Float and Capacitive Water-Level Sensing: Drift and Calibration

A brand received a spike of complaints that a humidifier shut off while the tank was still half full. This is a common field pattern. Reports say early shutoff, while engineers find a sensor that is behaving exactly as calibrated on a dry bench. This note covers why water-level sensing drifts and how to write a shutoff spec.

Calibration is only durable when the curve travels with the part number. A housing change, a new tank grade, or a cost-down that swaps the plastic can all move the baseline, and a stored curve that no longer matches the geometry will read early or late without any visible fault. Keeping the curve version, the tank part number, and the test condition together turns a silent drift into a traceable change. That matters most in the field, where the complaint is subjective and the sensor is behaving as calibrated. A short log of the trip point at full, half, and near-empty, taken with the water the market actually uses, is enough to show whether the setpoint moved or the surface changed. When a brand and a supplier share that log, the discussion moves from whether a unit is defective to which of the two failure directions is occurring, and that is a far easier conversation to resolve.

Why does a low-water shutoff trip early or late over time?

Mineral films and light biofilms change the surfaces the sensor reads. A float can hang on scale, and a capacitive electrode sees a different dielectric constant once a mineral layer builds up. The setpoint is only as stable as the surface in contact with the water.

Some drift is deliberate. Late shutoff protects output at the cost of running the transducer dry, while early shutoff protects the transducer at the cost of user complaints. The setpoint is a choice, so it should be stated, not left to the supplier.

What makes a float sensor drift in mineral-rich water?

Scale adds mass and friction to the float, so it needs more water to lift and more time to respond. In the worst case it sticks in a low position and holds the unit off. Sealing and surface finish of the float bore drive how fast this happens.

You can reduce drift with a smooth bore, a low-friction float material, and a geometry that self-cleans as the water level moves. It also helps to keep the sensing point away from the highest-solids zone, which is usually near the water surface.

How is capacitive sensing calibrated across tank geometries?

Capacitive sensing reads a change in capacitance as water enters the field of the electrode. Wall thickness, tank material, and nearby metal change the baseline, so each geometry needs its own calibration, not a shared constant. A housing change can invalidate the old curve.

Calibrate with the production tank, not a prototype, and store the curve with the part number. If the plastic supplier changes a grade with a different dielectric constant, the curve must be rechecked. This is a frequent cause of silent drift after a cost-down.

What is the field test for shutoff accuracy?

Fill to marked levels and log the trip point at full, half, and near-empty, then repeat after a defined number of cycles with the same water used in the field. Compare against the setpoint and report the spread, not a single value.

Run the test in the real enclosure, because a sealed unit holds mist and heat that change the local behavior. A bench fixture open to the room will not reproduce condensation on the sensing surface.

How do we write the shutoff spec so it is auditable?

State the setpoint, the tolerance, the water level definition, the test temperature, and the cycle count to the acceptance check. Add the required response time and the behavior after a power cycle. Then tie the verify steps to a named method so two labs get the same answer.

Prepared by 燕七.