A strain gauge bridge is four resistors, and every resistor changes with temperature. If they all changed equally the bridge would stay balanced, but they do not: the gauge on the tension side and the gauge on the compression side sit on different metal, see different strain, and very often sit at different temperatures.
Where the 0.02 % comes from
For a 500 kg cell at 2 mV/V, 0.02 % of full scale per 10 K is 0.1 kg per 10 K. Over a 30 K swing, that is 0.3 kg of apparent load with nothing on the platform. In a weighing application with a 0.5 kg resolution, the zero has moved more than half a division.
| Specification | Typical value | Effect on a 500 kg cell |
| Zero shift with temperature | 0.02 % FS / 10 K | 0.1 kg per 10 K |
| Span shift with temperature | 0.02 % FS / 10 K | 0.1 kg per 10 K at full load |
| Creep over 30 min | 0.02 % FS | 0.1 kg drift under constant load |
| Hysteresis | 0.02 % FS | 0.1 kg between load and unload |
| Non-linearity | 0.02 % FS | 0.1 kg at mid-range |
The compensation network
Two techniques are used together. The first is to put a temperature-sensitive resistor, usually nickel, in series with the bridge excitation. As temperature rises, the nickel resistance rises and the effective excitation falls, which cancels the sensitivity increase of the gauges.
The second is to add a fixed resistor in the bridge to shift the zero in a way that tracks temperature, chosen by trimming during calibration at two temperatures.
Why it drifts after years of service
The nickel resistor ages. Its temperature coefficient changes by a few percent over a decade, and the compensation that was exact at commissioning becomes approximate. The symptom is a zero that moves further each season, not a sudden failure.
Moisture ingress is the other cause. A cell that has taken water into the gauge area reads low insulation and shows a zero that wanders with the weather. Measuring insulation resistance, not just the output, catches it early.
Standards referenced: OIML R60 for load cell metrology, GB/T 7551 for the Chinese equivalent, IEC 61000-4-3 for the electromagnetic environment the cell must survive.
The specification sheet trap
Temperature effect is often quoted as a single number for the whole compensated range, say 0.02 % FS per 10 K. What matters in practice is the shape of the curve: a cell that drifts linearly can be compensated in the indicator, while one that drifts in a curve cannot.
Ask for the zero reading at three temperatures, for example 0, 20 and 40 °C, and plot them. Three points show whether the compensation is linear over the working range.
What the indicator can and cannot fix
Most weighing indicators can apply a temperature coefficient to the zero, entered as a number per kelvin. That handles linear drift. Non-linear drift, or drift that depends on load history, cannot be fixed in software.
Span drift is separate and usually smaller, but it is not zero: 0.02 % FS per 10 K on a 500 kg cell is 0.1 kg at full load, which matters when the acceptance tolerance is 0.2 kg.
The installation choices that help
Shade the cell from direct sun, keep it away from the heated side of a machine, and avoid mounting it on a surface whose temperature swings more than the cell is compensated for. These are free and often more effective than a better cell.
