Load Cells

What is a Load Cell

A load cell is a transducer that converts a force, most often the weight of an object, into a proportional electrical signal. It is the heart of nearly every industrial weighing system, from a 200 gram laboratory balance to a 120 tonne truck scale, and it is equally central to force test rigs, crane safety overload protection, and process tank batching. Unlike a mechanical spring scale, a load cell delivers a continuous, repeatable, and traceable electrical output that a controller can log, totalize, and act on, which is what makes automated weighing and force control possible.

Functionally a strain-gauge load cell has three parts. First, the spring element or flexure: a precisely machined block of high-strength steel, stainless steel (typically 17-4 PH), or aluminum alloy, shaped so that the applied load concentrates strain at a known, controlled location. Second, the strain gauges: thin metallic-foil resistors bonded to that location, usually four of them wired as a full Wheatstone bridge so that two gauges stretch in tension while two compress, doubling sensitivity and cancelling common-mode temperature effects. Third, the protective package: potting, a welded cover, or a hermetically sealed metal bellows that keeps moisture off the gauges, plus a cable carrying excitation in and signal out.

The physics rests on a chain discovered across the nineteenth and twentieth centuries. In 1843 Charles Wheatstone popularized the bridge circuit that measures tiny resistance changes. In 1856 Lord Kelvin observed that a metal conductor changes resistance when strained. In 1938 Edward Simmons at Caltech and Arthur Ruge at MIT independently invented the bonded wire strain gauge, the component that made practical load cells possible; the bonded foil gauge that dominates today followed around 1952 from Peter Jackson at Saunders-Roe. Through the 1950s and 1960s the strain-gauge load cell displaced mechanical lever scales in industry, and by the 1990s the digital load cell, with an analog-to-digital converter and microcontroller built into the housing, began replacing analog summing for large truck scales and silo systems.

The application scale spans roughly nine orders of magnitude of force. At the low end, micro force sensors resolve grams in medical infusion pumps and benchtop materials testing. In the middle band sit the bulk of industrial cells: platform scales, conveyor and check weighers, hopper and tank batching from a few kilograms to tens of tonnes. At the high end, column and canister cells under truck scales, rail weighbridges, and rolling-mill stands carry hundreds of tonnes per support. No single cell covers this whole span; selection is the act of matching a force range to a spring geometry, an accuracy class, and an environmental package.

Four engineering metrics dominate load cell quality across that whole range: combined error (the worst-case envelope of non-linearity, hysteresis, and repeatability), creep and creep recovery under sustained load, temperature effect on zero and on span, and the overload margin before permanent damage. These four, not the headline capacity, determine whether a cell holds its calibration in service and what it actually costs to own over a five to ten year production life.

Load Cell Types by Element Shape

Load cells are most usefully classified by the geometry of the spring element, because the shape decides how force enters the cell, how much side load and off-center load it tolerates, and which capacity band it serves. The five mainstream families below cover the great majority of industrial weighing and force work. Picking the wrong shape, for example mounting an S-type where a shear beam belongs, is a common first mistake that shows up as poor repeatability and sensitivity to mounting alignment.

Bending beam and single-point cells sense the bending strain in a cantilever loaded at its free end. The single-point variant is engineered with a special gauge layout and parallel-beam geometry so the reading stays correct regardless of where the load sits on the platform, which is why it dominates retail and bench platform scales up to roughly 600 by 600 mm. Capacities run from a few hundred grams to about one tonne. The HBM Z6, in production since 1972 and available from 5 kg to 1 t in stainless steel with hermetic bellows sealing, is the textbook example of this family.

Shear beam cells place the gauges at the neutral axis of a beam where shear strain, not bending strain, peaks. Shear sensing is inherently less sensitive to where along the beam the load is applied, which makes single-ended shear beams the workhorse of floor scales, conveyor weighers, and tank-leg weighing modules from about 25 kg to 10 t. The double-ended shear beam, supported at both ends and loaded in the center, scales the same principle up: it is the standard cell under medium and heavy truck scales and large hopper systems, commonly from 5 t to 200 t per cell.

S-type cells, named for the S-shaped flexure, are the only common family that measures both tension and compression equally well. They thread inline between two fixtures, so they suit suspended hoppers, hanging scales, crane lines, and material-testing load strings. Because they are loaded through a single axis, they demand good alignment and rod-end bearings to avoid side-load error. Column and canister cells compress a solid or hollow steel column and carry the heaviest loads, from a few tonnes to several hundred tonnes per cell, under truck weighbridges, silos, and rolling-mill stands. Their high overload tolerance and self-centering rocker mounts are the reason they survive the impact and thermal cycling of outdoor heavy weighing.

Sensing Principles

Element shape decides how force enters the cell; the sensing principle decides how deformation becomes a signal. Four principles cover essentially all load cells in service, and each owns a different combination of accuracy, capacity, cost, and environment. Strain gauge dominates by volume, but hydraulic, pneumatic, and capacitive cells survive in niches where no electrical excitation, intrinsic safety, or specific physics is required. The table compares the four on the metrics that drive selection.

Foil strain gauge is the dominant principle. Four metallic-foil gauges, typically constantan or Karma alloy, are bonded to the spring element in a full Wheatstone bridge. With excitation applied across one diagonal, load-induced resistance change unbalances the bridge and produces a millivolt-level output across the other diagonal, normalized to 2 mV/V or 3 mV/V at rated capacity. The full bridge is what gives strain-gauge cells their accuracy, from 0.25 percent of full scale on industrial cells down to 0.03 percent and better on precision and reference cells, because the opposing gauges cancel temperature drift and double the signal. The trade-off is that the gauges must be kept dry: moisture across the bridge is the leading field failure.

Hydraulic load cells need no electrical power. The load presses a piston onto an oil-filled diaphragm, and the resulting fluid pressure, read by a Bourdon gauge or pressure transmitter, is proportional to force. Because there is no electrical excitation at the cell, hydraulic cells suit explosion-hazard zones, lightning-prone outdoor sites, and remote tanks where running power and signal cable is impractical. They handle very high capacities, into the thousands of tonnes, but accuracy is modest, typically 0.25 to 1 percent of full scale, and the oil system adds temperature sensitivity and maintenance.

Pneumatic load cells balance the load with regulated air pressure through a nozzle-and-flapper feedback loop; the balancing air pressure is the measurement. They contain no electrical parts and are inherently safe in explosive and dust atmospheres, and being air-based they are clean and tolerant of temperature swings, which suits food and pharmaceutical lines. They are best at light to medium capacities and need a clean dry air supply. Capacitive and other niche principles, including vibrating-wire and silicon strain types, appear in OEM products, in cells that must survive extreme overload, and in cost-driven designs, but for general industrial weighing the foil strain gauge remains the default.

Accuracy Classes and Standards

Load cell accuracy is governed by two separate worlds of standards, and confusing them is a frequent procurement error. Legal-for-trade weighing follows OIML R60 (and the North American NTEP / Handbook 44 equivalent), which classifies cells by how many verification intervals they support. Force calibration and materials testing follow ISO 376 and ASTM E74, which classify force-proving instruments by their uncertainty over the loading range. A cell certified C3 to OIML R60 is not automatically a class 1 force transducer under ISO 376; the test regimes are different.

OIML R60 rates a cell by accuracy class (A, B, C, D) and the maximum number of verification intervals v it can support. Class A covers 50,000 intervals and up, class B from 5,000 to 100,000, class C from 500 to 10,000, and class D from 100 to 1,000. So a C3 cell supports 3,000 intervals and a C6 cell supports 6,000. The maximum permissible error is apportioned to the cell with a default factor p of 0.7 and applied as a single envelope covering linearity, hysteresis, temperature, and creep together. The table below shows the class C error band that C3 and C6 cells must meet.

The same R60 envelope sets the creep and temperature limits. For creep, the cell holds 90 to 100 percent of rated capacity for 30 minutes: the reading must not change by more than 0.15 times the maximum permissible error between the 20 and 30 minute marks, and the 30-minute reading must stay within 0.7 times the maximum permissible error of the start. For temperature, class A allows a dead-load output shift of p times v per 2 degrees C, while classes B, C, and D allow p times v per 5 degrees C, with a default operating range of minus 10 to plus 40 degrees C. This is why C6 is harder to achieve than C3: the same combined-error budget is divided across twice as many intervals.

ISO 376 and ASTM E74 govern force-proving instruments used to calibrate testing machines. ISO 376 sorts transducers into classes 00, 0.5, 1, and 2, from best to worst, based on reproducibility, repeatability, interpolation, zero, and reversibility errors over the loading range. The first usable calibration point cannot be smaller than 4,000 times the resolution for class 00, 2,000 for class 0.5, 1,000 for class 1, and 500 for class 2, so the better the class, the higher the minimum force at which the cell is trustworthy. Class 00 typically demands deadweight primary standards that achieve better than 0.003 percent of applied force. ASTM E74 is the parallel North American standard for the same purpose, and EN 10002 / ISO 7500 reference these instruments when verifying the machines themselves.

For procurement, the rule is to specify the standard that matches the duty. A vendor scale sold by weight to the public must use an OIML R60 or NTEP class C cell with a matching certified indicator. A laboratory force frame must use an ISO 376 or ASTM E74 transfer transducer of the required class. A simple process tank that batches material internally needs neither legal certification, only a stated combined error fit for the recipe tolerance, which is usually the most economical path.

Key Specification Parameters

A load cell datasheet may list 15 to 30 lines, but a handful drive the selection and the calibration math. The values below are typical industrial figures: precision and reference cells do better, and the cheapest OEM cells do worse. Read them as a band, and always confirm against the specific datasheet before committing.

Rated capacity (Emax) is the maximum load the cell is designed to measure, stated in kg, t, N, or lb. Rated output (full-scale output) is the bridge signal at rated capacity, expressed per volt of excitation: 2 mV/V or 3 mV/V are the near-universal defaults, with a typical unit-to-unit tolerance around 0.25 percent. A 3 mV/V cell at 10 V excitation therefore gives 30 mV at full load. Zero balance is the unloaded output offset, often specified at plus or minus 1 percent of rated output (around plus or minus 0.02 mV/V), which the indicator nulls out during setup.

Combined error is the single most important accuracy figure for industrial cells: the worst-case envelope of non-linearity, hysteresis, and non-repeatability, typically 0.02 to 0.25 percent of rated output. Manufacturers also list the components separately, with non-linearity and hysteresis often near 0.02 percent and repeatability near 0.01 percent of rated output on good cells. Creep is the output drift under constant rated load, commonly specified around 0.025 to 0.03 percent of rated output over 30 minutes, often paired with a creep-recovery figure.

Temperature effect is given as two numbers: the effect on zero (zero drift) and the effect on output or span (sensitivity drift), each typically around 0.001 to 0.004 percent of rated output per degree C, valid only inside the stated compensated temperature range, commonly minus 10 to plus 40 degrees C, inside a wider operating range such as minus 30 to plus 70 degrees C. The compensated range is where accuracy is guaranteed; outside it the cell still works but the numbers no longer hold. The table below gathers the parameters that should appear in any serious RFQ.

Bridge resistance is the electrical input and output impedance of the Wheatstone bridge; 350 ohm is the most common value, with 700 ohm and 1,000 ohm options that let more cells share one indicator without overloading its excitation supply. Excitation voltage is usually specified as a recommended value of 10 V with a maximum around 15 V; higher excitation lifts the signal but heats the gauges and can add drift. Safe overload, typically 150 percent of rated capacity, is the load the cell tolerates without permanent change; ultimate or breaking overload, often 300 percent, is the point of mechanical failure. Insulation resistance, usually specified above 2,000 megohm at 50 V DC, and the IP ingress rating together predict how the cell survives moisture, the dominant failure mode.

Selection Decision Factors

To turn the preceding five chapters into a specific model, work through the ordered sequence below. Most selection mistakes are not a single wrong number but a premature decision at the wrong level, for example fixing on a brand before settling the capacity and accuracy class. These eight steps double as a fixed RFQ template.

1. Force range and number of supports: Total the maximum load including dead weight, divide by the number of cells, and size each cell so the working load lands within 30 to 70 percent of rated capacity. Add margin for uneven distribution; a four-leg tank can load one corner far more than a quarter of the total during filling.
2. Element shape: Single-point or bending beam for small off-center platforms, single-ended shear beam for floor scales and tank legs, double-ended shear or column for truck and silo capacities, S-type for inline tension or suspended hoppers. The shape decides side-load tolerance and mounting.
3. Accuracy class and standard: Decide whether the duty is legal-for-trade (OIML R60 / NTEP class C, choose C3 or C6 by required divisions), force calibration (ISO 376 / ASTM E74 class), or internal process (a stated combined error fit for recipe tolerance). Each tier up in accuracy roughly raises the price.
4. Spring material and environment: Aluminum for light low-cost platforms, alloy steel (often nickel-plated) for dry industrial use, stainless 17-4 PH for wet, corrosive, food, or outdoor duty. Match the material to media and washdown chemistry.
5. Sealing and ingress protection: IP65 or IP66 potted cells indoors, hermetically sealed IP68 for outdoor, tank-leg, and standing-water service, IP69K stainless for high-pressure washdown lines. Moisture ingress is the leading field failure, so do not under-rate this.
6. Electrical interface: Analog mV/V into a junction box and weighing indicator, or a digital load cell with on-board ADC speaking RS-485 Modbus, CANbus, PROFINET, or EtherCAT for large multi-cell systems. Confirm bridge resistance is compatible with the indicator excitation supply.
7. Certifications and safety: Legal metrology approval (OIML / NTEP), hazardous-area approval (ATEX / IECEx) for explosive zones, EMC compliance, and where relevant functional-safety rating for crane overload and lifting protection.
8. Total cost of ownership (TCO): Purchase price plus mounting hardware, junction box, indicator, calibration, and the cost of downtime when a cell fails. A cheaper cell that drifts or lets moisture in often costs more within three years than a sealed stainless cell bought up front.

One last dimension that buyers routinely overlook is serviceability and mounting ecosystem: availability of matched mounting kits and rocker assemblies, junction boxes with corner trimming or digital calibration, replacement cells with the same rated output so a single failure does not force a full recalibration, and local calibration service traceable to national standards. These details look irrelevant at purchase but govern repair response and uptime over a five to ten year service life. HBM, Flintec, Vishay BLH Nobel, Mettler Toledo, Interface, and Zemic all maintain mounting-hardware ranges and calibration support, which makes them safe defaults for large multi-cell projects; Zemic, Keli, and Bongshin offer OIML and NTEP approved cells at lower cost for non-critical loops.