A measurement is useful; a history explains what happened

A data logger records measurements together with time information so they can be reviewed later. A display tells you the present temperature. A logger lets you investigate when a change began, how long it lasted and whether it happened again. Supported inputs determine whether it can record temperature, humidity, pressure, electrical measurements or pulse counts.

Some units include built-in sensors; others accept external probes, transmitters or readings from connected instruments. “Data logger” therefore does not describe one universal measuring device. Start with the required sensor inputs, rather than the size of the screen or the wireless logo.

What becomes a recorded value?

A sensor responds to a physical quantity. The logger reads its signal, digitizes it where necessary, applies the configured conversion and stores a value with a timestamp and channel identity. That is the basic recording process. Retrieving the results by USB, Ethernet or radio is a separate step.

Input compatibility matters. A probe input differs from a 4–20 mA input or a 0–10 V input. For a linear transmitter configured for 0–100 bar, a 12 mA signal represents 50 bar. Incorrect scaling can produce a smooth, convincing trend with entirely wrong units. For readings collected over Modbus / RS485, check the register address, data type, scaling and byte order as well.

Reading often does not mean saving every reading

The measurement interval tells you how often an input is read. The storage interval tells you how often results are written to a record. A logger might measure once per second and save a one-minute average. Depending on its capabilities, it can also save the minimum and maximum over that minute. An average can conceal a brief excursion. Configuration needs to specify what is retained, as well as how often.

Read the example: Orange represents readings taken every minute. Blue represents readings at minute 0 and minute 10 only, which miss the short rise between them. The dashed blue line simply joins two stored points; it does not prove the temperature stayed constant. The 8 °C limit is illustrative. Choose intervals around the duration of the event you need to capture, and account for sensor response time.

Use trends to investigate, not to invent certainty

Analyzing recorded data can reveal peaks, rates of change, time above a limit and relationships between channels. A door-open event coinciding with a temperature rise gives you a useful lead. Coincidence alone does not establish the cause. Missing samples should be shown as missing, rather than silently treated as zero.

In cold storage, a temperature record helps document conditions. It does not, by itself, establish food safety or spoilage; the product, exposure duration and applicable assessment method matter. Other common applications include transport monitoring, environmental surveys, process observation and field testing. Industrial use cases require hardware suited to the actual quantity and environment.

A live dashboard and an archive are different promises

Live monitoring means a current value reaches a display. Confirm separately that it is being retained as a usable historical record. Wireless loggers and remote access are convenient, but local buffering and recovery after a network outage vary by model. Losing internet access does not necessarily stop local recording.

A PLC may log values while controlling a machine, and SCADA may provide displays, alarms and historical archives. These functions can overlap. Decide which device must retain the evidence you need and what should happen during a failure. An alarm relay on a logger does not automatically make it a safety controller.

Estimate storage using the actual record format

For a simplified example, assume eight values at four bytes each, plus one eight-byte timestamp: that is 40 bytes per record. Saving a complete set every minute gives 1,440 records and 57,600 bytes per day, or about 1.73 MB over 30 days. Real storage requirements also include headers, status fields and format overhead; text-based CSV files need a different estimate.

Non-volatile storage such as flash memory can retain completed records without power. The latest readings that have not yet been committed depend on the device design. Check whether full memory stops recording or overwrites the oldest data. When comparing power options, include sensor power, reading frequency and radio activity in the battery-life calculation.

Questions that make a purchasing decision easier

Which quantities, how many channels and what signal types? Then define the required ranges, accuracy, recording duration, clock synchronization and export format. A 16-bit resolution specification is not an accuracy guarantee. Measurement accuracy depends on the complete chain: sensor, wiring, input circuitry, calibration and operating conditions.

Check the operating temperature, humidity exposure and enclosure when assessing durability. Plan maintenance around the battery, clock, probes and data integrity. Compare total cost including software, probes, calibration and any subscription. With humidity sensors, placement and condensation can matter as much as the headline specification.

Explore the practical questions

How a logger works
Typical applications
Live monitoring
Sensor options
Storage capacity
Working with recorded data
Configuration and programming
Wireless loggers
Power supplies
Cost and selection
Industries and use cases
Maintenance
Remote access
Durability
Measurement accuracy

Technical references: Campbell Scientific — Data Loggers · Campbell Scientific — Measurement and storage intervals