Engineering Reference
Measurement & Instrumentation

Force Measurement

Understanding Load Cell Specifications

A practical explanation of full-scale output, combined error, non-linearity, hysteresis, repeatability, temperature effects, electrical characteristics and overload ratings.

A load cell converts an applied force into a measurable signal. Depending on its construction, that signal may be electrical, pneumatic, hydraulic or mechanical. In industrial weighing, strain-gauge load cells are widely used because a small deformation of an elastic metal element can be translated into a change in electrical resistance and then into a voltage signal. The underlying principle is straightforward, but the specification sheet used to describe a load cell contains a series of terms that are easy to misinterpret when considered separately.

Specifications such as full-scale output, non-linearity, hysteresis and repeatability do not describe interchangeable properties. Each term identifies a different source of measurement variation. A cell may produce a stable signal under repeated loading while still showing hysteresis between the ascending and descending portions of a calibration cycle. Another cell may have low non-linearity but exhibit a measurable zero shift as temperature changes. Understanding these distinctions is necessary when evaluating whether a sensor is suitable for a platform scale, vessel, hopper, testing machine, vehicle-weighing system or process-monitoring installation.

The stated values also need context. A percentage written on a data sheet is usually referenced to rated output or full scale, rather than to the instantaneous applied load. This distinction becomes important at the lower end of the measuring range, where an error expressed as a percentage of full scale can represent a much larger percentage of the actual load being measured.

Scope

This article explains common specification terms and how they relate to one another. It does not replace the acceptance limits, calibration instructions or installation requirements stated for a particular instrument.

Rated Capacity and Measuring Range

Rated capacity is the maximum force or load for which the load cell is designed to provide its specified measurement performance. It may be expressed in grams, kilograms, tonnes, newtons, kilonewtons or another force-related unit. Rated capacity should not be confused with the load that causes mechanical failure. A cell may tolerate a force above its rated capacity without breaking, but its accuracy specifications generally apply only within the stated measuring range.

Selecting capacity involves more than matching the expected process load. The dead weight of platforms, tanks, mounting structures, piping forces and temporary process disturbances may all contribute to the force applied to the sensing element. At the same time, selecting a capacity far above the normal operating load can reduce the usable signal associated with the quantity of interest.

For example, a system operating mostly between zero and one tonne may obtain less useful resolution from a ten-tonne load cell than from a correctly protected two-tonne cell. The higher-capacity sensor may provide greater mechanical margin, but only a small portion of its electrical output range would be used during ordinary operation. Capacity selection is therefore a balance among usable signal, expected loads, abnormal loads, mounting conditions and required measurement uncertainty.

Full-Scale Output and Sensitivity

Full-scale output, commonly abbreviated as FSO, is the electrical output produced when rated mechanical load is applied. For analog strain-gauge load cells, it is often expressed in millivolts per volt, written as mV/V. This format indicates that the output depends on both the applied force and the excitation voltage supplied to the bridge circuit.

A load cell rated at 2.0 mV/V produces approximately two millivolts of output for every volt of excitation when the cell is loaded to rated capacity. With 10 V excitation, the nominal full-scale signal is therefore 20 mV. At half of rated capacity, an ideal linear cell would produce approximately 10 mV.

Nominal full-scale signal = sensitivity × excitation voltage

FSO describes sensitivity rather than accuracy. A sensor with a 3 mV/V output is not necessarily more accurate than one with a 2 mV/V output. It simply produces a larger nominal electrical signal for the same excitation voltage at rated load. A larger signal can make instrumentation design easier, but accuracy still depends on factors such as non-linearity, hysteresis, repeatability, temperature effects, electrical noise, installation and calibration.

The actual rated output may also differ slightly from its nominal value. A data sheet may state a nominal sensitivity together with a tolerance. The indicator, amplifier or controller must therefore be calibrated using the real installed output rather than assuming that every sensor produces exactly the nominal figure.

Combined Error

Combined error is a consolidated performance value representing the maximum deviation from an ideal reference relationship during a defined loading and unloading procedure. It is commonly expressed as a percentage of full-scale output. The exact components included in combined error can depend on the test method and the terminology used by the data-sheet publisher, so the accompanying definition should always be reviewed.

In many cases, the value incorporates the influence of non-linearity and hysteresis observed over an increasing and decreasing load cycle. It is useful as a broad indicator because it summarizes several deviations in a single number. However, that convenience also limits its diagnostic value. Combined error alone does not reveal whether the dominant contribution comes from curvature in the calibration response, a difference between loading directions or another included characteristic.

When two cells have similar combined-error values but are intended for different operating cycles, the separate specifications remain relevant. A filling process that always approaches a target from below may be influenced differently from a batching system that repeatedly loads and unloads through the same points. The combined figure is therefore most useful when read alongside the individual terms.

Non-Linearity

Non-linearity describes the maximum deviation of the calibration curve from a specified straight reference line. It is generally determined during increasing load and expressed as a percentage of full-scale output. In an ideal load cell, the output would rise in exact proportion to the applied force. A 25% load would produce 25% of full-scale signal, a 50% load would produce 50%, and so forth.

Real elastic elements, strain gauges, bonding layers and bridge circuits do not behave with perfect mathematical linearity. Their collective response may form a slightly curved calibration line. Non-linearity quantifies the largest difference between that measured curve and the selected straight-line reference.

The reference line itself matters. Depending on the stated method, it may be drawn between zero output and the output at rated capacity, or calculated as a best-fit line through a series of calibration points. Values established using different reference methods should not be compared without checking how they were derived.

Non-linearity can be especially important when one calibration factor is expected to represent the entire measuring range. Digital systems may compensate for repeatable curvature by using a multi-point calibration table or polynomial correction. Such compensation can reduce the displayed error, but it does not remove the need to control installation, temperature, creep, hysteresis and other independent effects.

Non-linearity concerns the shape of the calibration curve; hysteresis concerns the direction from which a load point is approached.

Hysteresis

Hysteresis is the maximum difference between output signals recorded at the same applied load when that load is reached from opposite directions. One reading is obtained while the force is increasing from zero, and the other is obtained while the force is decreasing from a higher load, often rated capacity.

Consider a calibration sequence in which a cell is loaded from zero to 25%, 50%, 75% and 100% of capacity, then unloaded through the same points. At the 50% point, the output recorded during loading may differ slightly from the output recorded during unloading. The difference is the hysteresis at that point. The largest observed difference over the stated test cycle is normally reported as the hysteresis specification.

Hysteresis can arise from the mechanical and material behavior of the elastic element, strain-gauge adhesive, protective compounds and mounting interfaces. Friction, side forces, binding, flexible piping and structural movement outside the load cell can also create system-level behavior that resembles sensor hysteresis.

The practical importance depends on the application. A process that always measures rising loads may be less affected than a system that must produce the same indication while alternately loading and unloading. Hysteresis is therefore not merely a static accuracy figure; it describes the influence of load history on the indicated result.

Repeatability

Repeatability is the maximum difference among output measurements obtained when the same load is applied repeatedly under nominally identical conditions. It is usually expressed as a percentage of rated or full-scale output. A repeatability test attempts to hold the relevant variables constant, including loading point, mounting condition, excitation, temperature, stabilization time and loading procedure.

Repeatability differs from absolute accuracy. A sensor may produce nearly identical readings each time and therefore have good repeatability, while all of those readings remain offset from the accepted reference value. Conversely, a system that is correctly calibrated on average may still show an undesirable spread among repeated readings.

Repeatability is particularly relevant to sorting, check weighing, dosing and process-control tasks in which the same operation is performed many times. It helps describe the consistency of the measurement system. The observed repeatability of an installed scale, however, includes more than the load cell. Mechanical vibration, product movement, electrical interference, resolution, digital filtering and the repeatability of the applied test load can all influence the result.

Accuracy

The closeness of a measurement to an accepted reference value under specified conditions.

Repeatability

The closeness of repeated measurements to one another when the test conditions are kept nominally unchanged.

Resolution

The smallest change in input that can be distinguished by the complete measurement system.

Linearity

The degree to which output follows the selected straight-line relationship across the measuring range.

Zero Balance and Zero Offset

Zero balance is the load-cell output measured with rated excitation applied and no intended mechanical load on the sensing element. It is often expressed as a percentage of full-scale output. An actual strain-gauge bridge rarely produces exactly zero differential output in an unloaded state. Small differences in gauge resistance, bonding, machining and residual stress create an initial offset.

Instrumentation normally compensates for this offset during zeroing or calibration, provided the value remains within the acceptable input range. A large zero offset can reduce the available signal range or indicate that the cell is carrying an unintended preload. It may also result from mechanical damage, wiring problems, moisture ingress or permanent deformation following overload.

In installed systems, “zero” includes the mechanical condition of the entire structure. The dead weight of a platform or empty vessel may be intentionally tared out, while piping, restraints, thermal expansion and product accumulation may cause the apparent zero to change. Distinguishing electrical zero balance from system zero is therefore important during troubleshooting.

Temperature Specifications

Temperature affects both the mechanical element and the electrical bridge. Data sheets commonly distinguish between a compensated temperature range and an operating temperature range. These terms do not have the same meaning.

The compensated temperature range is the interval over which the cell is adjusted or characterized so that specified limits for zero balance and rated output can be maintained. The operating temperature range is the wider interval within which the cell can function without permanent adverse effects. A cell may remain mechanically and electrically operational near the limits of the operating range while no longer meeting the tighter performance values stated for the compensated range.

Temperature Effect on Zero

Temperature effect on zero describes the change in no-load output caused by a change in temperature. It is usually stated as a percentage of full-scale output per degree or over a specified temperature interval. In a vessel-weighing installation, this effect may appear as a slowly changing displayed weight even though the actual contents remain unchanged.

Temperature Effect on Output

Temperature effect on output describes the change in sensitivity or span caused by temperature. If sensitivity changes, the same mechanical load may produce a slightly different electrical output at different temperatures. Zero compensation alone cannot correct this effect because it alters the slope of the load-versus-output relationship rather than merely shifting the starting point.

Temperature gradients may also matter. A cell that is uniformly stabilized at one temperature can behave differently from a cell whose body is heated on one side by sunlight, process material or nearby equipment. Calibration and verification procedures should therefore allow sufficient time for thermal stabilization when temperature-related uncertainty is significant.

Input Resistance, Output Resistance and Excitation

A conventional strain-gauge load cell commonly contains four resistive gauges arranged as a Wheatstone bridge. Input resistance is measured across the excitation leads, while output resistance is measured across the signal leads under the stated unloaded conditions. Common bridge resistance values include approximately 350 ohms, although other values are used.

Resistance values help determine the electrical load placed on the excitation supply. When several cells are connected in parallel through a junction box, their combined input resistance is lower than the resistance of one cell. The indicator must be capable of supplying the resulting bridge current without excessive voltage drop or instability.

Recommended excitation is the voltage at which the transducer is intended to operate while meeting its specifications. Applying more voltage increases the millivolt output, but it also increases power dissipation and self-heating in the bridge. Exceeding the stated limit may cause drift, measurement error or permanent damage. A lower excitation voltage reduces signal amplitude and may require instrumentation with lower noise and greater gain.

Insulation Resistance

Insulation resistance is measured between electrically isolated parts of the load cell, such as the bridge circuit, metal transducer body and cable shield. A high insulation resistance indicates that only a very small leakage current can pass along these unintended paths under the specified test voltage and environmental conditions.

Reduced insulation resistance may result from moisture, contamination, damaged cable insulation or deterioration of internal sealing. Leakage paths can create zero shifts, unstable readings and sensitivity to grounding conditions. Because moisture-related faults may vary with humidity and temperature, a sensor can appear satisfactory during one test and unstable under different environmental conditions.

Cable Length and Sense Conductors

Cable length is part of the electrical measurement system. Resistance in the excitation conductors causes a voltage drop, meaning the bridge may receive slightly less voltage than the instrument supplies. Changes in cable temperature can change that resistance and therefore alter the effective excitation at the cell.

Four-wire load cells provide two excitation conductors and two signal conductors. Six-wire cells add a pair of sense conductors connected near the bridge. Compatible instrumentation uses the sense voltage to compensate for excitation losses in the cable. This arrangement is useful where cable runs are long or where conductor resistance changes materially with temperature.

Shortening or extending a calibrated cable can alter the electrical conditions assumed in the original calibration, particularly with a four-wire design. Cable changes should therefore follow the instructions associated with the sensor and instrumentation rather than being treated as a purely mechanical installation detail.

Safe Overload and Ultimate Overload

Safe overload and ultimate overload describe different limits. Safe overload is the maximum load that can be applied without causing a permanent change beyond the stated performance requirements. Ultimate overload is the maximum load the structure can withstand without mechanical failure under the defined test conditions.

A load between these two limits may not cause immediate fracture, but it may permanently alter zero balance, sensitivity or linearity. The absence of visible damage does not establish that measurement performance has been preserved. After a significant overload event, zero output and calibration should be checked before the cell is returned to measurement service.

Overload ratings also do not account automatically for every real installation condition. Side load, bending moment, torsion, impact and uneven loading can stress the sensing element differently from the axial force used to establish the nominal capacity. Mechanical stops and mounting assemblies must be designed so that they protect the sensor without creating friction, binding or alternate load paths during normal measurement.

Important distinction

Safe overload is a performance-preservation limit. Ultimate overload is a structural-survival limit. Neither should be interpreted as an alternative normal operating capacity.

Spring-Element Material

The spring element is the structural part of the load cell that deforms elastically under force. Common materials include aluminum, alloy steel and stainless steel. Material selection influences rated capacity, stiffness, fatigue behavior, corrosion resistance, environmental sealing strategy and the mechanical dimensions required to obtain a measurable strain.

Material alone does not define environmental suitability. A stainless-steel body may provide useful corrosion resistance, but cable entry, welded seals, connectors, protective compounds and installation hardware remain part of the complete environmental boundary. Similarly, an aluminum cell may perform reliably in a protected platform-scale application even though it would not be selected for direct exposure to aggressive washdown chemicals.

The published material description should therefore be read together with the enclosure rating, sealing method, compensated temperature range and intended installation environment.

How Specifications Interact

Individual specifications are useful, but a load cell functions as a system. A low non-linearity value cannot compensate for a mounting structure that diverts part of the force around the sensor. High repeatability does not eliminate an incorrect calibration factor. A generous overload rating does not prevent measurement errors caused by side loading. Similarly, a stable bridge output at room temperature does not establish performance across a changing outdoor temperature range.

The following table summarizes the principal question addressed by each common specification.

Specification Primary Question Typical Reference
Full-scale output How much electrical signal is produced at rated mechanical load? mV/V at rated capacity
Combined error What is the maximum consolidated deviation during the stated test cycle? Percent of full-scale output
Non-linearity How far does the calibration response depart from the specified straight reference line? Percent of full-scale output
Hysteresis Does the output at the same load depend on whether the load is increasing or decreasing? Percent of full-scale output
Repeatability How closely do repeated measurements agree under nominally identical conditions? Percent of rated output
Zero balance What output is present with excitation applied and no intended load? Percent of full-scale output
Temperature effect on zero How much does the no-load output change with temperature? Percent of full scale per degree or range
Temperature effect on output How much does sensitivity change with temperature? Percent of full scale per degree or range
Safe overload What temporary load can be applied without a permanent performance change? Percent of rated capacity
Ultimate overload What load can be withstood before structural failure under the stated conditions? Percent of rated capacity

A Systematic Method for Reading a Data Sheet

A data sheet can be reviewed in a consistent order to reduce the risk of focusing on one attractive number while overlooking a limiting condition.

  1. Confirm the force mode. Determine whether the cell is intended for compression, tension, shear, bending or a defined combination of forces.
  2. Establish the real load envelope. Include normal process load, dead load, temporary overload, dynamic effects and forces introduced by the mounting structure.
  3. Review output and instrumentation compatibility. Check rated output, bridge resistance, excitation voltage, wiring configuration and the input range of the indicator.
  4. Separate the accuracy terms. Read non-linearity, hysteresis, repeatability, zero balance and combined error as distinct characteristics.
  5. Check environmental limits. Compare compensated and operating temperature ranges, material, sealing, insulation resistance and cable requirements with the installation conditions.
  6. Examine mechanical protection. Verify safe overload, ultimate overload and any restrictions on side load, shock, bending moment or mounting orientation.
  7. Review calibration requirements. Identify the required calibration procedure, test points, traceability, verification interval and stabilization conditions.

Specifications and Calibration

Specifications describe expected performance under stated conditions, while calibration establishes the relationship between applied reference loads and the output of an individual measurement system. Over time, mechanical stress, aging, environmental exposure, cable changes, misalignment and overload can alter that relationship. Periodic verification is therefore used to determine whether the installed system remains within its required limits.

A multi-point calibration is more informative than a single full-load check because it can reveal non-linearity and direction-dependent behavior. A typical procedure applies a sequence of increasing reference loads across the measuring range and may then repeat the points while unloading. Repeating the sequence helps assess consistency and distinguish a stable response from random variation.

The number and location of calibration points should reflect the required method and operating range. Five approximately distributed loading steps may provide a practical view of the response for many industrial checks, but the applicable procedure, uncertainty requirement and governing standard take precedence. Systems used near the lower end of a high-capacity range may require additional points in that region.

Calibration interval is not determined by time alone. Measurement importance, frequency of use, loading severity, environmental exposure, history of drift, legal requirements and the consequences of an incorrect result all affect the appropriate schedule. A system with a stable documented history may justify a different interval from one exposed to frequent impact, washdown, temperature cycling or structural modification.

Illustrative Interpretation

Consider an analog load cell with a rated capacity of 10 tonnes, a nominal output of 2 mV/V and an excitation supply of 10 V. Its nominal full-scale signal is 20 mV. If the stated non-linearity is 0.03% of full scale, that figure corresponds to an output deviation equivalent to 0.003 tonne, or 3 kilograms, when expressed against the ten-tonne full-scale range.

This does not mean every measurement will be accurate within three kilograms. Hysteresis, repeatability, temperature effects, calibration uncertainty, indicator error, mechanical installation and resolution must also be considered. At an applied load of 100 kilograms, a three-kilogram full-scale contribution would represent 3% of the measured quantity. This illustrates why a high-capacity cell may be unsuitable when the primary requirement is precision at very small loads.

The example also shows why percentages must be traced to their reference basis. A value stated as a percentage of full-scale output has a different implication from a percentage of the current reading. Without that distinction, two numerical specifications can appear comparable even though they represent different quantities.

Load-cell specifications form a technical vocabulary for describing sensitivity, consistency, calibration shape, temperature behavior, electrical compatibility and mechanical limits. Full-scale output states how much nominal signal is available. Non-linearity describes departure from a reference line. Hysteresis identifies the influence of loading direction. Repeatability describes the agreement among repeated measurements. Zero and temperature specifications address changes that can occur even when the applied force has not intentionally changed.

Overload, resistance, excitation, insulation, cable and material specifications address other parts of the measurement problem. None should be interpreted in isolation. The performance of an installed weighing system depends on the interaction of the sensing element, mounting structure, instrumentation, environment, calibration method and operating procedure.

A careful evaluation therefore begins by defining the actual measurement task and then reading each specification according to its stated reference conditions. This approach makes it possible to compare devices consistently, identify the likely sources of uncertainty and design calibration and verification procedures that correspond to the way the system is used.

Reference Framework

Commonly cited technical frameworks for force-measurement verification include ASTM E4, concerning force verification of testing machines, and ASTM E74, concerning calibration of force-measuring instruments used to verify testing-machine force indication.

Requirements for a specific legal-for-trade, laboratory, industrial or safety-related installation should be taken from the standards, regulations, calibration procedures and equipment documentation applicable to that installation.