A weighing system that no longer returns to zero is usually reported as "load cell drift". In practice the cell itself is the cause less than half the time. The rest is cable, moisture, mounting, temperature or the instrument. A structured check finds the cause in an hour without a set of test weights.
First, separate zero from span
Load the scale with a known mass and note the reading, then remove it. If zero has moved but the span error is unchanged, the fault is in the zero path: bridge balance, cable, connector or thermal zero shift. If both zero and span have moved, suspect a shunt path or moisture in the cell. This one observation halves the search.
Check the bridge electrically
| Excitation to output | Measure input and output resistance with the cell disconnected. Compare with the certificate; a large change means a damaged bridge or a shorted compensation resistor. |
| Bridge to body | Insulation resistance should read in the gigaohm range at 50 to 100 V. A few megohms means moisture ingress, which is the most common real cause of drift. |
| Zero balance | With rated excitation and no load, the output should be within the specified zero balance, typically a small percentage of full scale. |
Disconnect the instrument before these checks. A reading taken through the instrument measures the cable, the connector and the input circuit together.
Moisture: the usual suspect
A cell rated IP67 is protected against immersion for a limited time, not against a permanently wet cable gland. Water enters through a damaged jacket, a cable entry without a proper seal, or a connector that was opened and closed in the rain. The signature is a bridge-to-body resistance that falls after a humid night and recovers in dry weather. Replace the cable gland and the connector before condemning the cell; in many cases the cell is still within specification once it is dry.
Thermal zero shift
Every cell has a temperature effect on zero, often specified as a small percentage of full scale per 10 K. A 10 t cell with 0.02 % per 10 K moves 2 kg for a 10 K change. On a warm afternoon that looks exactly like drift. Confirm by recording zero against cell temperature over a day; if the two track, the cell is behaving as specified and the fix is mechanical or thermal, not electrical.
Shunt calibration without weights
A shunt resistor across one bridge arm produces a defined output change, which checks the span of the electronics without loading the cell. It does not verify the cell itself, but it separates an instrument problem from a mechanical one in minutes:
- Record the indicated value with no load.
- Apply the shunt and record the change.
- Compare with the calculated value from the certificate and the shunt resistance.
- Remove the shunt and confirm the indication returns to the original zero.
If the shunt result is right and the scale still weighs wrong, the problem is mechanical: load introduction, binding, or a damaged cell.
Mechanical causes that look electrical
- Side load and binding: a stay rod, a pipe or a guard touching the platform takes part of the load and releases it slowly.
- Mounting torque: uneven bolt torque distorts the cell body and shifts zero.
- Creep: a cell that continues to drift under constant load has a mechanical or strain-gauge problem, not a zero problem.
- Cable movement: a stiff cable pulling on the cell adds a real force.
When to replace
Replace the cell when insulation stays low after drying, when input or output resistance is outside the certificate tolerance, or when zero cannot be balanced within the instrument range with a known-good cable and instrument. Otherwise repair the installation. The load cell range lists capacity, accuracy class and ingress rating by series, and the digital instruments page covers the indicators used for these checks.
