Every load cell datasheet carries two temperature figures: temperature effect on zero and temperature effect on span, usually expressed as a percentage of full scale per 10 K. Both are small numbers, both are correctly measured, and both routinely fail to predict what happens on a shop floor.
What the number means
A cell rated at 0.02 %FS/10 K changes its zero output by 0.02 % of full scale for every 10 K of temperature change. On a 500 kg cell that is 0.1 kg per 10 K. If the workshop moves from 18 °C in the morning to 30 °C in the afternoon — a 12 K swing that is entirely normal near a loading door — the zero moves by about 0.12 kg before anything is placed on the scale.
That is inside the specification. It is also visible on a display that resolves 0.05 kg.
The three things the number does not include
1. The gradient, not the temperature. Compensation is characterized with the whole cell at a uniform temperature. In service, one end of the cell is bolted to a warm frame and the other end is in moving air. A gradient of 3 K across a 150 mm cell body produces a bending moment that the compensation network was never characterized for.
2. The time constant. A cell does not reach its new zero instantly. The thermal mass of the steel body means the reading keeps moving for 20 to 60 minutes after the air temperature changes. A reading taken ten minutes after the door opened is not wrong — it is early.
3. Creep and recovery. Temperature change and load change interact. A cell loaded at 80 % of capacity and then warmed will show a drift that is neither pure creep nor pure temperature effect, and neither compensation covers the combination.
What to do about it
| Measure | Cost | Effect |
|---|---|---|
| Zero the scale before each weighing batch | seconds | removes accumulated zero drift |
| Shield the cell from direct air movement | a sheet metal guard | removes the gradient and the fast transient |
| Check with a known test weight at start and end of shift | two minutes | turns an invisible drift into a number |
The third row is the one that changes behaviour. A cell whose zero has drifted 0.12 kg looks identical to a cell whose zero has not drifted at all, until a test weight is placed on it. Two minutes at the start and end of a shift turns the temperature question from a debate into a measurement.
Why a better cell is not always the answer
The obvious response to drift is to buy a cell with a tighter temperature specification. That works, up to a point. Below about 0.01 %FS/10 K, the remaining error is dominated by installation effects — gradient, air movement, mounting torque — rather than by the cell. Money spent on a tighter cell without fixing the installation buys a number that still does not describe the workshop.
The order that actually works is: fix the installation first, then measure the residual drift, then decide whether a better cell is needed. In most cases it is not.
A note on legal weighing
For instruments used in trade, the applicable requirements come from the OIML R 76 series and the national metrology regulations rather than from the datasheet. Those requirements specify maximum permissible errors at defined temperatures and often require the instrument to be tested across a temperature range, not at a single point. A datasheet figure of 0.02 %FS/10 K tells you what the cell does; it does not tell you whether the instrument passes.
