A load cell is loaded with:
5,000 N
.
After complete unloading, however, the evaluation system does not return to:
0 N
but instead shows, for example:
+18 N
.
After a few minutes, the value slowly drops to:
+7 N
and remains there.
Is the load cell defective?
Not necessarily.
A zero point that does not return completely can have various causes that must be assessed very differently from a metrological point of view.
These include:
- creep under constant load,
- creep recovery after unloading,
- hysteresis,
- mechanical stress,
- transverse or lateral forces,
- friction in the force introduction,
- cable or hose forces,
- temperature changes,
- electrical zero-point drift,
- a previous overload.
Observing the behavior over time is particularly important.
A signal that initially returns slowly after unloading behaves differently from a value that immediately settles at a new fixed zero point after unloading.
It is equally important to determine whether the deviation is still present when the load cell is removed from the machine and tested under defined reference conditions.
The most important diagnostic question is therefore not only: “How large is the zero-point error?”, but rather: “How does the signal change during loading, unloading and the subsequent waiting period – and does the error remain when mechanical side effects are excluded?”
Why Does a Load Cell Normally Return to Zero?
A conventional strain-gauge load cell has an elastic spring element.
Under load, this body is deformed very slightly.
Strain gauges attached to the spring element detect this deformation and change their electrical resistance.
A Wheatstone bridge circuit converts this into a load-dependent output signal.
Typical passive load cells, for example, provide a signal in:
mV/V
.
When the load is removed again and the permissible elastic range has not been exceeded, the spring element should largely return to its original state.
The output signal should therefore also return close to its original zero point.
In a real measurement, however, this return is not mathematically perfect.
Material properties, strain-gauge adhesives, temperature, mechanical installation and the complete force introduction cause small deviations.
Load cells therefore have specified characteristics such as:
- hysteresis,
- creep,
- zero balance,
- temperature effect on zero,
- repeatability.
Distinguishing Creep, Hysteresis and Zero Return
In practice, several effects are often grouped together under the term:
zero-point drift
.
For systematic troubleshooting, however, they must be considered separately.
| Effect | Typical Behavior | When Visible? |
|---|---|---|
| Creep | Signal changes slowly despite constant load | while the load remains applied |
| Creep recovery | Signal slowly moves back toward the original zero after unloading | after removing a load that had been applied for a longer period |
| Hysteresis | the same force value produces different signals during increasing and decreasing load | when comparing ascending and descending load sequences |
| Mechanical stress | zero point depends on mounting, friction or mechanical position | often immediately after a load change or movement of the mechanics |
| Overload damage | new permanent zero point or changed characteristic curve | after exceeding permissible forces or moments |
| Temperature drift | zero point changes with sensor or ambient temperature | during warm-up or temperature changes |
This distinction is important because the corrective measures are completely different.
Mechanical stress cannot be corrected by a new electrical zero adjustment.
Normal creep recovery, on the other hand, is not automatically an indication of a damaged load cell.
What Is Creep in a Load Cell?
Creep refers to the:
time-dependent change in output signal under a constant applied load
.
Example
A load cell is loaded with:
10,000 N
.
Immediately after the load is applied, the evaluation system indicates:
10,002 N
.
After 30 minutes under unchanged load, it indicates:
10,009 N
.
Although the mechanical load has remained unchanged, the signal has slowly changed.
This time-dependent change is creep.
It results from the time-dependent behavior of the complete elastic measuring system, including:
- spring element,
- strain gauges,
- adhesive or bonding layers.
The permissible magnitude depends on the specific sensor and is specified in the data sheet for a defined load and time period.
Creep is evaluated under constant load. A residual value after complete unloading is therefore not automatically “creep”.
What Does Creep Recovery After Unloading Mean?
After a load has been applied for a longer period, the output signal may continue to change over time even after the load has been removed.
This effect is referred to as:
Creep Recovery
.
Typical behavior
Before loading:
0 N
.
Immediately after unloading following a prolonged load:
+20 N
.
After a few minutes:
+9 N
.
After further stabilization:
+3 N
.
Such behavior differs significantly from a zero point that immediately jumps to:
+20 N
and then remains practically unchanged.
The signal behavior over time therefore provides important diagnostic information.
What Is Hysteresis?
Hysteresis describes the fact that the output signal at the same load can depend on the direction from which that load point is approached.
Example
A load cell has a measuring range of:
0...10 kN
.
During increasing load, at:
5 kN
the system measures, for example:
5.001 kN
.
The load cell is then loaded up to 10 kN and subsequently unloaded again.
During decreasing load, the same point of:
5 kN
now produces:
5.008 kN
.
The difference between the two measured values forms part of the hysteresis.
A proper hysteresis test therefore always requires at least:
one ascending load sequence
and:
one descending load sequence
.
A single zero-point value after one loading cycle is not sufficient to assess hysteresis.
Mechanical Stress as a Common Cause
In machines and weighing systems, poor zero return is often not caused by the sensor itself.
The load cell is part of a mechanical force path.
If this force path does not behave completely reversibly during loading and unloading, a residual force remains on the sensor after unloading.
Possible causes include:
- non-flat mounting surfaces,
- misaligned force-introduction components,
- stressed bolted connections,
- missing joints or self-aligning elements,
- distortion of the machine frame,
- lateral contact with stops,
- friction in guides,
- insufficient mechanical freedom of movement.
A decisive observation is:
If the zero point changes when screws are loosened, the mechanics are moved or the load cell is removed, this strongly indicates a mechanical influence.
Detecting Transverse Forces and Off-Center Loading
A force transducer is normally designed for a defined primary measuring direction.
Additional forces and moments can influence the signal.
These include:
- transverse forces,
- lateral forces,
- bending moments,
- torsional moments,
- off-center force introduction.
A force introduction in which the mechanics shift slightly during loading is particularly critical.
During unloading, the mechanism may then fail to return to exactly the same mechanical starting position.
The load cell therefore continues to experience a small force or moment.
This effect can incorrectly appear to be sensor hysteresis.
In high-accuracy test stands, the force introduction should therefore be designed so that unavoidable small alignment errors can be compensated by suitable force-introduction components or joints.
Eliminating Friction and Force Shunts
In weighing systems, the entire force to be measured should pass as completely as possible through the load cell or the intended load cells.
An unintended additional force path is referred to as a:
force shunt
.
Typical examples include:
- a vessel touching a rigid pipeline,
- a platform rubbing against a frame,
- a transport lock still making contact,
- a stop with insufficient clearance,
- a flexible connection that is too stiff,
- a guide that is binding.
Friction is particularly problematic.
A frictional connection may shift in one direction during loading and then remain in a slightly different position during unloading because of static friction.
The resulting force distribution is then no longer identical to the original condition.
This causes:
- poor zero return,
- apparent hysteresis,
- poor repeatability.
Why Even the Connection Cable Can Influence the Zero Point
From a mechanical perspective, a connection cable behaves like a small spring.
With large load cells, the force generated by it is often negligible.
With small measuring ranges or high-resolution force measurements, however, it can become clearly visible.
A cable is particularly unfavorable if it:
- is under tension,
- is sharply bent directly behind the sensor outlet,
- is fixed to a moving structure,
- changes its position during loading and unloading,
- is tightly secured with a cable tie.
After a load cycle, the cable may lie slightly differently than before.
This results in a different residual force acting on the sensor.
With small force ranges, this alone can already produce a visible zero-point offset.
Test
With the sensor completely unloaded, carefully change the position of the cable.
If the measured value changes reproducibly, the cable routing must be mechanically optimized.
The cable should be routed:
- without tension,
- with a suitable bending radius,
- with minimal mechanical influence on the sensor.
Temperature Change or Mechanical Error?
Temperature can also change the zero point of a load cell.
Load-cell data sheets therefore specify the:
temperature effect on zero
.
A typical problem occurs when:
- the load cell is brought from a cold storage area into a warm test room,
- a motor or hydraulic unit heats the test stand,
- sunlight heats only one side of the sensor,
- the sensor is thermally affected by the process itself.
The electrical self-heating of the strain-gauge bridge after the excitation voltage is switched on can also cause a stabilization period.
For accurate testing, the complete measuring system should therefore be allowed to reach thermal equilibrium before zero adjustment.
Typical indication of a temperature effect
The zero point changes slowly and correlates with the sensor or ambient temperature, regardless of whether a high mechanical load was previously applied.
Permanent Zero-Point Shift After Overload
A considerably more critical cause is mechanical overload.
If the permissible load of the spring element is exceeded, it may leave its purely elastic range.
This can result in permanent:
plastic deformation
.
After unloading, the sensor no longer returns completely to its original mechanical condition.
Typical indications include:
- sudden large zero-point shift after an event,
- changed sensitivity,
- poorer linearity,
- increased hysteresis,
- deviating bridge resistance values,
- visible mechanical damage.
Typical causes of overload are not limited to excessive axial forces.
The following can also be critical:
- impact loading,
- lateral forces,
- excessive moments,
- misalignment during installation,
- tightening screws while the assembly is already mechanically stressed.
Re-zeroing electrically does not reverse mechanical overload damage.
The zero value may then appear correct again, while linearity, hysteresis or structural integrity may already be impaired.
Checking the Electrical Measuring Chain Separately
Before removing the load cell, it should be determined whether the deviation actually originates from the sensor.
The complete signal chain may, for example, be:
load cell → cable → junction box → measuring amplifier → PLC / display
.
Possible electrical faults include:
- unstable bridge excitation,
- contact resistance,
- moisture in junction boxes,
- poor connector contacts,
- damaged cables,
- EMC interference,
- measuring amplifier drift.
A useful test
The load cell is completely unloaded and its raw signal is monitored as directly as possible at the sensor or at the measuring amplifier input.
If the raw signal remains stable while the PLC indication drifts, the cause is probably further downstream in the measuring chain.
With a passive strain-gauge load cell, a suitable:
mV/V simulator
can also be used.
This allows the measuring amplifier and evaluation system to be tested independently of the mechanical load cell.
Testing the Load Cell Outside the Machine
One of the most informative tests is to remove the load cell from the problematic mechanical setup.
It is then loaded in a defined test fixture.
This eliminates numerous influences such as:
- pipe forces,
- guide friction,
- frame stress,
- stops,
- machine movement,
- problematic process connections.
If the sensor shows:
- good zero return,
- good repeatability,
- a plausible ascending and descending characteristic
in the reference setup, the cause is very likely to be the original mechanical installation.
If the error remains even in the controlled test setup, a sensor fault or changed calibration becomes more likely.
A Simple Loading and Unloading Test
The following procedure can be used for an initial systematic assessment.
1. Allow the system to thermally stabilize
Operate the sensor and electronics for a sufficient period under stable ambient conditions.
2. Mechanically unload the sensor
Ensure that there is actually:
0 force
on the measuring axis.
3. Document the zero value
For example:
Z0
.
4. Increase the load in steps
For example:
0 → 25 → 50 → 75 → 100%
.
5. Hold the load at the upper point
Observe the measured value over a defined period.
This makes creep behavior visible.
6. Unload in steps
For example:
100 → 75 → 50 → 25 → 0%
.
The intermediate values allow hysteresis to be assessed.
7. Document the zero point immediately after unloading
For example:
Z1
.
8. Continue monitoring the zero point
Also document it after, for example:
1 min, 5 min, 15 min, 30 min
.
This shows whether creep recovery takes place or whether a stable residual offset remains.
9. Repeat the measuring cycle
A repeated load sequence shows whether the error is reproducible.
Correctly Interpreting the Signal Over Time
| Observation | Likely Direction for Troubleshooting |
|---|---|
| Signal slowly drifts under constant load | Creep, temperature, electrical stabilization |
| Signal slowly returns toward zero after unloading | Creep recovery |
| Ascending and descending values differ at the same load | Sensor hysteresis or mechanical friction/stress |
| Zero value changes when a cable or hose is moved | Mechanical force from the line |
| Zero point jumps when a screw is loosened | Mechanical stress |
| Zero point changes with temperature even without a load change | Temperature coefficient or thermal setup |
| Large permanent zero offset immediately after overload | Possible plastic deformation or sensor damage |
| Sensor operates correctly outside the machine | Error probably in the mechanical installation |
This assignment does not constitute a definitive diagnosis.
However, it helps to select the next tests in a targeted manner.
Practical Example: Force Sensor Remains at +120 N After a 20 kN Load
In an assembly test stand, compressive force is measured using a force transducer.
The measuring range is:
0...25 kN
.
After a load cycle up to:
20 kN
the system indicates after complete unloading:
+120 N
.
Step 1: Observe the signal over time
The value remains almost unchanged for ten minutes.
Pure creep recovery is therefore less likely to be the sole cause.
Step 2: Check the cable
The connection cable is moved carefully.
The zero point changes by only a few newtons.
The cable is therefore not the main cause.
Step 3: Check the force introduction
The upper compression plate is loosened.
The zero point immediately jumps from:
+118 N
to:
+14 N
.
This is a clear indication of mechanical stress.
Step 4: Inspect the mounting surfaces
The inspection reveals that the upper mounting part is not completely parallel to the support surface of the force transducer.
Tightening the screws therefore introduces an additional bending moment.
Step 5: Correct the force introduction
The support surface is corrected and a suitable force-introduction component is used.
Step 6: Repeat the measuring cycle
The same loading cycle is repeated.
After unloading, only a small offset initially remains and continues to decrease during the subsequent stabilization period.
The load cell itself was therefore not the primary cause.
This example shows why poor zero return after unloading should never be considered in isolation. An apparent sensor fault can originate entirely from the mechanical force introduction.
When Calibration Is Advisable
Calibration or a reference test is advisable if:
- the zero point remains permanently changed even in a controlled test setup,
- an overload is suspected,
- the ascending and descending characteristic is abnormal,
- the sensitivity no longer matches previous measurements,
- a specified calibration interval has been reached,
- a new assessment is required after repair or mechanical modification.
A calibration can show, among other things:
- zero-point deviation,
- sensitivity deviation,
- linearity behavior,
- hysteresis,
- repeatability.
However, it does not replace mechanical diagnosis of the machine.
A force transducer may perform perfectly in the calibration laboratory and still produce large measurement errors when installed if the force introduction is unsuitable.
Systematic Troubleshooting
- Allow the sensor and measuring electronics to thermally stabilize.
- Ensure that no force is actually acting on the measuring axis.
- Document the current zero point.
- Observe the signal under constant load for a defined period.
- Apply and remove the load in several steps.
- Compare ascending and descending values at identical load points.
- Document the zero point immediately after unloading.
- Continue recording the zero point during the subsequent recovery period.
- Check cables, hoses and flexible lines for mechanical influence.
- Inspect mounting surfaces and force introduction.
- Eliminate transverse forces, bending moments and off-center loading.
- Check frames, guides, stops and protective devices for friction.
- For vessel weighing systems, check pipes and flexible connections as possible force shunts.
- Check mounting screws and specified tightening torques.
- Compare the temperature trend with the zero-point trend.
- Check bridge excitation and the measuring amplifier.
- If necessary, test the electronics with a suitable mV/V simulator.
- If possible, test the load cell outside the machine under reference conditions.
- After an overload event, do not merely re-zero the system; check the complete sensor characteristic.
- Document the results and zero-point trend for future comparative measurements.
Common Diagnostic Errors
- Calling every residual value after unloading creep: Creep is evaluated while a constant load remains applied.
- Confusing creep recovery with hysteresis: Hysteresis is assessed using ascending and descending load sequences.
- Suspecting only the sensor: Mechanical friction and stress are very common causes.
- Re-taring before every test: This hides the actual zero-point trend.
- Testing only 0 and 100%: Hysteresis and characteristic errors are more clearly visible at intermediate points.
- Not documenting waiting times: Without time information, creep and creep recovery can hardly be compared meaningfully.
- Underestimating cable forces: With small measuring ranges, even a stiff connection cable can be relevant.
- Ignoring hoses and pipelines: In vessel and machine weighing systems, they can create significant force shunts.
- Mounting the load cell under angular stress: Additional transverse forces and moments impair zero return and repeatability.
- Electrically re-zeroing and ending the diagnosis: A mechanical cause or overload remains present.
- Considering overload only as excessive axial force: Impact, transverse force and moments can also mechanically damage a sensor.
- Ignoring temperature: Zero point and output can change during thermal stabilization.
- Never testing the sensor outside the system: This makes it impossible to determine whether the deviation originates from the sensor or the machine.
Suitable Force Transducer
For force measurements in test stands, materials testing machines and industrial test fixtures, the WIKA Type F2222 tension/compression force transducer is one suitable option.
The force transducer offers, among other things:
- measuring ranges from small forces up into the MN range,
- tension and compression force measurement,
- low installation height,
- high long-term stability,
- suitability for static and dynamic measuring tasks,
- IP66 ingress protection,
- optional force-introduction components,
- optionally integrated measuring amplifiers.
For reliable measurement, the force must be introduced centrally and as free as possible from transverse forces.
A flat mounting surface is also important during installation.
These mechanical conditions are particularly important for the troubleshooting discussed in this article.
A high-accuracy force transducer can only achieve its specified performance if the machine provides a reproducible force path.
Further load cells, tension/compression force transducers, shear beams and weighing electronics can be found under Force, Weighing, Speed and Torque Sensors at ICS Schneider.
Further information on mechanical installation can be found in the article Installing Load Cells Correctly: Avoiding Incorrect Force Introduction, Supports and Force Shunts.
For electrical installation, the article Extending Load Cell Cables: Correctly Accounting for 4- and 6-Wire Technology, Sense Lines and Voltage Drop is also helpful.
Conclusion
If a load cell does not return exactly to zero after unloading, there can be many different causes.
Creep describes the time-dependent change in the signal while a constant load remains applied.
After the load is removed, creep recovery can occur, during which the output signal gradually moves back toward the original zero point.
Hysteresis, on the other hand, is evident when the same force value produces different signals during ascending and descending loading.
In a real machine, additional mechanical influences must also be considered.
Stressed mountings, transverse forces, friction, stops, pipelines or even the connection cable can prevent the load cell from returning to exactly the same mechanical state after a load cycle.
A sudden and permanently large zero-point shift after a high load should, however, also raise the possibility of overload.
Simply performing another electrical zero adjustment is then not a sufficient test.
The most reliable diagnosis is achieved by performing a controlled loading sequence with a documented time axis and then comparing the installed sensor with a defined reference setup wherever possible.
The decisive question is therefore not only: “Does the load cell return to zero?”, but rather: “When does the deviation occur, how does it develop over time, and does it disappear once mechanical force introduction, cables, friction and temperature have been excluded as influencing factors?”
FAQ: Load Cell Does Not Return to Zero
Why does my load cell still show a value after unloading?
Possible causes include creep recovery, hysteresis, mechanical stress, friction, cable forces, temperature changes or a previous overload. The decisive factor is how the residual value changes over time and when the mechanical setup is modified.
Is a residual value after unloading automatically creep?
No. Creep describes the time-dependent signal change while a constant load remains applied. The time-dependent change after the load is removed is referred to as creep recovery.
What is hysteresis in a load cell?
Hysteresis is the difference between the output signals at the same load when that load point is approached once from a lower load and once from a higher load. Both ascending and descending load sequences are therefore required for testing.
How can I identify mechanical stress?
A typical indication is when the zero point changes significantly when mounting screws are loosened, the mechanics are moved or the load cell is removed. Poor repeatability can also indicate mechanical stress or friction.
Can the connection cable really change the measured value?
Yes. Especially with small force ranges, a cable that is under tension or routed too rigidly can exert an additional mechanical force on the sensor. The cable should therefore be routed without tension and with sufficient freedom of movement.
Can a vessel connection influence the zero point of a load cell?
Yes. Pipelines, hoses, electrical cables and other connections between the weighed structure and a fixed part of the plant create additional force paths. If these forces change during filling or emptying, zero-point and repeatability errors can occur.
How can I distinguish creep from temperature drift?
Creep is observed under constant load and as stable ambient conditions as possible. If the zero point changes in parallel with the sensor or ambient temperature even without a load change, a temperature effect is more likely.
What indicates an overloaded load cell?
A suddenly and significantly shifted zero point after an unusual loading event may be an indication. Sensitivity, linearity, hysteresis or repeatability may also change. A corresponding reference test is then advisable.
Can I simply re-zero an overloaded load cell?
A new zero adjustment can hide the displayed offset, but it does not repair possible plastic deformation of the spring element. After a suspected overload, the complete measurement characteristic should therefore be checked.
How do I correctly test zero return?
First, document the stabilized unloaded zero value. Then apply a defined load, hold it if necessary and remove it completely. Record the zero point immediately after unloading and again after defined waiting periods.
Why should I measure both ascending and descending load points?
This is the only way to determine whether the same force value is indicated differently depending on the loading direction. This difference is essential for assessing hysteresis.
When should the load cell be removed and tested separately?
If the mechanical cause cannot be clearly identified in the installed condition or if a sensor fault or overload is suspected. A defined test setup separates the sensor from influences such as frame stress, friction, pipe forces and stops.
When is calibration advisable?
Calibration is advisable if a permanent zero-point offset, changed sensitivity or abnormal hysteresis is detected, after a suspected overload, or according to the specified calibration interval. It should complement, not replace, the mechanical inspection of the installation.