The core mechanical difference is a crown pressed into both the hub teeth and the elastomer legs: curved-jaw zero-backlash couplings machine a radial curvature into the jaw face and a matching axial crown on the spider so the element sits under pre-compression, eliminating the rotational air gap that a straight trapezoidal jaw leaves between hub flank and spider leg [S1][S3]. Lovejoy (a Timken company) lists a maximum bore of 200 mm (7.63 in) and a maximum nominal torque of 27,996 Nm (247,800 in-lb) on its CJ-series curved-jaw line, with angular misalignment capability up to 1.3° on the standard product [S1]. Miki Pulley transmits up to 387 ft-lb (525 Nm) through its jaw/spider family and quotes low-to-zero backlash depending on the model selected [S5].
Straight-jaw couplings still dominate general industrial power transmission, including pumps, gearboxes, conveyors, compressors, and generator sets, because the open trapezoidal profile gives the elastomer more room to flex under misalignment [S1][S2]. The two designs share the same three-piece architecture (two hubs plus an elastomeric spider), the same fail-safe behaviour when the spider wears (the metallic jaws lock metal-on-metal and keep the load turning, which matters on elevators and fire pumps), and the same lubrication-free operating mode that distinguishes jaw couplings from gear or grid couplings [S2][S4].
Jaw profile: how the geometry creates or kills backlash
The standard L-type (straight-jaw) spider is a solid Nitrile Butadiene (Buna-N) element at roughly 80 Shore A; it fills the gap between two flat-flanked jaws, but a small clearance fit on the bore and the natural compression set of the elastomer leave a measurable rotational dead band before torque is transmitted [S4]. Curved-jaw hubs machine a precise curved tooth profile that press-fits the spider rather than compressing it into a trapezoidal pocket; Ruland describes this as a press-fit that removes the angular play normally seen in a straight-jaw assembly [S3], and Lovejoy pairs the curved hub with both radial and axial crowning of the elastomer to give the "improved torsional softness" that lets the coupling still damp vibration despite the tighter fit [S1].
Backlash is not the only variable. Curved-jaw designs trade some misalignment capacity for repeatability: Lovejoy's curved-jaw angular limit is 1.3°, while a Hytrel spider in a straight-jaw coupling is generally limited to about 0.5° of angular misalignment regardless of jaw profile [S1][S4]. For positioning axes where the lost-motion arc directly becomes a servo following error, this trade is the entire reason the curved-jaw geometry exists. Engineers comparing it against other flexible shaft couplings should treat backlash as a separate specification row from misalignment tolerance, not a single combined number.
Spider material and Shore hardness: the second axis of selection
Spider selection drives damping, torque capacity, temperature range, and cyclic-load survivability, and it interacts with jaw profile rather than overriding it [S2][S4]. Buna-N at roughly 80 Shore A is the general-purpose default: -40 to 212 °F (-40 to 100 °C), oil resistant, good damping, tolerant of start/stop duty, and typically rated near 1° angular misalignment [S4]. Urethane delivers about 1.5× the torque capacity of Buna-N at the cost of damping and a narrower -30 to 160 °F (-34 to 71 °C) window; it is not recommended for cyclic or start/stop service [S4]. Hytrel pushes the temperature ceiling to 250 °F (121 °C) and the floor to -60 °F (-51 °C), with the same ~0.5° angular limit and the same "no cyclic load" caveat [S4].
Miki Pulley's product matrix makes the profile-versus-hardness interaction explicit. The ALS-R curved-jaw hub uses a 97 Shore A Type "R" elastomer for high torque, high torsional stiffness, and zero backlash with limited misalignment; the ALS-Y curved-jaw hub uses a 90 Shore A Type "Y" element for a zero-backlash connection with moderate torsional stiffness and more misalignment headroom; the ALS-B straight-jaw hub uses a 97 Shore A Type "B" spider for higher misalignment with minimal backlash; and the AL SPRFLEX pairs a straight-jaw profile with an NBR element for high flexibility and high damping at low backlash [S5]. Two takeaways for spec writing: (1) higher Shore A on a curved-jaw hub stiffens the drive and tightens the position loop, and (2) straight-jaw with a soft spider is a different product than straight-jaw with a hard spider, even though both share the same jaw geometry.
Decision matrix: curved-jaw zero-backlash vs straight-jaw
Four criteria, drawn from the cited product data, line the two profiles up against each other. (1) Rotational backlash: zero on a press-fit curved-jaw assembly per Ruland and Lovejoy [S1][S3], versus measurable but reducible on a straight-jaw unit, with Miki Pulley's ALS-B and AL SPRFLEX explicitly marketed as "minimal backlash" and "low backlash" rather than zero [S5]. (2) Angular misalignment: up to 1.3° on Lovejoy's curved-jaw CJ type [S1], versus roughly 1° on a Buna-N or urethane straight-jaw spider and 0.5° on a Hytrel spider [S4]. (3) Torque capacity: up to 27,996 Nm (247,800 in-lb) on Lovejoy's CJ curved-jaw [S1] and up to 525 Nm (387 ft-lb) on Miki Pulley's jaw/spider family [S5], with urethane giving about 1.5× the torque of Buna-N in equivalent sizes [S4]. (4) Temperature window: -40 to 212 °F for Buna-N, -30 to 160 °F for urethane, -60 to 250 °F for Hytrel, with the jaw profile itself imposing no additional thermal limit [S4].
Use-case mapping follows directly. Specify a curved-jaw zero-backlash coupling when the driven load is a servo or stepper axis, a ball-screw lead, a packaging indexer, or any closed-loop position control where lost motion becomes a following-error budget item [S2][S3][S5]. Specify a straight-jaw coupling when the drive is a pump, fan, compressor, conveyor, or generator set feeding a line-shaft, where some backlash is acceptable in exchange for the wider misalignment window and the lower unit cost [S1][S2]. Mixing the two (a curved-jaw hub with a soft Buna-N spider) is valid when the application needs both zero backlash and the damping of a softer elastomer; the press-fit is created by the jaw profile, not by the Shore hardness [S3][S5].
Limits, failure modes, and standards to keep in view
Every cited curved-jaw product is also a fail-safe device: when the elastomer eventually fatigues, the metal jaws interlock and continue to drive the load, generating noise and wear but not an instant stop, which is the same behaviour as a straight-jaw coupling [S2][S4]. Once that fail-safe event has occurred, both hubs must be replaced along with the spider, because hub-on-hub running gall's the teeth and the assembly is no longer dimensionally correct for a fresh insert [S4]. Cyclic or start/stop service is the documented weak point for urethane and Hytrel spiders regardless of jaw profile; Buna-N remains the default for reversing drives [S4].
Standards and certifications are largely hub-and-bore driven rather than profile driven: Lovejoy publishes a CE-marked European catalogue and a separate German-language catalogue for the CJ curved-jaw line, indicating conformity with the EU machinery directive's coupling provisions [S1]. Bore tolerance is a clearance fit on general-purpose jaw couplings, so any application requiring an interference fit for zero axial play must specify it on the purchase order rather than assume the catalogue default [S4]. For servo sizing, follow the coupling manufacturer's per-size torque derating rather than combining catalogue maximum torque with the jaw profile name; backlash and torque are tested at the size level, not the family level [S1][S5].
Procurement signals worth tracking
Three signals are worth a follow-up check before specifying. First, Miki Pulley's continued release of 90 and 97 Shore A curved-jaw elements shows the market is pushing harder elastomers into zero-backlash assemblies to lift torque density without sacrificing position repeatability [S5]. Second, Lovejoy's CJ curved-jaw catalogue tops out at 200 mm bore and 27,996 Nm, so any drive above that envelope is no longer a jaw-coupling decision and must move to a gear coupling or disc coupling comparison [S1]. Third, urethane's 1.5× torque multiplier over Buna-N at narrower temperature limits means that engineers retrofitting a curved-jaw coupling into a hotter enclosure should re-derive the size, not just swap the spider [S4]. Cross-checking the chosen jaw profile against the driven equipment's actual position-loop bandwidth and the spider's cyclic-load rating before issuing a PO will catch most of the field failures that show up as "the coupling has backlash" tickets.
See also our earlier report, Automatic Level Coarse Setup: Tubular Vial vs Circular Bubble for the First Centering Pass.