Sprag One-Way Clutch

What a sprag clutch actually is

Most clutches are deliberate: a pedal, a lever, or a hydraulic piston decides when two shafts couple. A one-way clutch takes that decision away from the operator and hands it to the geometry. It couples automatically in one direction of relative rotation and decouples automatically in the other. The sprag clutch is the highest-performing way to do that.

Strip away the housing and a sprag clutch is almost embarrassingly simple: a smooth cylindrical inner race, a smooth cylindrical outer race, and an annular gap between them filled with a ring of little steel cams called sprags. There are no teeth anywhere. Each sprag is an asymmetric, roughly figure-eight (peanut) shaped block. The trick is that a sprag is taller across one diagonal than the other. Across its tall diagonal it is slightly bigger than the radial gap between the races; across its short diagonal it is slightly smaller. A light ribbon or garter spring keeps every sprag tipped so it always touches both races.

Now spin the inner race. Friction at the two contact points drags each sprag and tries to rotate it about its own centre. In one direction that rotation rolls the sprag toward its tall diagonal — but the tall diagonal does not fit, so the sprag jams between the races. It cannot stand up, it cannot lie down, so instead it transmits force: the inner race, the sprag, and the outer race become one rigid body and torque flows straight through. Reverse the relative rotation and friction rolls the sprag the other way onto its short diagonal, which is smaller than the gap, so the sprag tips out of contact, the races slip past each other, and the clutch overruns — it freewheels. That is the entire principle: a wedge that builds itself when you push one way and dissolves itself when you push the other.

The geometry — strut angle and the self-locking condition

The single most important number in a sprag clutch is the strut angle (also called the wedge or gripping angle), φ. Picture the straight line that connects the inner contact point to the outer contact point on one sprag — the "strut." The strut angle is how far that line tilts away from pure radial. A purely radial strut (φ = 0) could never grip; a steep strut grips easily but risks denting the races. The sprag locks only if the friction available at the two contacts can supply the tangential force the load demands. That gives the classic self-locking inequality:

Self-locking condition (steel on steel):

   tan(φ / 2)  <  μ

   where  φ = strut (wedge) angle
          μ = friction coefficient at the contacts

Hardened, ground steel races, light oil:  μ ≈ 0.05 – 0.12
   →  φ_max ≈ 2·atan(0.08) ≈ 9.1°
   →  practical design range:  φ ≈ 6° – 9°

Read that inequality physically. The load demands a tangential (drag) force F_t at each contact. To stay wedged, the sprag must develop a normal force F_n such that the friction μ·F_n meets or exceeds F_t. The geometry ties F_t and F_n together through the strut angle: a shallow strut angle means a small F_t produces a large F_n, so friction wins easily and the sprag locks. As you open the strut angle, F_n shrinks for the same load and friction loses — the sprag rolls over and skids.

The elegant part is that a sprag rolls as it engages. Designers shape the contact profiles so the strut angle decreases as torque increases. So the harder you load it, the more deeply it wedges and the bigger its safety margin against slipping — a self-stabilising grip. The cost of going too far is real: if the strut angle is too small, peak Hertzian contact stress at the two lines climbs past the brinelling limit (roughly 1,500–2,000 MPa for case-hardened bearing steel), the races take permanent dents, and once dented the wedge geometry is ruined.

Torque capacity and a worked example

Because every sprag in a full-complement ring shares the load, torque scales with the number of sprags. The per-sprag normal force is fixed by the brinelling limit of the line contact; multiply by the friction lever arm and the count, and you get capacity.

Rated torque  ≈  N · F_t · r_m

   N   = number of sprags around the ring
   F_t = allowable tangential force per sprag (set by brinelling limit)
   r_m = mean race radius

Example — a torque-converter stator sprag clutch:
   Mean race radius      r_m = 30 mm = 0.030 m
   Sprag count           N   = 24
   Allowable F_t/sprag       ≈ 280 N  (line contact, 60 HRC races)

   T_rated ≈ 24 × 280 N × 0.030 m ≈ 200 N·m

Compare a 6-roller ramp clutch in the same 60 mm bore:
   N = 6,  F_t/roller ≈ 280 N,  r_m = 0.030 m
   T_rated ≈ 6 × 280 × 0.030 ≈ 50 N·m   → about 4× less

That 4-to-1 difference is the whole reason sprag clutches exist. By replacing a handful of rollers-on-ramps with two-to-three dozen sprags-between-smooth-races, you pack far more torque into the same diameter, and you do it with far less lost motion (backlash) because there is no ramp gap to take up — the sprags are already touching both races, energised by their spring, waiting to wedge.

Why a sprag clutch beats a ratchet-and-pawl

The oldest one-way device is the ratchet and pawl — a toothed wheel and a spring-loaded finger. It is positive, cheap, and impossible to back-drive. But it has two fatal drawbacks for high-speed machinery. First, it is noisy: every overrunning revolution the pawl clicks over each tooth (think of a bicycle freewheel ticking). Second, it has index error: when load reverses, the pawl must travel to the next tooth before it catches, so there is a sudden lash of up to one tooth pitch before torque transmits. At 6,000 rpm that lash becomes a destructive hammer-blow.

A sprag clutch has neither problem. It is silent because nothing skips over teeth — the sprags simply lie down and slip on smooth races. And it has near-zero index error because every sprag is already in contact, so engagement is effectively instantaneous and happens at whatever angular position the reversal occurs. The trade is that a ratchet is positive (it cannot slip), while a sprag holds by friction (it can slip if you abuse the strut angle, contaminate the oil, or brinell the races).

Sprag clutch versus the alternatives

Property	Sprag clutch	Roller ramp clutch	Ratchet & pawl	Wrap (band) spring clutch
Holding principle	Friction wedge (cam)	Friction wedge (roller on ramp)	Positive tooth	Friction (spring grips shaft)
Engagement lag	< 1 ms, near-zero index	Small ramp take-up	Up to one tooth pitch	A few degrees of wind-up
Overrun noise	Silent	Silent	Audible clicking	Mild rub
Torque density	Highest (20–40 contacts)	Medium (6–12 rollers)	Low–medium	Medium
Backlash	Very low	Low–medium	One tooth	Low
Can slip / back-drive	Yes, if abused	Yes, if abused	No (positive)	Yes
Dirt / misalignment tolerance	Lower	Higher	High	Medium
Typical use	TC stator, backstops, helicopter freewheel	Starter Bendix, cheap freewheel hubs	Bicycle freewheel, winch pawl	Printers, appliance drives

Where sprag clutches actually show up

• Torque-converter stator. The classic high-volume use. Inside an automatic-transmission torque converter the stator redirects oil to multiply torque at low speed, reacting against a sprag clutch grounded to the housing. As the turbine catches up to the pump (the "coupling point"), the oil would try to spin the stator backwards — so the sprag clutch releases and lets the stator freewheel, removing a parasitic drag that would otherwise cost several percent in efficiency at cruise.
• Engine starter drives. A sprag (or roller) overrunning clutch lets the starter motor crank the engine, then disengages the instant the engine fires so the engine cannot back-drive and over-speed the small starter motor to destruction.
• Helicopter and turboprop freewheel units. A sprag pack sits between each engine and the main rotor gearbox. In normal flight the engine drives the rotor; if an engine quits or a multi-engine ship loses one, the freewheel decouples the dead engine so the rotor can keep turning, enabling autorotation for a survivable landing.
• Conveyor and hoist backstops. A large sprag clutch (a "holdback") prevents a loaded inclined belt or bucket elevator from running backwards if the drive loses power. Mine and quarry conveyor backstops routinely hold hundreds of kilonewton-metres.
• Indexing and feed drives. Paper-handling rollers, printing presses, packaging machines, and ATM cash dispensers use sprag clutches to drive forward on one stroke and freewheel on the return, giving precise, repeatable indexed motion.
• Overrunning alternator decoupler (OAD) pulleys. Modern serpentine-belt accessory drives put a small sprag-type one-way clutch (plus a spring) inside the alternator pulley to absorb crankshaft torsional pulses, reducing belt chirp and tensioner wear.
• Two-speed and dual-drive machinery. A sprag lets a slow auxiliary motor drive a shaft while a faster main motor automatically takes over and overruns the auxiliary the moment it spins faster — no clutch pedal, no shifting logic.

Variants and design choices

• Full-complement vs. caged sprag rings. A full-complement ring crams in the maximum number of sprags for peak torque density. A caged design spaces the sprags in a retainer for better high-speed overrun behaviour and to control the energising spring force precisely.
• Centrifugal lift-off (phasing) sprags. At high overrun speed the sprags can be designed to tip outward under centrifugal load and lift clear of the inner race, eliminating drag wear during long freewheel periods — vital for helicopter freewheels that overrun for hours.
• Ribbon spring vs. garter spring energising. The spring's only job is to keep every sprag lightly touching both races so it is ready to wedge. Too weak and a sprag misses engagement; too strong and freewheel drag and wear climb.
• Indexing (logarithmic-spiral) sprag profiles. Precision indexing clutches use specially profiled sprags that engage with near-zero lost motion, so the output shaft starts moving the instant the input reverses — critical in high-speed printing and assembly.
• Sprag clutch with bearing support. Most production units integrate ball or needle bearings so the unit also locates the two races radially, turning the clutch into a single drop-in cartridge (the standard form in starter and gearbox modules).

Failure modes — where sprag clutches actually break

• Rollover / "pop-out" wear. In the freewheel direction each sprag drags lightly on both races, polishing a wear track and slowly altering the wedge geometry until the strut angle drifts and engagement turns unreliable. The cure is centrifugal lift-off designs, clean light oil, and minimising overrun time.
• Brinelling. Shock torque drives the line contacts past the surface yield limit (~1,500–2,000 MPa), denting the races. A dented race no longer presents a smooth wedging surface, so the sprag skips or locks erratically. Cure: keep peak contact stress well under the limit and harden the races to ≥60 HRC.
• Spring fatigue. The energising ribbon spring flexes every revolution and eventually relaxes. Sprags then lie down and miss engagement, producing intermittent or delayed lock-up.
• Galling from lubrication starvation. Engagement is a brief, high-slip event. With no oil film the contacts gall, raising friction unpredictably and tearing the surface. A clean, low-viscosity oil is mandatory; many failures trace to the wrong fluid or a clogged feed.
• Thermal softening. Repeated high-energy engagements (starter abuse, torque-converter stall) heat the races and can temper the hardened layer below 60 HRC, after which brinelling accelerates.
• Slip / runaway. The worst case for a safety backstop: a worn or contaminated clutch with too large an effective strut angle cannot generate the normal force to hold, the sprags roll over, and a loaded conveyor runs backwards. This is why backstop clutches are on a mandatory inspection schedule.

Common pitfalls when applying a sprag clutch

• Specifying the average torque, not the peak. Shock and torsional spikes brinell long before steady torque is a problem. Size to peak contact stress with margin.
• Ignoring overrun duty. A clutch that locks briefly but freewheels for hours (helicopter, alternator decoupler) lives or dies by its lift-off and drag behaviour, not its torque rating.
• Using the wrong lubricant. The friction coefficient at the contacts is the design. An EP gear oil with the wrong additives can cut μ enough to push the clutch out of the self-locking window and make it slip.
• Mounting on a soft or out-of-round race. Sprags need true, hard cylindrical surfaces. A turned-soft or distorted race lets sprags brinell or sit at uneven strut angles, so a few carry all the load.
• Treating it as positive. A sprag clutch holds by friction, not by a tooth. For life-safety holdbacks, add margin and, where codes require it, an independent secondary holding device.