directly on the contact form.
The short version: pick the reader by the hands that will use the door, then check that the product still opens when the reader says no.
Top comments (0)