How Does a Torque Wrench Work? Click, Beam, Digital

Some click wrenches do without the heavy coil spring. In a split-beam wrench, the load goes through a main beam and a catch holds a second beam alongside it. Olsa Tools' product manager Gus Lange explained it to EngineLabs: "As torque is applied, the force is transferred through the main beam, and the catch keeps the main beam and anchor beam together. Once the desired torque is reached, the catch releases, causing a loud click as the two beams separate."

Because the setting doesn't keep a big spring compressed in the drawer, Precision Instruments, which says it launched the split-beam click wrench market in 1974, says its C series "can be stored at any value." Stahlwille's MANOSKOP 730 uses a bending rod as its measuring element, and Stahlwille says it needs "no reset to '0'" either. The C series has a trade-off: it works clockwise only, apart from the C4D600F series, whose ratchet can be flipped. Our split-beam torque wrench guide covers current models.

A beam wrench is the simplest design. Norbar's Philip Brodey describes it: "The body of the wrench is designed to bend as load is applied to the handle while an indicating bar, routed at the head of the wrench, but not bending, moves relative to a scale." The indicating bar carries no load, so it stays straight and points to the torque as the main beam bends away beneath it.

A dial wrench develops the same idea, turning the deflection into the movement of a needle on a dial, usually with a slave pointer that stays at the highest reading. There's no spring to take a set, but you have to watch the scale as you pull. Our beam vs click vs digital guide compares the types in use and picks the best beam wrenches.

A digital wrench measures electronically instead of with a spring. "In electronic wrenches, measurement is by means of a strain gauge attached to a beam, which flexes as torque is applied," Brodey explains, and the signal is converted to a torque unit and shown on the display.

Because the reading is electronic, the wrench can warn you as you approach the target. Norbar's NorTronic beeps as the torque approaches the target, slowly at first and then faster, and vibrates when the target is reached. Wrenches such as GEARWRENCH's 85078 offer a peak mode, which holds the highest reading, and a tracking mode that shows the torque live. Many add a gyroscope to measure angle, for torque-plus-angle work such as cylinder head bolts, and some store readings for download. Our digital torque wrench guide compares current models.

Tightening a bolt stretches it by a tiny amount, and that stretch is what clamps the joint. But most of the torque you apply never becomes stretch. It's used up in friction, under the bolt head or nut and in the threads.

Sources split it slightly differently, but they agree on the scale: Bolt Science says friction usually takes 85–95% of the applied torque, and the commonly quoted split is about 50% under the head, 40% in the threads and 10% stretching the bolt. That's why oil, rust, plating or a dirty thread changes the clamp load you get from the same torque, and why torque is, in Portland Bolt's words, "a very indirect indication of tension." Our bolt torque chart shows how much the thread condition changes the torque to use.

Two different accuracies matter. The first is the wrench's own: how close the click or the reading comes to the true torque. ISO 6789-1:2017 sets the limits used to certify hand torque tools, and they depend on the type and on the tool's capacity.

US makers that rate click wrenches to ASME B107.300 typically quote ±4% clockwise, often ±6% counter-clockwise, from 20% to 100% of the scale. Below 20%, the allowance doesn't shrink with the setting: Snap-on explains that the older ANSI standard allows an error equal to 0.8% of the wrench's capacity there. On a 250 ft-lb wrench that's 2 ft-lb, or ±10% at a 20 ft-lb setting, which is why a wrench is best used in the middle of its range. Some makers rate tighter than the standards: Tohnichi rates its QL-MH click wrench at ±3%, and Norbar its NorTronic electronic wrench at ±1% of reading from 20% to 100% of capacity.

The second accuracy is the clamp load in the bolt, and it's much looser. NASA's Fastener Design Manual quotes the Industrial Fasteners Institute's figure of ±25% for setting bolt load with a torque wrench, against ±35% by feel. Most of that spread comes from friction, which varies from bolt to bolt: a good wrench removes one source of error; it can't remove friction.

Makers' manuals agree on most of this list. If a wrench has suffered any of it, check it before you trust it; our calibration guide shows how.

Torque wrenches appear in US patents from the 1930s. John H. Sharp filed for a "torque-measuring wrench" in 1931, granted in 1935, and Conrad Bahr and George Pfefferle filed theirs in 1935: a ratchet wrench for bolted pipe joints, assigned to S.R. Dresser Manufacturing. Sturtevant Richmont credits Walter P. Chrysler with inventing the beam torque wrench and licensing Paul Sturtevant to make it, and dates that to 1924.

You'll often read that Bahr invented the torque wrench in 1918, while working for New York City's water department. We found no primary source for that: the magazine article Wikipedia cites says only that he was working there in 1918.

The mechanism, accuracy and care guidance on this page comes from these documents.