An Almond Angular Shaft Coupling is a torque-transmitting joint that connects two intersecting shafts using an almond- or lens-shaped sliding block trapped between two slotted yokes fixed to each shaft. As the input shaft rotates, the block slides inside both yoke slots simultaneously while the slot axes pivot through their fixed shaft angle, transferring rotation across the intersection. Designers use it where a Cardan joint is too bulky and a flexible coupling can't hold positional accuracy. You'll find it inside watch fusee mechanisms, drafting instruments, and small geared scientific apparatus where shaft angles between 5° and 30° are fixed by housing geometry.
Almond Angular Shaft Coupling Interactive Calculator
Vary shaft angle, input speed, and rotation phase to see the Cardan-like output speed ripple in an almond angular shaft coupling.
Equation Used
This calculator models the almond coupling velocity ripple using the Cardan-like relation noted in the article. The shaft intersection angle beta sets the size of the once-per-revolution speed variation, while theta selects the current input rotation phase.
- Uses the Cardan-like velocity relation described for larger almond coupling angles.
- Shaft angle is fixed and shafts intersect at one point.
- No backlash, slot wear, compliance, or friction losses are included.
- Average speed over one revolution equals input speed.
How the Almond Angular Shaft Coupling Actually Works
The almond coupling is a close cousin of the Oldham coupling, but instead of compensating for parallel offset between two shafts, it compensates for angular intersection. Two yokes — each a flat slotted disc — sit on the ends of the two shafts. Between them rides the almond, a lens-shaped sliding block whose two flat faces engage the two slots at the shaft intersection angle. As the assembly rotates, the almond slides back and forth inside both slots once per revolution, and that sliding action is what allows the two yokes to share angular velocity even though their rotation axes meet at, say, 15°.
The geometry is unforgiving. The almond's two working faces must lie on planes that bisect the shaft intersection angle, and the slot widths must match the almond thickness within roughly 0.02 mm on instrument-grade builds. If the slot is too tight, the almond binds at the extremes of its travel and the coupling stalls or chatters. If the slot is too loose, you get backlash that shows up as a periodic angular error — once per revolution — which is fatal in any positioning application. This is why precision builds use hardened tool steel almonds lapped to size against the bronze yoke slots.
Unlike a constant velocity joint, the almond coupling does not deliver perfectly uniform output velocity at large angles. Above about 20° intersection the angular velocity ratio between input and output develops a measurable second-harmonic ripple, similar to a single Cardan joint. Below 10° the ripple is negligible for most instrument work. Common failure modes are slot wear (showing as growing backlash over service life), almond chipping at the slot edges if the coupling is shock-loaded, and seizure if lubrication fails — the sliding speed at the slot faces is low but the contact pressure is high.
Key Components
- Driving Yoke: The slotted disc fixed to the input shaft. The slot is machined diametrically across the face, typically 3 to 8 mm wide on instrument couplings, with a width tolerance of ±0.01 mm against the almond. Slot depth must exceed half the almond's maximum sliding travel or the block walks out of engagement at the top of stroke.
- Driven Yoke: Identical in geometry to the driving yoke but mounted on the output shaft at the intended intersection angle. The two slot axes must be coplanar with the plane that bisects the shaft angle, otherwise the almond cocks under load and wear accelerates dramatically.
- Almond (Sliding Block): The lens-shaped intermediate element. The two flat working faces are ground parallel to within 5 µm and angled to match the shaft intersection. Hardened tool steel (58–62 HRC) running in bronze slots is the standard pairing for long service life — typical running clearance is 0.01 to 0.03 mm per face.
- Retaining Cage or Housing: A light cage or housing that keeps the almond captive between the yokes when the coupling is unloaded or vertical. Without it the block can fall out during assembly or when a shaft is disconnected for service.
Industries That Rely on the Almond Angular Shaft Coupling
You won't find almond couplings on a CNC spindle or a conveyor drive — they live in low-power, high-precision territory where shaft angles are dictated by housing constraints and a Cardan joint would be too bulky or too imprecise. The almond's appeal is compactness and zero-backlash potential when ground to instrument tolerances. Below about 50 W of transmitted power and 30° of intersection, it competes directly with miniature universal joints and bevel gears.
- Horology: Mechanical watch fusee-and-chain mechanisms in vintage Breguet and modern Lange & Söhne pieces use almond-style angular couplings to transfer torque from the mainspring barrel to the going train across small intersection angles.
- Scientific Instruments: Carl Zeiss microscope stage drives historically used almond couplings between the focus knob shaft and the rack pinion shaft where the housing forced a 12° to 18° angle.
- Drafting and Drawing Tools: Pantograph linkages and parallel-rule drafting machines from Kern Aarau and Haff used almond couplings to transfer rotation between the drawing arm and the scale knob through a fixed angle.
- Aerospace Instrumentation: Mechanical altimeter and gyro-repeater shaft drives in legacy avionics — Sperry and Smiths Industries panels — used almond couplings to route low-torque rotation around tight panel geometry.
- Tabletop Robotics and Education: Small geared mechanism kits and university kinematics demonstrators use almond couplings as a teaching example for intersecting-shaft motion transfer alongside Oldham and Schmidt couplings.
- Optical Equipment: Theodolite and surveying-instrument fine-adjust shafts from Wild Heerbrugg and Kern used almond couplings where the eyepiece prism geometry forced the adjustment shaft into a non-coaxial path.
The Formula Behind the Almond Angular Shaft Coupling
The output-to-input angular velocity ratio of an almond coupling varies through each revolution exactly like a single Cardan joint, because both share the same underlying intersecting-shaft geometry. The formula tells you how much velocity ripple to expect at a given shaft angle β. At small angles (β under 10°) the ripple is below 1.5% peak-to-peak — invisible in most instrument work. At the nominal range (15° to 20°) ripple reaches 4–7%, still acceptable for hand-driven adjustment knobs. Push past 25° and ripple climbs above 10%, which you'll feel as a pulsing resistance at the input knob. The sweet spot for almond couplings is 8° to 18°.
Variables
| Symbol | Meaning | Unit (SI) | Unit (Imperial) |
|---|---|---|---|
| ωout | Instantaneous angular velocity of the driven shaft | rad/s | RPM |
| ωin | Angular velocity of the driving shaft (constant) | rad/s | RPM |
| β | Intersection angle between the two shafts | degrees or radians | degrees |
| θ | Instantaneous rotation angle of the driving shaft | degrees or radians | degrees |
Worked Example: Almond Angular Shaft Coupling in a vintage theodolite fine-focus drive
You're restoring a Wild T2 theodolite and the almond coupling between the focus knob and the prism-housing pinion needs replacement. The housing geometry fixes the shaft intersection angle at 15°. Input speed at the knob is a steady 60 RPM during a focus sweep. You need to know the velocity ripple the user will feel at the pinion across the typical operating range of 8° to 22° so you can decide whether the replacement coupling needs re-shimming.
Given
- ωin = 60 RPM
- βnom = 15 degrees
- βlow = 8 degrees
- βhigh = 22 degrees
Solution
Step 1 — at the nominal 15° intersection, compute the ratio at θ = 0° (where the slot lies in the plane of the shaft angle, giving the maximum ratio):
Step 2 — at θ = 90° the ratio reaches its minimum:
So peak-to-peak ripple at 15° is (1.0364 − 0.9659) / 1.0 ≈ 7.0%. At 60 RPM input, the output oscillates between 57.95 and 62.18 RPM twice per revolution. A user turning the knob feels this as a faint pulsing — noticeable on a precision focus but not objectionable.
Step 3 — at the low end of the typical range, β = 8°, the ripple shrinks dramatically:
That's effectively imperceptible — the focus would feel as smooth as a coaxial coupling. At the high end, β = 22°:
That's the regime where the user feels distinct pulsing every half-revolution and the focus knob develops a "sticky" character on slow sweeps. The Wild factory chose 15° deliberately — far enough below 22° to keep ripple acceptable, far enough above 8° to fit the prism housing.
Result
Nominal velocity ripple at 15° is approximately 7% peak-to-peak, with the output speed swinging from 57. 95 to 62.18 RPM around the 60 RPM input. To the user this feels like a faint, even pulsing through the focus knob — present but not distracting. At 8° the ripple collapses to under 2% (effectively invisible), while at 22° it climbs above 15% and turns into a noticeable cogging sensation. If your restored coupling feels worse than this — say, a hard catch once per revolution rather than a smooth pulse — check three things in order: (1) the almond's two working faces are not parallel to within 5 µm (cocking under load), (2) the slot width has worn beyond 0.04 mm clearance per face giving second-harmonic backlash, or (3) the yoke slots are not coplanar with the shaft-angle bisector plane, which manifests as a hard spot at one specific θ angle.
When to Use a Almond Angular Shaft Coupling and When Not To
The almond coupling competes with three other intersecting-shaft solutions in the low-power precision space. Each handles the trade-off between angle range, velocity uniformity, backlash, and cost differently. Pick on the dimension your application is most sensitive to.
| Property | Almond Angular Coupling | Single Cardan (Universal) Joint | Miniature Bevel Gear Pair |
|---|---|---|---|
| Practical shaft angle range | 5° to 25° (sweet spot 8°–18°) | 0° to 45° | Fixed by gear cut, typically 90° (or 45° for mitres) |
| Velocity ripple at 15° | ~7% peak-to-peak | ~7% peak-to-peak (identical kinematics) | 0% (constant ratio) |
| Backlash when new | 0.01–0.03 mm slot clearance, near-zero with lapping | 0.05–0.15 mm at the cross trunnions | 5–15 arc-minutes typical, 1–2 arc-min on ground gears |
| Torque capacity at 10 mm OD | ~0.5 N·m instrument-grade | ~2 N·m commercial miniature | ~3–5 N·m for steel mitre pair |
| Axial length of joint | Very short — typically 1× to 1.5× shaft diameter | Medium — 3× to 4× shaft diameter | Short — depends on gear face width, often 1× to 2× |
| Cost (instrument grade, single piece) | Moderate — lapped almond and matched yokes | Low — mass-produced miniature U-joints from NB or Belden | High for ground precision pairs (Reliance Precision, KHK) |
| Lifespan in continuous service | 10⁶–10⁷ revolutions before slot wear shows | 10⁵–10⁶ revolutions before trunnion wear | 10⁷–10⁸ with proper lubrication |
Frequently Asked Questions About Almond Angular Shaft Coupling
A smooth ripple comes from the inherent cos(β) / (1 − sin²(β)cos²(θ)) variation and is symmetric across the revolution. A notch at one specific angle means asymmetry — almost always one of two causes. First, the two yoke slots are not coplanar with the shaft-angle bisector plane; one slot is rotated a degree or two around its shaft axis, so the almond hits a tight spot once per revolution. Second, the almond's two working faces are not symmetric about its centreline — a manufacturing error in the lens grind. Mark the input shaft, rotate by hand, and note exactly where the notch occurs. If it's at the same θ on both forward and reverse rotation, it's geometric; if it shifts, it's lubrication or a debris particle.
You can, but the design envelope shrinks fast. The sliding velocity at the slot face is proportional to input RPM and to sin(β). At 60 RPM and 15° the slot face sees a peak sliding speed of only a few mm/s, well within boundary-lubrication regime. Push to 600 RPM and you're at 50–60 mm/s, where the bronze slot starts to need a proper grease film or it will gall. Above ~1000 RPM almond couplings simply aren't the right tool — switch to a Cardan joint or a Schmidt offset coupling depending on your geometry.
Three deciding factors. (1) Backlash — if your application is positioning-critical and the load is bidirectional, the almond wins because a lapped slot fit gives essentially zero backlash whereas a miniature U-joint always has trunnion play. (2) Axial length — the almond is roughly a third the length of a U-joint of the same shaft size, which matters in instruments. (3) Torque — if you need more than about 1 N·m through a 10 mm shaft, the U-joint's pin-and-trunnion geometry beats the almond's line contact. Watch and instrument work goes almond; hobby robotics and small mechatronics goes U-joint nine times out of ten.
The formula assumes rigid yokes, zero slot clearance, and exact angle alignment. Real-world deviation usually comes from elastic deflection of the yokes under the contact load — instrument couplings often use 0.5 mm thick brass yokes that flex visibly under torque, adding a flexure-dependent ripple component on top of the geometric one. Check by spinning the coupling unloaded and comparing to the loaded measurement; if the difference is mostly load-dependent, it's yoke flex. The fix is thicker yokes or a stiffer yoke material, not a tighter almond fit.
Classic symptom of slot-bottom contact. Under load the almond rocks slightly inside the slot and its tip touches the bottom of the slot once per revolution at the maximum-travel point. Slot depth needs to exceed half the almond's peak sliding travel, which scales with shaft radius × tan(β). At 15° on a 5 mm shaft that's about 0.67 mm of one-side travel — your slot needs at least 1.5 mm depth with margin. If you've matched the original spec exactly and still see stalling, the almond is oversized in its long axis — measure across the lens tips and compare to the original drawing.
It self-centres kinematically because both slots constrain it, but only if the two yokes are positioned at the correct axial spacing — the spacing where the almond's two flat faces simultaneously contact both slots without preload. Set the spacing too tight and the almond is jammed (stalls). Set it too loose and the almond rattles axially and chips its edges within hours. The correct spacing is the almond's face-to-face distance measured along the shaft-angle bisector, which is shorter than its straight thickness by a factor of cos(β/2). Shim the bearing housings to within 0.05 mm of this dimension on instrument builds.
References & Further Reading
- Wikipedia contributors. Coupling. Wikipedia
Building or designing a mechanism like this?
Explore the precision-engineered motion control hardware used by mechanical engineers, makers, and product designers.
