Note: This page describes the Cal Table in the discontinued Windows application Accordion Pilot. The current interface is the browser-based Web UI – Kalibrering. The concepts below apply to both.
Overview
A calibration table holds the data that converts between raw hardware values and engineering units (volts, amps, °C). Without calibration, readings may carry offset and gain errors; with it, measurements are accurate.
What a calibration table shows
A linear calibration is defined by a gain and an offset per key (y = kx + m). A table-based calibration maps several raw points to calibrated values, for example:
| Raw value | Calibrated value | Unit |
|---|---|---|
| 0 | 0.000 | V |
| 1024 | 1.250 | V |
| 2048 | 2.500 | V |
| 4095 | 5.000 | V |
The details shown include the channel name, calibration type (linear, table or polynomial), the gain/offset or coefficients, the valid raw range and the engineering unit.
Verifying calibration
Apply a known reference (for example 2.500 V from a calibrator), read the channel and compare — the difference should be within specification. For critical measurements, verify at several points (0 %, 25 %, 50 %, 75 %, 100 % of scale) and compute the error at each: error = (measured − actual) / actual × 100 %.
Types of error
- Offset error — all readings shifted by a constant (correct with the offset term).
- Gain error — error proportional to the value (correct with the gain/slope).
- Non-linearity — error varies across the range (correct with multi-point or polynomial calibration).
- Hysteresis — reading depends on direction (average up/down readings).
Calibration interval
Recalibrate critical measurements every 6–12 months, general use annually, and always after repair or after exposure to out-of-range conditions. Calibration requires traceable reference equipment (precision voltage/current sources, a calibrated DMM, and a temperature chamber for temperature sensors). Back up calibration files before any change, and verify after loading.