Keeping humidity sensors honest

A capacitive humidity sensor is a polymer film, and polymer films age. Typical drift is on the order of one percent relative humidity per year, faster in air with organic solvents or persistent high humidity. Over five years that is enough to turn a compliant store into an apparent problem, or to hide a real one.

Calibration is how you tell instrument decline from building change. It is not difficult and it does not need equipment you cannot improvise.

The salt solution method

A saturated solution of certain salts, in equilibrium with excess undissolved salt, holds the air above it at a known and reproducible relative humidity. Seal a sensor in a small container above such a solution, wait for equilibrium, and compare the reading to the reference value.

Salt	Nominal RH at 20 °C	Useful for
Lithium chloride	about 11 percent	Low-end check, dry stores, metals
Magnesium chloride	about 33 percent	Low-mid, winter heating conditions
Sodium chloride	about 75 percent	High-end check, damp stores, mould threshold
Potassium chloride	about 85 percent	High-end, rarely needed

The values shift slightly with temperature, so the manufacturer’s table for the salt you use is worth having rather than a single memorised figure. Lithium chloride is toxic and should be handled accordingly; sodium chloride is table salt and magnesium chloride is a de-icing salt, both benign.

Two points across the range is the minimum that tells you anything, because a sensor can be accurate at 50 percent and badly wrong at 75. Sodium chloride and magnesium chloride together cover most collection conditions.

Doing the check

1. Make a slurry: enough salt that undissolved crystals remain after stirring in distilled water. It should look like wet sand, not like a solution.
2. Put the slurry in a small dish inside an airtight container, with the sensor suspended above it and not touching the liquid.
3. Leave it somewhere with a stable temperature. Temperature drift during the test is the largest source of error, so a cupboard in an unheated interior room beats a windowsill.
4. Wait. Twelve to twenty-four hours is usual for a logger with a filter cap; a bare sensor equilibrates faster.
5. Read the logger, note the temperature, and compare against the reference value at that temperature.

A deviation within the instrument’s stated accuracy is a pass. A deviation outside it means the instrument either needs adjustment, if it supports one, or needs to be treated as having a known offset, or needs replacing.

Recording an offset instead of adjusting

Many loggers in this class cannot be adjusted. That is not fatal. A sensor that reads consistently 4 percent high is still a perfectly good instrument for detecting change, provided the offset is written down and applied when the data is read.

What is fatal is an unrecorded offset, because the next person will take the numbers at face value. Keep a calibration log per instrument with the date, the salts used, the temperature and the deviation found. The log is part of the climate record, not an administrative extra.

When to pay for a certificate

An in-house salt check is verification. A laboratory calibration is traceable evidence, and the two are not interchangeable when someone else has to rely on the number.

Pay for the certificate when a lender requires evidence of conditions, when an insurance claim depends on the environment, when a funder wants monitoring data as a deliverable, or when a dispute with a contractor turns on whether the plant held the specification. In those cases a reference instrument with a certificate, kept aside and used only to check the working loggers, is a sensible compromise: one calibrated instrument and a documented comparison procedure covers a whole fleet. Thermohygrometers covers what makes a good reference instrument.