Sensors & PLC / Load Cells & Weighing Systems

Why Does My Scale Weight Reading Keep Drifting?

Doc: load-cell-reading-drifting
Section: sensors-plc/load-cells
Revised: 2026-08-21

Drift means the displayed weight keeps moving while the real load stays unchanged. The direction, timing, and load condition matter: a zero that walks with an empty deck is not the same fault as a reading that changes only under a test weight. Record those clues before touching calibration, because a new zero can hide the evidence without fixing anything.

Key Specifications

Specification	Value	Source
Empty-scale trend	A continuing change with no applied load indicates zero drift or an unwanted changing load path; compare against the indicator and load-cell manuals rather than a universal limit.	Rice Lake Weighing Systems, Advanced Load Cell Troubleshooting
Stable-load trend	A gradual output change under a constant load is creep; acceptance must use the exact load-cell datasheet, test duration, load level, and temperature condition.	OIML R 60:2021, Metrological regulation for load cells
Bridge resistance	Measure excitation and signal resistance with the cell isolated; compare readings with the calibration certificate or model datasheet because nominal resistance and tolerance vary by cell.	Interface, Load Cell Troubleshooting, Bridge Circuitry and Zero Balance
Insulation to body or shield	Low or unstable insulation resistance points to moisture, contamination, or pinched wiring; stay within the cell manufacturer's permitted test voltage.	Interface, Load Cell Troubleshooting, Insulation Resistance Tests
Mechanical freedom	Debris, rigid piping, tight check rods, poor alignment, or damaged mounts can cause no-return-to-zero, nonlinearity, and drift as the structure moves.	Rice Lake Weighing Systems, RoughDeck CS Operator and Installation Manual, Troubleshooting

Types & Variations

Single-cell bench or platform scale

One cell supports the platform, so off-center loading, overload stops, platform contact, and cable damage affect the whole reading directly.

Use: Remove the platter and inspect the center load path, mounting faces, overload stops, and cable before condemning the indicator.

Multi-cell floor, hopper, or tank scale

Several cells feed a summing junction box; one wet channel, tight corner, or restrained vessel connection can pull the total reading.

Use: Compare corners or individual cell signals, then substitute or disconnect one channel at a time according to the manufacturer procedure.

Four-wire versus six-wire connection

A six-wire circuit returns sense leads from the bridge so a compatible indicator can correct excitation drop in a long cable run.

Use: Prefer six-wire remote sense for long runs or changing cable temperature when both the load cell and indicator support it; do not improvise sense connections.

Maintenance Steps

Make the machine safe and preserve the evidence
Record the as-found display, raw counts or mV/V when available, temperature, load state, and drift direction. Remove product only through the approved operating procedure. Apply lockout/tagout (LOTO) to electrical, hydraulic, pneumatic, gravity, and stored-energy sources, block suspended or raised equipment, then bleed down and verify zero pressure before entering the scale structure.
A vessel, deck, hopper, or conveyor can move after power is removed. Verify isolation at the point of work and use rated blocking; never rely on a valve, cylinder, or indicator reading as the sole restraint.
Free the mechanical load path
Remove packed debris without striking the cells. Check mount seating, fasteners, overload stops, check rods, flexible joints, hoses, conduit, and piping for contact or preload. Confirm the live structure can move as its design intends and that each load still passes through the sensing axis.
Do not loosen a mount, check rod, or stay rod until the supported structure is secured against tipping, sliding, and gravity movement.
Inspect and dry the wiring system
With excitation isolated, open the junction box in dry conditions. Look for water tracks, corrosion, loose strands, a shield landed at the wrong point, pinched seals, poor cord-grip seating, and crimps closed on insulation. Replace compromised cable or glands, restore drip loops, and close the enclosure on clean seals.
Never perform resistance or insulation tests on a load cell that remains connected to an energized indicator. Observe the manufacturer's insulation-test voltage limit to avoid damaging the strain-gage bridge.
Divide the system with measured tests
Verify indicator excitation at the terminal, compare raw output at zero and under a stable load, and use a compatible simulator to separate the indicator from the field wiring. On multi-cell systems, test one channel at a time and label every lead before moving it. Compare bridge, zero-balance, and insulation results only with the certificate or datasheet for that exact cell.
Calibrate and prove the repair
Allow the equipment to reach a stable operating temperature, exercise the scale, then perform zero and span calibration under the indicator manufacturer's procedure. Verify return to zero, repeatability, and corner or section response with approved test loads. Save both as-found and as-left results.
Keep personnel clear while test weights are hoisted or moved. Use rated lifting gear, control the suspended load, and follow the site lift plan.

Zero drift and span error are two different faults

Movement with an empty deck and a wrong number under a test weight share almost no causes, so name which one you have before touching anything. Zero drift means the display walks with nothing on the scale: a changing parallel load path, water in a cable or junction box, thermal shift in the cell and the steel around it, or a bridge that has been overloaded. Span error means zero holds but a certified test load reads short or long: excitation lost in a long four-wire run, a junction box trim adjusted to hide a bad corner, or a structure that shunts part of the load around the cells. Creep sits between the two. Apply a stable load and the output walks while temperature and mechanics stay still, then zero recovers slowly once the load leaves. One zero adjustment destroys the evidence for all three, which is why calibration belongs at the end of this job and not the start.

What the timing and the load state already tell you

Empty the platform, confirm nothing touches it, and record the reading at startup, after the indicator manufacturer's warm-up period, and again once room or process temperature has moved. Then set a known load near the center, leave it still, and log the indication at regular intervals. Smooth one-way movement at both zero and load is thermal, moisture, or unstable excitation. A loaded reading that moves and then settles is creep, acceptable only inside the cell datasheet limit. Random jumping is mechanical vibration, a loose termination, radio-frequency interference, or a shield landed at both ends. Movement that appears when the load shifts from one corner to another is binding, poor mount seating, or one bad cell in a multi-cell platform. Readings that track sun, washdown, steam, or batch heat usually mean the structure grows into a check rod, a hard conduit, or process piping and hands part of the load to something that is not a load cell. Not every moving digit is drift: an indicator division set finer than the mechanical system supports will display noise that no repair removes.

Substitute a simulator to split the field wiring from the indicator

Disconnect the cells and connect a compatible load-cell simulator at the junction box or the indicator. A steady simulator reading sends the search back out to cells, cable, moisture, and mounting; drift that survives the simulator belongs to the indicator, its supply, grounding, or the electrical environment around it. On a multi-cell system, substitute one channel at a time — the fault usually vanishes when the bad leg leaves the summing circuit. With excitation isolated, measure excitation and signal resistance at each cell and compare against that cell's calibration certificate rather than a generic figure, since nominal bridge resistance and tolerance vary by model. Insulation resistance from the bridge to the cell body and to the shield exposes moisture and pinched wiring, and stays within the manufacturer's permitted test voltage. Also verify excitation at the indicator terminals under load: a voltage that sags when the batch reaches full weight is a wiring problem the calibration routine cannot see.

Free the load path first, calibrate last

Clear packed debris around the deck and mounts, restore free movement, and set check rods and stay rods to the equipment manufacturer's clearance. Hoses, conduit, and process piping must not become a second load path. Dry a wet junction box, replace damaged cable and seals instead of taping the jacket, restore cord-grip seating, and add drip loops. Tighten terminals with power isolated and tug each conductor to catch a crimp closed on insulation instead of copper. Route signal cable away from motor leads, contactors, drive output cable, and welding returns, and use remote sense where both the cell and the indicator support it. A bridge outside the manufacturer's resistance, zero-balance, or insulation limits needs replacement, not adjustment. Only when the reading holds still should the scale be leveled, exercised with load, calibrated at zero and span with approved test weights, and verified for return to zero, repeatability, and corner response. Keep the as-found numbers with the as-left ones; an as-left-only record proves someone adjusted the scale but cannot show whether the original fault was zero drift, span change, or a bad corner.

FAQ

Why does my digital scale drift when nothing is on it?

An empty-scale drift usually comes from mechanical contact, trapped material, temperature change, moisture leakage, unstable excitation, or a shifted load-cell zero. Remove all hidden loads, trend raw counts after warm-up, and use a simulator to decide whether the fault sits in the indicator or the field system.

Can temperature make a load cell reading drift?

Yes. Temperature can move the cell's zero, change cable resistance, alter excitation at the bridge, and expand the supporting structure until piping or check rods bind. A reading that tracks heating and cooling needs both an electrical check and a mechanical clearance check.

How can I tell whether the load cell or indicator is bad?

Replace the field signal with a compatible load-cell simulator. A stable reading with the simulator points toward the cell, junction box, cable, or mounting. Drift that remains with the simulator points toward the indicator, its power supply, grounding, or the local electrical environment.

Should I recalibrate a scale that keeps drifting?

No. Calibration changes the relationship between signal and displayed weight; it does not remove moisture, binding, electrical noise, or a damaged bridge. Stabilize and repair the system first, then calibrate and verify zero return, repeatability, span, and corner response.

References

Spotted an error or have a question about this guide? Email contact@fixtheplant.com — corrections are prioritized.