The Constant-Velocity Joint: Transmitting Power Around a Bend

The problem: torque around a moving corner

In a front-wheel-drive car the same two wheels must both steer and drive. Each halfshaft runs from the transaxle differential out to a wheel hub that swings through roughly 40° of steering lock while the suspension simultaneously moves the hub up and down through its travel. The shaft therefore has to deliver full engine torque — enough to spin the tire — through a joint that is continuously changing angle, and it must do so without changing the wheel's rotational speed relative to the shaft.

The last requirement is the hard one. Any simple hinge that carries rotation across an angle tends to speed the output up and slow it down within each revolution. If the wheel's angular velocity oscillated even a few percent per turn, that oscillation would feed back as a torsional vibration into the axle, the transmission, and ultimately the steering wheel and floorpan — a droning, shuddering nuisance that gets worse with steering angle and speed. A joint that carries torque through an angle while holding the speed ratio at exactly 1:1 is called a constant-velocity or homokinetic joint. Building one is a problem in pure geometry as much as in metallurgy.

The Cardan joint's velocity sin

The oldest angle-transmitting coupling is the universal joint, also called the Cardan or Hooke joint: two forked yokes linked by a cross-shaped spider. It is cheap, strong, and used by the billion — but it is not constant-velocity. If the input turns through angle θ at steady speed, the output angle obeys tan θ_out = tan θ_in / cos β, where β is the bend angle. Differentiating gives the instantaneous speed ratio:

• ω_out/ω_in = cos β / (1 − sin²β · cos²θ_in)

The output speed swings between cos β (its slowest) and 1/cos β (its fastest) twice per revolution. At a 30° bend that is 0.87× to 1.15× — a ±15% fluctuation at double the shaft frequency. The associated angular acceleration produces inertial torque pulses that grow with both angle and rpm, which is exactly the vibration a driven, steered wheel cannot tolerate.

Engineers can cancel the error with a double Cardan joint: two Hooke joints in series sharing a centre yoke whose two sets of ears lie in one plane (so each joint's driving yoke sits 90° from the next), each running at half the total angle. Because the second joint's fluctuation is the mirror image of the first, the two errors annul and the assembly delivers constant velocity — provided the two halves stay at equal angles, which a centring ball and socket enforces. Double-Cardan units are common in steering shafts and in the front driveshafts of four-wheel-drive trucks, but they are bulky. For the tight, high-angle package of a car's front halfshaft, the automotive industry turned instead to the ball-and-cage joint.

The homokinetic plane: geometry of equal speed

The single principle behind every CV joint is the bisecting-plane theorem. Consider two shafts whose axes cross at a point O with an angle β between them, coupled by some moving contact point (a ball, a roller, a tongue). The instantaneous angular velocity of each shaft equals the tangential speed of that contact point divided by its perpendicular distance from that shaft's axis. For the two angular velocities to stay equal at every instant of the revolution, the contact point must remain equidistant from both axes — and the locus of points equidistant from two crossing lines is precisely the plane that bisects the angle between them.

This bisecting plane is the homokinetic plane. Any coupling element that is geometrically forced to live in it will transmit rotation at a constant 1:1 ratio, no matter how the joint is bent. A Cardan joint fails because its spider pins do not stay in that plane; they trace a tilted circle instead. The engineering challenge of a CV joint is therefore mechanical: how do you constrain the torque-carrying elements so that, as the shaft swings, they are automatically herded onto the bisector of an angle that keeps changing? The Rzeppa joint answers with a beautifully simple groove geometry.

Inside the Rzeppa joint: balls, cage, and offset grooves

The dominant automotive CV joint is the Rzeppa joint, developed by Ford engineer Alfred Hans Rzeppa, who patented a caged six-ball joint in 1926 and refined it in the early 1930s into the self-centring offset-groove form used today. It looks like a steel flower bud. An outer race — a bell-shaped housing splined to the wheel hub — has six curved grooves cut into its inner spherical surface. A smaller inner race, splined to the driveshaft, carries six matching grooves on its outer sphere. Between them ride six hardened steel balls — eight in some modern high-angle designs — one per pair of grooves, all held in the windows of a thin cage so that every ball sits in a common plane. Torque passes from shaft to inner race, through the balls, into the outer race and the hub — carried entirely by ball-on-groove contact.

The magic is in the groove shape. The centers of curvature of the inner and outer grooves are offset by equal amounts on opposite sides of the joint center O. Each pair of grooves therefore forms a shallow wedge whose narrow end lies on one side and whose wide end lies on the other. When the joint articulates by angle β, this wedge steers each ball toward the intersection of the two grooves — a point that the offset geometry places exactly on the homokinetic bisecting plane. The cage then holds all six balls coplanar so none can wander off. The result is that the balls automatically self-center in the angle bisector at every steering position, satisfying the constant-velocity condition mechanically, without any external control.

The parts are punishingly loaded. The balls are precision bearing balls; races are made from case-hardening low-alloy steels such as carburized SAE 8620 or 20MnCr5, and the balls from through-hardened AISI 52100 (100Cr6) bearing steel, finished to a surface hardness of 58–62 HRC over a tough core. Under peak torque, the Hertzian contact stress where a ball presses into its groove reaches on the order of 2–3 GPa, which is why the raceways must be hard, clean, and precisely ground: a soft or contaminated raceway spalls in thousands of miles rather than the intended 150,000-plus. Because the balls mainly roll rather than slide as torque is transmitted (some sliding and ball spin remain, growing with articulation angle), the fixed Rzeppa joint is extremely efficient (well above 98%) and quiet.

Outboard fixed, inboard plunging: dividing the labor

A car's halfshaft actually uses two different joints, because the two ends have different jobs. The outboard joint, at the wheel, must swing through huge steering angles but does not need to change length — so it is a fixed Rzeppa ball joint good for roughly 45–50° of articulation at full lock. The inboard joint, at the transmission, sees only modest angles (about 18–26°) but must let the shaft lengthen and shorten as the suspension arcs change the straight-line distance between the diff and the hub. That axial motion is called plunge.

The classic inboard design is the tripod (or tripode) joint: a three-armed 'spider' pressed onto the shaft, each arm carrying a needle-roller-mounted spherical roller that runs in one of three axial tracks inside a cup-shaped 'tulip' housing. As the suspension moves, the rollers simply slide along their tracks, giving up to about 50 mm of plunge. The tripod is cheap, robust, and handles plunge naturally. It transmits nearly constant velocity, but the orbiting motion of its three rollers generates a small cyclic internal axial force at three times shaft frequency (the 'generated axial force'), a known source of low-speed launch shudder in front-drive cars. Some vehicles instead use plunging ball joints — cross-groove (VL) or double-offset (DO) designs — which trade a little cost for lower vibration. Either way, the design split is deliberate: high-angle purity at the wheel, plunging tolerance at the transmission.

Boots, grease, and the clicking failure

A CV joint is a bath of precision-ground steel that must stay clean and lubricated for the life of the car, in the spray zone behind a wheel. It survives thanks to a convoluted boot — a bellows of thermoplastic elastomer (Hytrel-type TPE), neoprene, or silicone — clamped over the joint and packed with a heavy molybdenum-disulfide or lithium-complex extreme-pressure grease. The boot flexes through the full articulation range, seals out water and grit, and holds the grease against the slinging of rotation. Boots are validated across a brutal window — roughly −40 °C to +120 °C, ozone exposure, and millions of flex cycles.

By far the most common failure is a torn boot. Once it splits, grease flings out and abrasive road grit gets in; the raceways and balls wear and pit, and the joint develops radial play. The textbook symptom is a rhythmic clicking or knocking that appears only during turns — the worn outboard joint rattling under load at high angle — while a worn inboard tripod tends to produce a shudder on hard acceleration. Other failure modes include brinelling (permanent raceway indentation from shock loads) and false brinelling (fretting wear from vibration while the car sits parked or is trailered), fatigue spalling of the contact tracks, and spline wear. Manufacturers qualify joints on rigs that spin them at articulation angle under cyclic torque to failure, and specify a static torsional strength typically two to three times the rated dynamic torque — a passenger-car halfshaft joint carrying several thousand newton-metres before yielding. Catch a split boot early and a joint that would have lasted the life of the car is saved by a $20 rubber bellows and fresh grease.

From Weiss and Tracta to GKN

The Rzeppa joint did not appear in a vacuum. Two rival constant-velocity designs were already in service by the late 1920s. The Bendix-Weiss joint (Carl Weiss, patented 1925) used four balls plus a small centering ball rolling in curved grooves, and became the standard front-axle drive joint of World War II 4×4s such as the Willys Jeep and GMC trucks. The Tracta joint (patented by Pierre Fenaille in 1926), a nested double tongue-and-groove coupling, powered the front-drive Tracta cars Fenaille built with Jean-Albert Grégoire and was also widely licensed for military vehicles. Rzeppa's six-ball, offset-groove design proved the most compact and high-angle of the three, and it won the market decisively when mass-market front-wheel-drive cars exploded in the 1970s and 1980s.

Today the technology is dominated by a handful of specialists — above all GKN Automotive (heir to Hardy Spicer, Birfield, and the German Löbro / Löhr & Bromkamp lines), which supplies a large share of the world's driveshafts. Their catalog names — RF (fixed ball), UF (undercut-free high-angle, reaching about 50°), AC (angular-contact fixed), VL cross-groove and DO double-offset plunging joints — are simply refinements of the same 1926 idea: hold the torque-carrying elements in the plane that bisects the angle, and rotation flows around the bend at perfectly constant speed.

Shaft couplings that transmit torque through an angle, and how faithfully they preserve speed

Joint type	Output velocity	Max angle	Axial plunge	Typical location
Cardan (Hooke) single	Fluctuates ~±15% at 30°, twice per rev	~30–35° mechanical limit; only a few ° continuous at speed	None (needs slip yoke)	Rear driveshaft
Double Cardan	Constant if the two halves run at equal angles, centre-yoke ears in one plane	~30–40°	None	Steering columns, 4×4 front shafts
Rzeppa ball (fixed)	Constant at any articulation	~45–50°	None	Outboard (wheel-side) halfshaft
Tripod (plunging)	Near-constant; small cyclic axial force at 3×	~18–26°	~50 mm	Inboard (transmission-side) halfshaft
Bendix-Weiss (historic)	Constant (four balls + centering ball)	~30–32°	None	WWII 4×4 front drive axles