Capacitive Touch Water Rejection: Distinguish Droplets, Films and Flowing Water

Capacitive touch water rejection is often specified as a single yes-or-no feature. In practice, a droplet, a continuous film and flowing water produce different signals, and a controller tuned for one can fail the others. This guide separates the water conditions, adds grounding and glove variables, and defines the recovery and false-contact records that make a wet-touch claim testable.

What must capacitive touch water rejection discriminate?

It must separate water from a deliberate finger.

Water rejection must distinguish a droplet, a film and flowing water from a real touch, while still allowing a wet finger if the product requires it.

Water changes the capacitance of the sensor in several ways. A droplet covers a small area and can look like a finger. A film spreads across many nodes and shifts the baseline. Flowing water changes continuously and can produce a moving pattern. A wet finger adds the water path to a real contact, so the controller sees both at once.

The rejection feature has to classify those cases, not merely detect that something is present. A controller that rejects all large contacts may also reject a gloved hand; a controller that accepts all small contacts may accept droplets. The requirement should state which cases must be rejected, which must be accepted and which may be either.

A controller that reports the same event for a droplet and a finger has not solved the classification problem; it has only changed the threshold. The requirement should also state the acceptable delay before a rejected water contact is cleared, because a long suppression feels like a frozen screen. Record the water condition, the contact and the controller’s response. A test that only says water rejected cannot distinguish a correct classification from a blanket suppression.

How do isolated droplets differ from a continuous film?

A droplet is local; a film shifts the baseline.

A droplet changes a few nodes and can look like a touch, while a film shifts the baseline across many nodes and may suppress all input.

An isolated droplet changes the capacitance at a small group of nodes. If it is the size of a fingertip, the controller may report a contact. A continuous film covers many nodes and changes the baseline everywhere; the controller may interpret it as a large object, reject all input or slowly adapt the baseline until the film becomes invisible.

A slow-moving film can be tracked by the baseline algorithm and become invisible, while a fast-moving droplet is reported as a touch. A film that is present at power-up is especially difficult, because the controller may treat it as the new baseline. The two cases need different tuning. A droplet test that passes does not prove that a film is handled, and a film test that passes may have suppressed the droplet response. Run both, and record the number of false contacts and the time to recover after the water is removed.

The film may also leave a residue after drying, which changes the next test. Clean and dry the surface according to a defined procedure before each condition.

Which grounding conditions change the observed response?

Grounding changes whether water looks like a finger.

Water connected to earth through a bezel or grounded fixture provides a return path a floating droplet lacks, so the same amount of water can behave differently.

Water is conductive, and its effect depends on where the charge can return. A droplet on an isolated cover glass has a different coupling from a film that reaches a grounded metal bezel or a mounting screw. A product on a grounded fixture can behave differently from the same product on an insulated surface.

A grounded bezel can also act as a large electrode, changing the sensor’s self-capacitance even when no water is present. The test matrix should include a floating product, a product bonded to earth, a product held by a user and a product on a conductive surface. If the enclosure has a metal bezel, include a case where the water bridges the bezel and a case where it does not.

Record the grounding arrangement with each result. A water-rejection claim without the grounding condition is incomplete, because the same controller can pass in one arrangement and fail in another.

When do glove settings conflict with water rejection?

Glove sensitivity and water rejection conflict.

Raising sensitivity for a thick glove also makes a water film more likely to register, so tune and test the two requirements together.

A glove reduces the capacitance change from a finger, so the controller may need higher gain or a lower threshold to detect it. The same setting makes a water film, a droplet or a nearby object more likely to cross the threshold. A product that must work with gloves and in wet conditions has a narrower tuning window than one that needs only one of them.

The Infineon CAPSENSE design guide and the Microchip capacitive touch guide describe overlay, grounding and moisture considerations for their controller families. The tuning values are device-specific.

The glove and water settings may be separate modes in the controller, so the product must define which mode is active in each use case. Test the glove and water conditions as a combination, not as separate features. A glove test on a dry screen and a water test with a bare finger do not cover the case where both are present.

How should recovery after drying be evaluated?

Recovery must be measured after complete drying.

Touch should return to its dry baseline after the water is removed, and recovery time and residue effects should be recorded rather than assumed.

After the water is removed, the controller should return to its dry baseline and accept normal touches. The recovery time depends on the water path, the drying method and the controller’s baseline tracking. A film that evaporates slowly may keep the baseline shifted for minutes.

A recovery test should include a deliberate touch immediately after drying, because the baseline may be valid while the threshold is still shifted. Record the time from water removal to the first correct touch and the time to the dry baseline. The drying method should be part of the procedure, because wiping and air drying leave different residue. Check the surface for residue and clean it if the procedure requires. A recovery test that starts before the surface is dry measures a different condition.

The projected capacitive touch design guide covers the sensor and overlay choices that affect how water leaves the surface.

Which sensor regions deserve separate testing?

Test edges, seams and the connector area separately.

Water collects at bezels, seams and connector openings, so the response there can differ from the centre and must be measured separately.

The centre of the sensor is usually the easiest region to test and the least representative. Water collects at the bezel, in a seam, around a connector opening or under a gasket. Those regions may have a different ground path and a different overlay thickness. A touch near the edge may be rejected or accepted differently from the same touch in the centre.

Divide the sensor into regions and test each with the same water condition. A region that is never exposed to water still belongs in the baseline, so the comparison is fair. Record the region and the result. A centre-only test can miss an edge failure that the user meets every time the product is wiped.

A region near a screw boss or a grounding spring is particularly sensitive to water bridging. Include the area around the FPC and any exposed connector, because water can reach the electronics there and change the coupling as well as the insulation.

What performance table should report misses and false positives?

Report misses, false positives and recovery time.

The table should show missed deliberate touches, false contacts and recovery time for each water condition and grounding state.

A water-rejection table needs both error types. A missed touch is a deliberate contact that the controller ignored; a false contact is an input with no deliberate touch. A controller that rejects everything has zero false contacts and many misses, which is not a good result. The table should show both, plus the recovery time.

The table should also record the water volume and the time the water was present, because a short splash and a long soak are different tests. The counts should be taken from the controller’s own report, not from a screen recording. Add the water type, the volume or film condition, the grounding arrangement and the cleaning procedure. A row without those fields cannot be compared with another run.

A change in the controller firmware can alter the classification without any hardware change, so the version belongs in the record. Record the controller configuration and the firmware version. A threshold change can move the result from one row to another.

What tuning changes require renewed environmental validation?

Any tuning or stack change reopens water validation.

A firmware threshold, new cover glass, gasket change or grounding change can alter water behaviour, so each should trigger a re-test.

Water rejection depends on the whole stack: the sensor pattern, the cover glass, the adhesive, the gasket, the bezel and the grounding. A change to any of them can change the coupling and the baseline. A firmware threshold change can also move the boundary between accepted and rejected contacts.

The record should name the person who approved the re-test and the build that was tested. Define the changes that trigger a re-test and record the reason. A new cover-glass coating, a different adhesive, a new gasket or a changed chassis bond should not be treated as cosmetic.

The Linux multi-touch protocol defines how the resulting contacts are reported to the host; the rejection decision happens in the controller, but the host still sees the result.

Water-rejection test matrix

Touch condition	Glove, water or ground state	Controller configuration	Misses and false contacts	Approved scope
Dry bare finger	Floating product	Baseline settings	Reference counts	Reference configuration
Isolated droplets	Dry, floating	Baseline settings	False contacts and misses	Approve only if classified
Continuous film	Wet, grounded bezel	Water-rejection mode	False contacts and recovery	Approve with recovery time
Flowing water	Wet, user holding product	Water-rejection mode	False contacts and misses	Approve only for defined flow
Wet glove	Wet glove, grounded	Glove and water settings	Misses and false contacts	Separate product requirement

The USB HID usage tables and the Linux I2C documentation cover the bus that carries the controller data; they do not define the water-rejection behaviour.