Goubet's Universal Shaft Coupling is a double-yoke jaw coupling patented by French engineer Goubet in the 19th century to transmit rotary motion between two shafts whose axes meet at an angle. Unlike a rigid sleeve coupling that demands near-perfect alignment, it uses a central cross or intermediate block pivoting in two yokes so the driven shaft can sit at up to 15° off the driver. It solved the problem of mill line shafts forced through awkward building geometry, and it kept thousands of factory countershafts running where flexible coupling sleeves would have torn themselves apart within a shift.
Goubet's Universal Shaft Coupling Interactive Calculator
Vary two shaft operating angles and see the estimated twice-per-revolution output speed ripple for a single Goubet/Hooke joint.
Equation Used
The calculator estimates the output speed ripple of a single Goubet universal shaft coupling. In the article example, a 10 deg joint gives about +/-1.5% speed variation and a 20 deg joint gives about +/-6%. The rounded relation ripple_% ~= 0.015 beta_deg^2 captures that practical Hooke-joint behavior for small to moderate shaft angles.
- Single Goubet universal shaft coupling with constant input speed.
- Rounded small-angle Hooke-joint ripple approximation used for the article examples.
- Ripple occurs twice per shaft revolution.
- Torque loss, bearing friction, and shaft compliance are ignored.
The Goubet's Universal Shaft Coupling in Action
Goubet's design is a Hooke joint built for the mill floor — two forked yokes, one keyed to the driver shaft and one to the driven shaft, both pivoting on a central trunnion cross. When the driver turns, each arm of the cross rocks in its bearing as the angle between the shafts forces the geometry to flex. Power passes through the cross from one yoke to the other regardless of whether the shafts are perfectly collinear or tilted. The whole point is that you can route a line shaft around a mill column or down into a basement gear pit and still deliver torque cleanly to the next bearing block.
The mechanism is a Cardan coupling at heart, so it inherits the same kinematic quirk: at any non-zero shaft angle the output speed varies sinusoidally through each revolution even when the input is constant. At a 10° operating angle the output speed swings roughly ±1.5% twice per turn. At 20° it swings ±6%. That is why Goubet, like Cardan before him, recommended pairing two of his couplings back-to-back through a short intermediate shaft with the yokes phased 90° apart — the second joint cancels the speed ripple of the first and you get an effective constant-velocity drive.
Tolerances matter more than the geometry suggests. The trunnion bores in each yoke must match within about 0.05 mm or the cross sits cocked under load and you get a vibration the millwrights used to call "hammering" — a once-per-rev thump felt right through the bearing pedestal. Run a coupling at a higher angle than rated and the trunnion bushings wear oval inside a few hundred hours, the yoke arms start to flex, and eventually a cross arm shears off at the root fillet. Lubrication on the original Goubet pattern was an oil cup feeding each trunnion; let it run dry and the bronze bushing welds itself to the steel pin.
Key Components
- Driver yoke: Forked iron or steel fitting keyed and taper-pinned to the input shaft. The two arms carry the trunnion bores that hold one axis of the central cross. Bore concentricity to the shaft axis must hold within 0.05 mm TIR or the joint introduces its own runout.
- Driven yoke: Mirror of the driver yoke, keyed to the output shaft and oriented 90° around its axis from the driver yoke. It carries the second pair of trunnion bores. Phasing between the two yokes is what creates the angular misalignment capacity.
- Trunnion cross (spider): Four-armed forging — sometimes a cast iron block in lighter Goubet variants — that pivots in all four yoke bores simultaneously. The arms see pure bending in service; root fillet radius is critical, typically 3-5 mm on a 60 mm cross to keep stress concentration below 2.0.
- Trunnion bushings: Bronze sleeve bearings pressed into each yoke bore, riding on the hardened steel cross arms. Diametral clearance around 0.04-0.08 mm. Anything tighter binds at angle, anything looser hammers under reversing load.
- Oil cups or grease nipples: Each trunnion has its own lubrication port. Goubet's original drawings show four small brass oil cups, one per arm. Modern restorations usually convert to grease fittings serviced every 200 operating hours.
- Retaining keys and taper pins: Hold the yokes axially on their shafts. A loose key here lets the yoke walk on the shaft and the joint angle changes during operation, which destroys the bushings within a week.
Industries That Rely on the Goubet's Universal Shaft Coupling
Goubet's coupling found its home anywhere torque had to turn a corner — and in a 19th century factory built around a single steam engine and miles of overhead line shafting, that was almost everywhere. Modern uses are narrower because flexible couplings and gearmotors took over short transmissions, but the joint still earns its keep wherever you have a fixed angular offset between two rotating shafts.
- Paper mill restoration: Recommissioned overhead line shafts at the Robert C. Williams Paper Museum, where a Goubet-pattern double joint connects the main jack shaft to a beater drive offset 12° from the bull gear axis.
- Heritage machine shops: Driving back-geared lathe countershafts off ceiling line shafts in working museums like the Hagley Mill in Delaware — the Goubet joint accommodates ceiling joist offsets without forcing the line shaft to zigzag.
- Marine auxiliary drives: Coupling steam donkey engines to capstan and winch shafts on preserved vessels, where engine and load sit at fixed angles dictated by the deck plan.
- Agricultural PTO drives: Direct descendant of Goubet's design appears as the modern tractor PTO universal joint, transmitting up to 100 HP at angles up to 25° on implements like a John Deere rotary cutter.
- Rolling mill spindles: Heavy-duty descendants of the Goubet coupling drive roll spindles on Steckel mills and small bar mills where the work roll axis shifts during pass changes.
- Vintage textile mill rebuilds: Reconnecting carding line drives at sites like Quarry Bank Mill in Cheshire, where the original 1830s line shaft routing demands several angled segments.
The Formula Behind the Goubet's Universal Shaft Coupling
The kinematic behaviour of a single Goubet joint is captured by the Cardan equation relating output angular velocity to input angular velocity and the operating angle. This is the equation you reach for when you want to know how much speed ripple your line shaft will feel at a given misalignment. At the low end of the typical operating range — say 5° — the ripple is so small (under 0.4%) that you can ignore it for most mill drives. At a nominal 10-15° you are in the design sweet spot: ripple stays under 4% and the joint runs cool. Push past 20° and the ripple climbs fast, the trunnion forces double, and bushing life collapses.
Variables
| Symbol | Meaning | Unit (SI) | Unit (Imperial) |
|---|---|---|---|
| ωout | Instantaneous angular velocity of the driven shaft | rad/s | RPM |
| ωin | Angular velocity of the driver shaft (assumed constant) | rad/s | RPM |
| β | Operating angle between driver and driven shaft axes | rad or ° | ° |
| θ | Rotational position of the driver yoke (0° when driver yoke arms lie in the plane of the two shafts) | rad or ° | ° |
| Δω/ωin | Peak-to-peak speed ripple ratio over one revolution | dimensionless | dimensionless |
Worked Example: Goubet's Universal Shaft Coupling in a heritage flour mill line shaft restoration
A working heritage flour mill in Saint-Hippolyte Quebec is recoupling an overhead line shaft from a 22 kW geared electric motor down to a millstone runner drive that sits 14° below horizontal because of the timber-frame gallery layout. The motor turns at 220 RPM. The restoration team is using a single Goubet-pattern coupling and wants to know the worst-case speed ripple the millstone will see, and whether they should switch to a phased double joint.
Given
- ωin = 220 RPM
- β = 14 °
- θ range = 0 to 360 °
Solution
Step 1 — at the nominal 14° operating angle, find the maximum and minimum output speeds. Maximum occurs at θ = 0° and 180° where the cos(θ) terms maximise the denominator effect:
Step 2 — minimum output speed occurs at θ = 90° and 270°:
Step 3 — peak-to-peak speed ripple as a percentage of input:
At the low end of a sensible operating range — 5° — the same calculation gives a ripple of only 0.4%, which a millstone runner would not even register; the stone simply turns smoothly. At a nominal 14° you get 6.0% ripple, which is felt as a faint twice-per-revolution surge but is still inside what a granite stone and its rynd can tolerate without chatter. Push the geometry to 25° and ripple climbs to nearly 20% — the stone would visibly pulse, the bedstone clearance would vary every half turn, and you would see streaky flour with inconsistent grind.
Step 4 — check whether a phased double joint solves it. If the team installs a second Goubet coupling on a short intermediate shaft, with the second yoke rotated 90° from the first and the intermediate shaft at the same 14° angle to both ends, the speed ripples cancel and output ripple drops below 0.1% regardless of operating angle.
Result
Nominal answer: a single Goubet coupling at 14° produces a 6. 0% peak-to-peak speed ripple at 220 RPM, swinging the millstone between roughly 213.5 and 226.7 RPM twice per shaft revolution. In practice the miller will feel a faint surge through the stone hutch and may see fine streaking in the flour at this level — tolerable for demonstration milling, marginal for production. Across the operating range the difference is dramatic: under 1% ripple at 5°, the 6% nominal at 14°, and a punishing 20% at 25° where the joint becomes unusable for stone milling. If you measure ripple higher than the predicted 6% the likely causes are a bent driven shaft adding runout that stacks with the geometric ripple, taper-pin slop letting the driven yoke phase-shift on its shaft during each revolution, or trunnion bushing clearance grown past 0.10 mm so the cross floats and adds a once-per-rev knock on top of the twice-per-rev kinematic surge. The cleanest fix on this build is the phased double joint described in step 4 — drop ripple below 0.1% and the millstone runs as smoothly as a direct drive.
When to Use a Goubet's Universal Shaft Coupling and When Not To
Goubet's coupling is one of three classical options for connecting misaligned shafts, and the right pick depends on operating angle, speed, torque ripple tolerance, and how often you need to service it. Here is how the Goubet (single Hooke joint) compares to a flexible disc coupling and a constant-velocity Rzeppa-style joint on the dimensions practitioners actually search on.
| Property | Goubet's Universal Coupling (single Hooke) | Flexible disc coupling | Rzeppa-type CV joint |
|---|---|---|---|
| Maximum operating angle | 25° practical, 15° preferred | 1-3° | 45-50° |
| Output speed ripple at 15° angle | ~7% peak-to-peak | Negligible (< 0.1%) | Negligible (< 0.5%) |
| Torque capacity (60 mm bore) | 1500-3000 Nm continuous | 200-800 Nm continuous | 1000-2500 Nm continuous |
| Typical operating speed | Up to 1500 RPM | Up to 6000 RPM | Up to 4000 RPM |
| Service interval (lubrication) | 200 hours grease, weekly oil cup | Inspection only, no lube | 5000 hours sealed grease |
| Typical service life at rated load | 8000-15000 hours | 30000+ hours | 5000-10000 hours |
| Cost relative index | 1.0 (baseline) | 0.6 | 3-5 |
| Best application fit | Mill line shafts, PTO drives, fixed angular offsets | Direct motor-to-pump or motor-to-gearbox | Vehicle drive shafts, varying high angles |
Frequently Asked Questions About Goubet's Universal Shaft Coupling
Because the cancellation only works if both joints operate at the same angle and the intermediate shaft phases the yokes correctly. On a real mill floor the upstream and downstream shaft angles are rarely equal — you might have 14° at the motor end and 9° at the stone end because of how the bearing pedestals landed. With unequal angles the two joints don't fully cancel, and you can end up with worse ripple than a single joint plus a twice-per-rev bending moment in the intermediate shaft that wears it out fast.
Rule of thumb: if you can guarantee the two angles match within 1°, run a double joint. If not, run a single joint at a smaller angle or rework the bearing positions.
The two yokes mounted on the intermediate shaft must be in the same plane — that is, the fork arms of one yoke and the fork arms of the other point in the same direction along the intermediate shaft. The yokes on the driver and driven shafts then end up 90° rotated relative to each other, which is what people usually mean by "phased 90°."
Get this wrong by 90° and you have stacked the ripples instead of cancelling them — output ripple roughly doubles. The diagnostic check is simple: scribe a reference line down the intermediate shaft passing through both yoke arm centres. If you can lay a straightedge along the shaft and it touches one arm of each yoke, the phasing is correct.
Almost always a misaligned yoke bore or a bent yoke arm. The Cardan kinematics put equal load on opposite arm pairs over a revolution, so symmetric wear is normal. If one arm is wearing two or three times faster, the cross is sitting cocked and that arm is taking a side load it was never designed for.
Pull the joint, set the cross on a surface plate, and check that all four arms are coplanar within 0.1 mm. Then check the yoke bores for concentricity to the shaft. The other usual suspect is a missing or under-sized woodruff key letting the yoke rock on its shaft under torque reversal — tighten that and the wear pattern usually evens out.
At 4° and 90 RPM, a flexible disc coupling is the better answer. The Goubet would give you about 0.25% speed ripple, which is fine, but the disc coupling delivers near-zero ripple, no lubrication schedule, and a 30,000+ hour service life with nothing to grease. The Goubet earns its place above roughly 5-6°, where a disc coupling starts to fatigue the disc pack.
The crossover point in our shop's experience is around 4-5°: below that, disc; above that, Goubet or one of its modern descendants. If you have 4° today but suspect the building will settle and the angle will grow, install the Goubet now and save yourself a rebuild.
Heat at the trunnions at modest angles points to lubrication starvation or excessive bushing preload. The original Goubet design relied on continuous gravity feed from oil cups — if those cups have run dry, are blocked with old grease, or someone converted to grease fittings without enlarging the bushing clearance, you get metal-on-metal contact at the trunnion arm.
Stop the drive, let it cool, and check bushing clearance with a feeler gauge. You want 0.04-0.08 mm diametral. Anything below 0.03 mm and the bushing is binding — the cross can't articulate freely, friction climbs, and heat builds. Re-ream to clearance, refill the oil cups, and the temperature should drop within 30 minutes of running.
Treat any unmarked vintage joint as derated until proven otherwise. The original cast iron crosses had ultimate tensile strength around 150 MPa — about a third of modern cast steel. Measure the cross arm root diameter, calculate bending stress at your expected torque, and keep the result below 25 MPa to give yourself a 6× safety factor against a sudden shock load.
For a 50 mm cross arm, that works out to roughly 600 Nm continuous, which is well below what a modern equivalent would handle. If the application needs more, have a new cross machined from 4140 steel — the rest of the joint usually survives fine, it's the cross that fails first on these antiques.
References & Further Reading
- Wikipedia contributors. Universal joint. Wikipedia
Building or designing a mechanism like this?
Explore the precision-engineered motion control hardware used by mechanical engineers, makers, and product designers.
