A load cell measures force by allowing itself to be deformed by it. Strain gauges bonded to a machined flexure register that deformation, a Wheatstone bridge turns the resulting resistance change into millivolts, and a traceable calibration converts millivolts into pounds.
Which means every load cell is a compromise between two things it cannot both have. It must yield enough for the gauges to sense — and yielding is exactly what you do not want, because the sensor is now a spring you have installed in someone's machine. How that compromise is settled is the single most informative thing about a load cell design, and a shear web built for stiffness settles it firmly at one end.
Deflection of one thousandth of an inch at full load is roughly half what conventional designs of the same capacity give, and everything else about this class of sensor follows from that one number.
Low deflection means low stress, and low stress is what buys fatigue life. This is the connection almost nobody makes. Manufacturers of fatigue-rated sensors describe the recipe openly: design stresses at about half those of standard models, deflection at about half, and internal stress concentrations — sharp corners, edges — polished away so there is nowhere for a crack to start. A stiff sensor and a long-lived one are the same sensor described from two directions.
Stiffness sets how fast the sensor can answer. A structure that barely moves under load returns to rest quickly when disturbed, so it can follow rapidly changing forces that a softer design would smear.
Stiffness resists what you did not intend to measure. Bending moments and side loads deflect a rigid structure very little, and a shear web geometry is arranged so that axial force produces shear in the web while off-axis effects largely do not. That is why this shape is chosen where the loading is imperfect and the result still has to be right.
And a stiff sensor leaves the system it measures alone. Insert a compliant load cell into a machine and you have changed the machine — added travel, altered its natural behaviour, softened a stack that was designed to be rigid. One thousandth of an inch changes almost nothing.
Sensors built this way belong in aerospace structural testing, automotive durability and component evaluation, materials testing including fatigue work, precision industrial automation, stringent quality verification, research and prototype assessment, and specialised machinery where the measurement has to be right and the sensor has to survive.
Stiffness is not free, and the interesting part of this design is how the bill is paid.
The problem stiffness creates. Strain gauges respond to strain, and a structure that deflects less produces less of it. Everything gained in rigidity, fatigue life and extraneous load rejection is taken back as a smaller electrical signal — unless the design does something about it.
What a 700 ohm bridge is for. Most load cells use a 350 ohm bridge. Doubling that resistance halves the current the bridge draws at any given excitation, and published guidance on excitation notes that higher bridge resistances are less sensitive to excitation variation and allow transducers to be run at larger excitation voltages. That is precisely the point: this series accepts 10 to 20 VDC rather than a fixed 10, and because output is a ratio of what is supplied, running at the top of that range roughly doubles the signal — around forty millivolts at full load instead of twenty. The stiffness is recovered electrically rather than surrendered.
The obligation that comes with a range. Where excitation is a fixed number, matching it is automatic. Where you can choose, you have to choose correctly: the instrument must be told what it is supplying, and the sensor should be operated at the excitation its certificate was written at. This is not a nicety — ASTM E74 and ISO 376 both require that a substituted meter match the original excitation voltage, and laboratory measurements comparing 5 and 10 volts on a shear web cell found a consistent difference of about 0.01% across most applied forces. Small, but the same order as the specifications you are paying for.
Fatigue life, and why reversing the load costs you half of it. The published figures are 100 million cycles in one direction and 50 million fully reversed. That two-to-one gap is real metallurgy rather than caution: a cycle that swings from tension to compression puts the material through twice the stress range of one that returns to zero, so fully reversed duty is substantially more punishing. Industry guidance expresses the same relationship the other way round, noting that a fatigue-rated cell loaded in one direction only can be worked to roughly 133% of its bidirectional fatigue rating. If your test alternates, use the lower figure.
Load in at the middle, out around the bolts. Force arrives through a tapped centre hole and reacts through a ring of hold-down bolts — eight, twelve or sixteen depending on capacity. What screws into the centre therefore matters as much as the sensor: a load button for compression, and for in-line tension a matched tension base, which exists to supply the hardness, flatness, stiffness and alignment that an improvised fixture will not.
This is the most capable force sensor we build, and it is specified rather differently from the rest of the range — there are choices here that do not exist elsewhere. Our application engineers would rather work through them with you than have you discover one of them after delivery.
Establish first whether you need this at all. Be honest about it, because the price reflects genuine engineering. If your loading is steady, one-directional, well aligned and modestly demanding, a conventional design will serve you perfectly well for considerably less. What earns the premium is some combination of cyclic duty, imperfect load introduction, a requirement for very low deflection, or an accuracy specification that has to be defended.
Count your cycles before you choose a capacity. On fatigue work this matters more than the peak force does. Establish how many cycles the sensor will see over its life and whether the load reverses, then size against the fully reversed figure if it does. A sensor asked for more cycles than it was built for does not announce the problem; it drifts and eventually cracks.
Note that two different sensors share the 5,000 lb rating. One uses the compact body shared with the lower capacities — smaller, cheaper, and with a notably higher ringing frequency. The other uses the larger body shared with the 10,000 and 20,000 lb models, which brings a much bigger centre thread and more hold-down bolts, and therefore a more substantial load path and mounting. Neither is simply better. Tell us the duty and we will tell you which one it wants.
Order the tension base with the sensor if you are pulling. For in-line tension work this is not an accessory but part of the measurement, and it is sized to the capacity band rather than being universal. Sourcing an equivalent locally is a false economy: the point of the base is that its hardness, flatness and alignment are controlled, and those are exactly the properties a workshop fixture does not have.
Check your instrument against a 700 ohm bridge and a variable supply. Some conditioners assume 350 ohms, and some supply a fixed excitation. Neither is a problem if it is known about in advance and a nuisance if it is not. Confirm what your equipment provides, and set it to the excitation the certificate specifies.
Then decide the direction and the readout. The certificate supplied covers compression; add the tension option if that is the direction your results have to stand on. For the readout, an amplifier signal conditioner module supplies a voltage or current output, or a digital display adds a local reading with setpoints and logging — and on a sensor this good it is worth making sure the instrumentation is not the limiting element. Where units rotate through several rigs, Cal-Teds plug and play saves someone keying in constants and getting one wrong.
Give us the peak force, the cycle count, whether the load reverses, and what the sensor bolts to — that is enough for us to put a model and a fitting list in front of you. We keep the range on the shelf, and there is a discount for university and research purchases.
SWP Series Load Cell Applications.
The Transducer Techniques SWP Series low profile shear web load cells are specialized force sensors engineered for applications demanding ultra stiffness and resistance to extraneous bending and side load forces.
- Aerospace Testing: The aerospace industry relies on SWP Series load cells for structural testing of aircraft components. These load cells offer stiffness and precision, making them ideal for ensuring the safety and reliability of aerospace structures.
- Automotive Testing: Manufacturers use these load cells for quality control and vehicle component testing, including suspension systems and chassis evaluation.
- Materials Testing: SWP Series load cells are employed in laboratories for tensile, compressive, and fatigue testing, where stiffness and resistance to extraneous forces are valuable for accurate determination of material properties.
- Industrial Automation: These load cells are integrated into machinery for force monitoring and control in manufacturing and assembly processes where precision is critical.
- Quality Control: Industries with stringent requirements use these load cells to verify that products meet force and load specifications.
- Research and Development: Engineers evaluate performance, durability, and structural integrity during product development and prototyping.
- Custom Machinery: SWP Series load cells are integrated into specialized testing and manufacturing equipment requiring ultra-stiff, precise force measurement.
- Tension and Compression Force Measurement: These load cells accommodate both measurement directions, often paired with load buttons or compression-only accessories.
- In-Line Tension Applications: SWP Series load cells are combined with the tension base options (-TB) for optimal hardness, flatness, stiffness, and alignment.
The Transducer Techniques SWP Series shear web load cells are known for their stiffness, precision, and resistance to extraneous forces, making them valuable tools for force measurement applications where accuracy and reliability are paramount.
Frequently Asked Questions
What is a shear web load cell and why does the design matter?
A shear web takes the force arriving at the centre of a disc and carries it outward through a machined web to a mounting ring, with the gauges reading the shear in that web rather than the bending of a beam. Two things follow. It is extremely stiff, deflecting a thousandth of an inch at full load, which is roughly half what conventional designs give. And it is arranged so that axial force produces the strain being measured while bending moments and side loads largely do not — which is why it is the geometry chosen where load introduction cannot be made perfect and the answer still has to be right.
What capacities does the SWP Series cover and what do they cost?
Eight models across seven capacities: SWP-1K, SWP-2K, SWP-3K and SWP-5K-4 at $1,180; SWP-5K, SWP-10K and SWP-20K at $1,470; and SWP-50K at $2,700. The two 5,000 lb entries are not a misprint — they are different sensors of the same capacity built on different bodies, covered in a separate answer below. Prices are flat within each body size, so within a band there is nothing to save by specifying lower.
Why are there two 5,000 lb models, and which should I choose?
The SWP-5K-4 uses the compact 4.125" body shared with the 1,000 to 3,000 lb models: eight hold-down bolts on a 3.500" circle, a 5/8-18 UNF centre thread, $1,180, and a ringing frequency of 8,700 Hz. The SWP-5K uses the 6.000" body shared with the 10,000 and 20,000 lb models: twelve bolts on a 5.125" circle, a much larger 1 1/4-12 UNF centre thread, $1,470, and 6,500 Hz. So the smaller one is more compact, cheaper and faster; the larger one gives a substantially heavier load path and mounting for the same rated force. Which suits depends on the duty and the fixture — tell us both and we will advise.
How accurate is the SWP Series?
Nonlinearity, hysteresis and nonrepeatability are each 0.05% of rated output, with zero balance at 1.0%. That makes it the most accurate load cell we offer at these capacities — twice as good as our low profile and shear web alternatives on linearity and hysteresis, and five times better than the thru-hole and tension link series. On a product this good, the limiting factor is frequently the instrumentation or the load introduction rather than the sensor, which is worth bearing in mind when you specify the rest of the chain.
What is the SWP's fatigue life?
100 million cycles loaded in one direction, and 50 million fully reversed. That is a specification very few load cells publish at all, and the two-to-one difference reflects real metallurgy: a cycle swinging from tension to compression subjects the material to twice the stress range of one returning to zero. If your test alternates direction, plan against the 50 million figure. If it does not, you have the higher one. Either way, count the cycles your application will actually accumulate before you choose — on fatigue work that number governs the selection more than the peak force does.
Why is the bridge 700 ohms rather than 350?
A 700 ohm bridge draws half the current of a 350 ohm bridge at the same voltage, and higher bridge resistance is what makes it practical to run at a higher excitation. That is why this series accepts 10 to 20 VDC where most of our products specify a fixed 10. Since output is expressed per volt of excitation, working at the top of that range roughly doubles the signal — about forty millivolts at full load rather than twenty — which recovers the signal that a very stiff, low-deflection design would otherwise give up. Do check that your instrument is comfortable with a 700 ohm bridge, as some assume 350.
What excitation voltage should I use?
The one the calibration certificate specifies. Anywhere from 10 to 20 VDC is acceptable to the sensor, and higher gives you more signal, but the reading is only strictly comparable to the certificate at the voltage the certificate was written at. Force calibration standards take this seriously — both ASTM E74 and ISO 376 require a substituted instrument to match the original excitation — and published laboratory measurements comparing two excitation levels on a shear web cell found a consistent difference of about 0.01%. That is small, but it is the same order as the accuracy you are buying. Decide the voltage before calibration, then leave it alone.
What is the tension base option and do I need it?
The tension base is a matched fixture for in-line tension work, supplied in three versions sized to the capacity bands: TB0-SWP135 for the 1,000 lb through 5K-4 models at $220, TB1-SWP1020 for the 5,000 to 20,000 lb models at $285, and TB2-SWP50K for the 50,000 lb at $550. Its job is to give the sensor a mating surface with controlled hardness, flatness, stiffness and alignment. If you are pulling rather than pressing, we recommend it — the accuracy this sensor is capable of assumes the load arrives properly, and a fixture made locally is unlikely to control those four properties.
What is the natural ringing frequency of each model?
Published per model: 4,000 Hz on the SWP-1K, 5,700 Hz on the 2K, 7,300 Hz on the 3K, 8,700 Hz on the 5K-4, 6,500 Hz on the 5K, 7,400 Hz on the 10K, 8,500 Hz on the 20K and 10,000 Hz on the 50K. Note that it does not simply rise with capacity, because two body sizes overlap in the middle of the range — the compact 5K-4 rings faster than either the 10K or the 20K. Usable frequency content is a fraction of the ringing figure rather than all of it, so compare your test frequency against the row for the specific model you intend to buy.
What are the SWP's dimensions and mounting patterns?
Three body sizes. The 1,000 to 3,000 lb models and the 5K-4 are 4.125" in diameter with eight hold-down holes of 0.281" on a 3.500" bolt circle and a 5/8-18 UNF centre thread. The 5,000 to 20,000 lb models are 6.000" in diameter with twelve 0.406" holes on a 5.125" circle and a 1 1/4-12 UNF thread. The 50,000 lb is 8.000" in diameter with sixteen 0.531" holes on a 6.500" circle and a 1 3/4-12 UNF thread. Design the fixture around the specific model, since a plate drilled for one body will not accept another.
Questions From The Field
Is a stiffer load cell always the better choice?
No, and it is worth being clear about that. Stiffness costs money, and it costs signal — less deflection means less strain, which is why this design needs a higher-resistance bridge and a higher excitation to get back to a comfortable output. Where it earns its keep is cyclic duty, imperfect load introduction, applications that cannot tolerate the sensor adding travel to the load path, and fast events. Where the load is steady, well aligned and one-directional, a conventional design will do the job for considerably less and you should not feel obliged to buy stiffness you have no use for.
My test is fully reversed. How should that change what I order?
Two adjustments. Use the 50 million cycle figure rather than the 100 million one, and check it against the total your programme will accumulate rather than against a single run. Then look hard at the fixture, because a reversing load finds every joint that can move: threads, pins, clevises and anything merely tight rather than positively locked will take up and release on each reversal, and the sensor will report that as a disturbance around the crossing. Bidirectional duty is also the case where tension calibration is most likely to be worth having, since half your data is on that side of zero.
Can I press directly on the load cell face for compression?
You can, and it will cost you a good deal of what you paid for. Two flat surfaces are never quite parallel, so contact lands on an edge or a high spot and the force arrives eccentrically, varying with how the fixture happens to sit that day. A load button gives compression a defined, slightly domed contact that finds its own centre. On a sensor specified at 0.05% it is a false economy to introduce the load through an uncontrolled interface — ask us about compression accessories when you order.
My readings changed slightly after I increased the excitation voltage.
That is expected and it is not a fault. Output is a ratio of excitation, so a correctly configured instrument should scale automatically — but there is a small residual effect of the excitation level itself, measured in one published laboratory comparison at around 0.01% across most applied forces. First check the obvious thing: that the instrument knows what voltage it is now supplying, since a conditioner still scaled for the old figure will be out by the ratio of the two. Then settle on one excitation voltage, have the sensor calibrated at it, and stay there.
Do I need to do anything special about the mounting surface?
Yes, and on this sensor it matters more than on most, because the whole reaction passes through the bolt circle into whatever it is fixed to. The surface must be flat and stiff enough not to deform appreciably under load — a base that dishes even slightly distorts the flexure before any force is applied, and the sensor cannot distinguish that from load. Tighten the hold-down bolts evenly rather than working around the circle in order, and zero the system after mounting rather than before. Buying 0.05% and bolting it to a plate that flexes gets you neither.
How do I know if my instrument will work with this sensor?
Check three things. Whether it will drive a 700 ohm bridge, since some conditioners assume 350 and their excitation supply is sized accordingly. What excitation voltage it supplies and whether that is adjustable, given this sensor's 10 to 20 VDC range. And whether its own resolution and noise are good enough not to become the limiting element — there is little point buying a 0.05% sensor and reading it with something coarser. Tell us what you have and we will confirm the combination, or suggest something that suits.
Should I have the sensor calibrated with the tension base and adapters fitted?
Where the results have to withstand scrutiny, yes. Every fitting in the load path affects how force reaches the sensing element, and calibrating the assembly as a unit removes a variable that is otherwise very difficult to quantify afterwards. It also means the calibration describes the thing you actually use rather than a component of it. The corollary is that the assembly should then stay together — changing an adapter after calibration changes the measurement, and it is worth marking the parts so nobody separates them helpfully during a clean-down.
How often should an SWP be recalibrated, and what would tell me it needs it sooner?
Annually is the usual interval, tightened where the results carry weight or the duty is hard. Bring it forward for any suspected overload or impact, any change to the adapters, mounting or excitation voltage, and any move to a different fixture. On fatigue duty specifically, track accumulated cycles against the published figures as well as elapsed time — a sensor approaching its cycle count deserves checking regardless of how recently it was calibrated. The most useful early indicator is the unloaded reading: compare it against the zero balance on the certificate, and treat a growing offset as a reason to investigate.