How does a datalogger work?
From a sensor signal to a useful history
A data logger takes measurements and stores them with time information according to its configuration. The process involves reading an input, digitizing an analog signal or receiving digital data, converting it to engineering units, processing it and saving a record. A logger creates a history that can explain how conditions changed, rather than simply displaying the latest reading.
1. Read the correct input
Internal or external sensors measure temperature, humidity, pressure and other quantities. Pt100 resistance, a thermocouple voltage and a 4–20 mA transmitter need different input arrangements. Input type, range and wiring must match the sensor. The RTD-versus-thermocouple choice affects the measurement circuit as well as the probe.
For an analog input, signal conditioning prepares the signal and an analog-to-digital converter (ADC) produces a numerical value. With a digital sensor, the logger receives an already digitized reading through a supported interface such as Modbus. That reading does not need another analog conversion at the logger. Supported digital inputs may also record contact states or count pulses.
2. Convert the reading into engineering units
Scaling maps the input to units such as bar or °C. For a linear pressure transmitter configured as 4–20 mA = 0–10 bar, an input of 12 mA represents 5 bar: (12 − 4) ÷ (20 − 4) × 10 = 5. This assumes the transmitter and logger settings use the same range.
RTDs use their resistance–temperature relationship; thermocouples need the correct type and cold-junction compensation. Not every sensor uses a simple linear multiplier. Digital devices also require the correct register interpretation, data type and scale. Measurement accuracy depends on the sensor and input circuitry together.
3. Decide what to sample and what to save
The sampling interval sets how often a measurement is taken. The logging interval determines which results reach persistent storage. Some loggers separate these settings; others use one interval. A device may sample every second but save a one-minute average, minimum and maximum. A record every minute does not necessarily mean a measurement every minute.
Suppose four temperature samples are 20, 21, 25 and 22 °C. Their average is 22 °C, the maximum is 25 °C and the final sample is 22 °C. Saving only the average does not preserve the original samples. Choose the records needed for later analysis before starting the deployment.
Why can a brief event disappear from the record?
In this illustrative example, pressure normally sits at 3 bar. It rises to 8 bar from second 2 until second 3, then returns. The table assumes ideal instantaneous point measurements and ignores sensor response time and filtering. Only selected times are shown.
| Time | Actual pressure | Sample every 5 seconds | Sample every 1 second |
|---|---|---|---|
| 0 s | 3 bar | 3 bar | 3 bar |
| 1 s | 3 bar | No sample | 3 bar |
| 2 s | 8 bar | No sample | 8 bar |
| 3 s | 3 bar | No sample | 3 bar |
| 5 s | 3 bar | 3 bar | 3 bar |
The five-second point-sampling schedule misses the excursion. A maximum calculation only finds the largest sample actually measured; it cannot reconstruct an unmeasured peak. Faster sampling, suitable event capture or dedicated measurement hardware may be needed. A supported hardware pulse counter can count pulses between scans, which is different from taking periodic analog point measurements.
4. Timestamp and store the result
A real-time clock provides timing for measurements or reporting periods. Check which instant the stored timestamp represents. A record may also contain channel information and status flags. Available memory, channel count and record format determine storage duration. Depending on the model, a full memory may stop recording or overwrite the oldest records.
The power source may be a battery or an external supply. Sleep modes and switched sensor power can reduce battery consumption where supported. Retaining old records in nonvolatile memory does not keep new measurements running during a power failure. Provide suitable backup power where uninterrupted recording is required.
5. Retrieve, display or transmit the data
Records can be collected through USB, serial communication, Ethernet or supported wireless links. Live viewing and remote access are model-dependent features. Standalone local logging may work without internet access or a continuously connected computer. SCADA or cloud integration can be added where the system supports it. Retrieved records become tables, plots and reports.
Check the first recording before relying on it
Compare the input against a known reference, then verify the clock, units, channel names and timing settings. Download a short trial record and inspect its timestamps and values. Maintenance, environmental suitability and running costs should also fit the application. A successfully created file does not, by itself, prove a correct measurement setup.
Technical references: Campbell Scientific — data logger functions · Campbell Scientific — scan and storage intervals · Onset — logging modes and memory behavior