Choosing a data logger

A data logger is the cheapest instrument in preventive conservation that produces evidence rather than opinion, and the specification differences between a good one and a bad one are not obvious from the photograph. Five things decide whether the record it produces is worth anything.

Accuracy, and where it is quoted

Nearly all humidity sensors in this class are capacitive: a polymer film changes its dielectric constant as it takes up water. Good ones claim something around ±2 to ±3 percent RH, and that claim is nearly always made at a comfortable mid-range point, typically around 20 °C and 50 percent.

Collection stores are frequently neither. At 10 °C and 75 percent, the same sensor may be two or three points worse, and a specification that gives one headline figure with no range is hiding that. Look for an accuracy curve or at least a stated band. Temperature accuracy is rarely a problem in this class of instrument; humidity is where the money goes.

Drift, and whether you can do anything about it

A humidity sensor is not a ruler. It ages, and it typically reads progressively differently over years. If the instrument has no way to be checked and adjusted, then after three years you have a record of the sensor’s decline superimposed on the room’s climate, and no way to separate them.

Two things make this manageable: a manufacturer-stated drift figure, which implies they expect recalibration, and a documented calibration procedure. Salt solution checks are cheap and adequate for verifying a logger in the field, and the calibration page covers how.

Memory depth, and the download you will forget

This is arithmetic, not judgement. Memory divided by interval gives the time until the logger is full. Then find out what happens when it fills. Roll-over memory keeps recording and discards the oldest readings, so a missed download costs you the beginning of the period. Stop-when-full behaviour ends the record silently, and you find out three months later.

For a monthly download routine almost anything works. For a quarterly one, check the arithmetic before buying.

Readout, and what your staff will actually do

The most accurate logger in the world produces nothing if nobody downloads it. A cable readout means removing the logger from position, which for a sealed case is a real disturbance. A radio readout to a handheld or phone lets you walk the building without opening anything.

Against that, weigh software risk. Proprietary desktop software that runs on one machine and has not been updated in years is a genuine hazard to a twenty-year record. Ask what the export format is before you ask anything about the app.

Siting, which matters more than the model

A logger reads its own immediate surroundings. On an external wall it reads the wall, on top of a case in a sunbeam it reads the sun. Any of the instruments in this class will give you a defensible record if placed properly, and none of them will if not. Sensor placement covers this, and it is the page to read before spending anything.