A load cell with nothing on it should read zero. At 20 °C it usually does. At 40 °C, many cells read a few counts, and the question is whether that is a defective cell or normal behaviour that the datasheet already promised.
The specification is written as a percentage of full scale per 10 K. On a 300 kg cell with 2 mV/V output, 0.02 % per 10 K is 0.06 kg per 10 K, which is 6 counts on a 10,000 division instrument.
Where the drift comes from
| Source | Mechanism | Signature |
| Bridge resistor tolerance | Four resistors do not track with temperature | Smooth, repeatable drift proportional to temperature |
| Strain gauge mismatch | Gauge temperature coefficients differ | Drift that depends on how the cell was glued |
| Mechanical restraint | Mounting bolts expand and preload the cell | Drift that appears only when installed |
| Cable thermocouple effect | Dissimilar metals at the terminals | Small random offset, sensitive to junction temperature |
| Instrument zero drift | Amplifier offset, not the cell | Present with the cell disconnected |
The order of diagnosis follows the table. Disconnect the cell and see whether the zero still moves; then unbolt it and repeat; only then blame the bridge.
What compensation can and cannot do
A load cell is trimmed with two networks: a thermistor in series with the bridge supply to compensate span, and a resistor in the output to compensate zero. Both are set at the factory, usually at two temperatures, and both assume the cell is mounted the way the factory assumed.
- Compensation covers the cell, not the installation. A cell bolted to a steel frame inherits the frame expansion.
- Two points do not make a curve. A cell compensated at 20 °C and 40 °C can still drift at −10 °C.
- The zero trim has a range. A cell mounted with preload may need the instrument to do the rest.
- Check the certificate. If the drift is inside the stated 0.02 % per 10 K, the cell is doing what it promised.
The practical fix in a weighing system
If the drift is inside specification, the answer is procedural: zero the instrument at the working temperature, or use a dead-load compensation. If it is outside, the cell is a warranty case, and the evidence needed is a log of zero readings at three temperatures with the cell unloaded and the instrument at a stable temperature itself.
Our load cells are supplied with a temperature test sheet covering −10 °C to +40 °C, and the technical notes cover the mounting rules that stop the frame from being the source of the drift.
Standards referenced: OIML R60 (metrological regulation for load cells), GB/T 7551, IEC 60751 for the temperature sensors used in compensation.
A three-point check anyone can run
Load nothing, note the zero at 20 °C, warm the cell to 40 °C in a chamber, note it again, then cool to 0 °C for a third point. Three readings, 4 hours, and the drift is a number instead of an opinion. On a 300 kg cell, 0.02 % per 10 K shows as 0.06 kg per 10 K, which is 6 counts on a 10,000 division indicator.
If the third point falls off the line between the first two, the compensation network is the problem, not the mounting.
