A Hirth joint is a coupling with radial triangular teeth machined into two opposing flat faces, meshing concentrically when bolted together to transmit torque between coaxial shafts. A well-cut Hirth joint repeats centring within 2-3 µm and handles torques north of 25,000 Nm in aerospace shaft stacks. The radial geometry forces self-alignment under axial preload, eliminating the slop you get with keyways or splines. You see them inside Pratt & Whitney turbine rotors, Schaublin lathe spindles, and bicycle crank spindles such as the Campagnolo Ultra-Torque.
Hirth Joint Interactive Calculator
Vary torque, mean radius, flank angle, safety factor, and bolt preload to see required clamp force and preload margin.
Equation Used
The calculator estimates the axial preload required to keep a Hirth joint seated under torque. Torque creates a separating force through the serrated flank angle; multiplying that force by the safety factor gives the required clamp preload. The design torque shows what the entered bolt preload can carry with the same safety factor.
- Flank angle theta is the effective wedge angle converting torque load into axial separating force.
- Tooth contact is clean, concentric, and uniformly shared around the joint.
- Friction, fatigue, local tooth stress, face flatness error, and bolt relaxation are not included.
How the Hirth Joint Actually Works
A Hirth joint works by pressing two faces of identical radial teeth together along the shaft axis. Each tooth has a triangular profile cut from the centreline outward, so when you bring the faces together with a centre bolt or stud stack, the flanks slide against each other and the joint self-centres as the axial preload climbs. That self-centring is the whole point — there is no separate dowel, no key, no fitted bore. The teeth do the locating job and the torque-transmitting job at the same time.
The geometry only works if the tooth flanks are ground to matching pressure angles and the face is flat to within a couple of microns. We typically see flank angles of 60° on coarser industrial joints and 30° on precision spindle work where higher tooth count gives finer indexing. If the face is bowed by even 5 µm, the outer teeth carry all the load and the inner teeth float — you get fretting on the OD teeth and the joint loses concentricity within a few hundred load cycles. Same problem if the bolt preload drops: under reversing torque the faces micro-slip, the flanks polish, and runout creeps up until the assembly chatters.
Failures almost always trace back to three causes. Insufficient preload — the bolt clamp must exceed the peak torque-induced separating force, usually with a 1.5× safety factor. Contamination between faces — a single 20 µm chip will tilt the joint and ruin centring. And re-cutting a worn face without re-cutting its mate, which leaves a tooth-pitch mismatch the eye cannot see but the runout indicator will scream about.
Key Components
- Radial Face Teeth: Triangular teeth cut radially from the shaft centreline outward on both mating faces. Tooth count typically runs 24 to 360 — higher count gives finer angular indexing but smaller individual tooth load capacity. Flank angle is usually 60° for power transmission and 30° for precision indexing applications.
- Centre Bolt or Tie Rod: Provides the axial preload that locks the teeth together. Preload must exceed peak separating force by at least 1.5× — for a 200 mm Hirth carrying 5,000 Nm that means roughly 80-120 kN of bolt clamp. Use a 12.9-grade fastener and torque to spec, not by feel.
- Mating Hub Faces: The flat reference surfaces the teeth are cut into. Flatness must hold within 2-5 µm across the face diameter. We grind these after tooth cutting on a CNC face grinder — any waviness translates directly into runout and uneven tooth load.
- Pilot Diameter (optional): Some Hirth designs add a short cylindrical pilot at the centre to ease assembly alignment. Clearance is typically 5-10 µm on diameter — tight enough to guide the teeth into mesh, loose enough that the teeth still do the final centring job rather than the pilot.
Real-World Applications of the Hirth Joint
Hirth joints earn their place wherever you need to take a coupling apart repeatedly without losing concentricity, or where torque density is too high for a key. The repeatable indexing is what sells them in machine tools and turbines — pull the joint apart, service the part behind it, bolt it back together, and the runout returns to within a couple of microns of where it started. That is something a tapered fit or a keyway simply cannot do.
- Aerospace: Compressor and turbine disc stacks in jet engines such as the Pratt & Whitney F100 and Rolls-Royce Trent series, where multiple discs are tied together by a single central bolt and the Hirth (or close cousin curvic) couplings transmit shaft torque between stages.
- Machine Tools: Indexing heads and spindle nose couplings on precision lathes — Schaublin 102 and 125 spindles use Hirth-style face couplings for tool-changer interfaces that must repeat within 2 µm.
- Cycling: Campagnolo Ultra-Torque and Fulcrum crankset spindles split at the bottom-bracket centreline and rejoin via a Hirth joint clamped by a single 10 mm bolt, giving stiffness equal to a one-piece spindle.
- Industrial Gas Turbines: Siemens SGT-800 and GE Frame 7 rotor shafts use stacked disc Hirth/curvic couplings to transmit 50+ MW of shaft power while allowing the rotor to be assembled from individually forged discs.
- Heavy Machinery: Large rolling-mill drive shafts and ship propeller shafts use Hirth flange couplings up to 1 m diameter, transmitting torques over 500,000 Nm with bolt circles of 24 to 48 fasteners.
- Robotics: Tool-change interfaces on industrial robots — ATI Industrial Automation tool changers use Hirth-style face teeth on the master/tool plate interface for sub-10 µm repeatability across thousands of swap cycles.
The Formula Behind the Hirth Joint
The number that matters most to a Hirth joint designer is the torque capacity per unit of bolt preload — how much shaft torque the tooth flanks can carry before they start to slip apart axially against the clamp. At the low end of the typical preload range, the joint is undersized and slips under peak transient torque. At the high end you start yielding the centre bolt or crushing the tooth tips. The sweet spot lives in the middle third, where the clamp force gives you a 1.5-2× margin over peak torque-induced separation force. The formula below tells you where that margin sits.
Variables
| Symbol | Meaning | Unit (SI) | Unit (Imperial) |
|---|---|---|---|
| T | Maximum transmissible torque before flank slip | Nm | lb·ft |
| Fp | Axial bolt preload clamping the joint | N | lbf |
| Rm | Mean radius of the tooth pattern (average of OD and ID) | m | in |
| α | Tooth flank included angle (full angle between two flanks of one tooth) | degrees | degrees |
| η | Friction efficiency factor accounting for flank friction (typically 0.85-0.95) | dimensionless | dimensionless |
Worked Example: Hirth Joint in a CNC indexing rotary table
You are sizing the Hirth coupling on the rotary B-axis of a 5-axis CNC mill — a 200 mm diameter face coupling between the indexer body and the worktable, with a 60° flank angle, mean tooth radius of 85 mm, and a single M20 12.9-grade centre bolt providing axial preload. Spindle peak cutting torque demands 2,200 Nm at the workpiece, and you need to know whether the joint holds at nominal preload, at the low end of practical preload, and at the high end before the bolt yields.
Given
- Rm = 0.085 m
- α = 60 degrees
- η = 0.90 dimensionless
- Fp,nom = 150,000 N (nominal M20 12.9 preload at ~75% proof)
- Trequired = 2,200 Nm
Solution
Step 1 — compute the geometric factor from the flank angle. With α = 60°, the half-angle is 30°, so we need tan(90° − 30°) = tan(60°):
Step 2 — compute the nominal torque capacity at the design preload of 150 kN, which is what an M20 12.9 bolt gives you when torqued to roughly 75% of proof load:
That is nearly 9× the 2,200 Nm cutting demand, which is exactly the comfort margin you want on a precision rotary axis where any flank slip means a scrapped part.
Step 3 — at the low end of the practical preload range, say 60 kN (a soft-torqued bolt or one that has lost preload from gasket creep behind the joint):
Still 3.6× the cutting demand, but the margin is closing. At this preload a heavy interrupted cut or a tool crash sends the joint into the danger zone — you would see micro-fretting on the tooth flanks within a few hundred operating hours.
Step 4 — at the high end, around 180 kN, you are pushing the M20 bolt past 90% of proof load:
Theoretical capacity climbs to almost 24 kNm but you have lost most of your bolt fatigue margin. Any cyclic axial load — thermal growth, tool-change shock — will start the bolt down a fatigue crack. The sweet spot stays at the nominal 150 kN preload.
Result
Nominal torque capacity is 19,876 Nm against a 2,200 Nm cutting demand — roughly a 9× safety factor, which is appropriate for a precision rotary axis where flank slip ruins concentricity permanently. At 60 kN preload capacity drops to 7,950 Nm and the joint enters the fretting-risk zone under interrupted cuts; at 180 kN you reach 23,851 Nm but the centre bolt loses fatigue margin and becomes the weak link. If your indexer measures runout above 5 µm after a few hundred cycles, the most common causes are: (1) bolt preload loss from gasket or paint film creep behind the joint flange, drop preload by 30-40% in the first 50 cycles, (2) a single 10-20 µm chip trapped between faces tilting the entire joint, or (3) tooth-pitch mismatch from re-grinding one face without re-cutting its mate, which loads only the OD teeth and polishes them visibly within hours.
When to Use a Hirth Joint and When Not To
Hirth joints compete with curvic couplings, keyed shafts, and tapered shrink fits whenever you need to transmit torque between coaxial shafts. The right choice depends on whether you need disassembly repeatability, raw torque density, or low cost — these three drive most decisions.
| Property | Hirth Joint | Curvic Coupling | Keyed Shaft |
|---|---|---|---|
| Re-assembly concentricity | 2-3 µm typical | 1-2 µm typical (ground in matched pairs) | 20-50 µm — keyway slop dominates |
| Peak torque capacity (200 mm joint) | ~20,000 Nm | ~25,000 Nm | ~8,000 Nm before key shears |
| Manufacturing cost | Moderate — single-pass radial cutter | High — requires Gleason curvic grinder | Low — standard keyway broach |
| Indexing positions | Equal to tooth count, 24 to 360 | Equal to tooth count, typically 24-60 | 1 (single key) or 2-4 (multi-key) |
| Sensitivity to face contamination | High — 20 µm chip ruins centring | High — same vulnerability | Low — key takes load regardless |
| Typical service life under reversing torque | 10⁸+ cycles with proper preload | 10⁸+ cycles, ground convex form helps | 10⁶-10⁷ cycles before keyway elongates |
| Best application fit | Precision spindles, indexers, bike cranks | Aero turbine rotors, gas turbines | Low-cost industrial drives, agricultural |
Frequently Asked Questions About Hirth Joint
The most likely cause is contamination on one or two tooth flanks rather than wear. A single chip or fingerprint of grease in the wrong place tilts the joint by a few arc-seconds, which translates to several microns at the OD. Strip the joint, clean both faces with solvent and a soft brush — never wipe with a rag because lint catches in the tooth roots — and re-assemble dry or with a thin film of light oil only.
If runout still measures high after cleaning, check face flatness with a granite reference and dial indicator. Heat from grinding sometimes leaves a residual bow that only shows up after the part fully cools, hours after machining.
Curvic gives you slightly better repeatability and higher torque density because the teeth are ground with a convex form that distributes load along the flank length. The trade is cost — curvic requires a Gleason curvic grinder, which not every shop has. Hirth teeth can be cut on a CNC mill or hobbed on a horizontal mill with a single-tooth radial cutter, putting them in reach of any decent toolroom.
For aero gas turbines and ground-power turbines above ~10 MW, curvic earns its premium. For machine tool spindles, indexers, and shaft couplings under 50 kNm, Hirth is the right answer almost every time.
Almost always this is gasket or paint creep on the back side of one of the joint flanges, not bolt thread relaxation. Anything compressible in the clamp stack — a paper gasket, a thick paint film, even a thermal interface pad — will creep under preload and drop bolt tension by 20-40% over the first dozen thermal cycles.
The fix is to keep the clamp stack metallic. Mask paint off the mating face area, eliminate gaskets where possible, and re-torque the bolt after the first heat cycle. On critical assemblies, use a load-indicating washer or measure bolt stretch directly rather than torque.
Use a friction efficiency η of 0.85 as a conservative default for steel-on-steel teeth in the absence of test data. Rearrange the torque equation to solve for Fp, multiply the result by 1.5 as a clamp safety factor, and check that the resulting preload sits between 60% and 80% of the bolt's proof load. If it falls below 60%, you are oversized on the bolt and can drop a size; if it sits above 80%, step up the bolt grade or diameter rather than torquing harder.
For reversing-torque applications, use 2.0 as the safety factor instead of 1.5 — fatigue from cyclic flank slip is what kills these joints, not single-event overload.
No, and this is one of the most common shop-floor mistakes with these joints. The two faces share a tooth pitch that was originally cut as a matched pair. If you re-cut one face by even 0.1 mm of stock removal, the tooth pitch on that face shifts relative to the original geometry, and only the OD teeth or only the ID teeth will mesh — never both at once.
The correct repair is to re-cut both faces in one setup, ideally on the same machine within the same thermal window. If only one half is recoverable, scrap both and remake the pair.
The creak comes from micro-slip on the Hirth tooth flanks under reversing pedal load. Two causes account for almost all cases: the centre bolt has lost preload (Campagnolo specs 42 Nm — check it with a calibrated wrench, not a multi-tool), or moisture has worked its way between the tooth flanks and is causing stick-slip as the steel teeth fret.
The fix is to disassemble, clean both faces with solvent, apply a thin film of anti-seize or grease specifically rated for steel-on-steel fretting (Loctite LB 8150 works well), and re-torque to spec. The grease film does not reduce torque capacity meaningfully but it kills the fretting that causes the creak.
Tooth count drives indexing resolution directly — 360 teeth gives you 1° indexing, 72 teeth gives you 5°, 24 teeth gives you 15°. Pick the coarsest count that still hits every angular position you need, because finer tooth count means smaller individual teeth and lower torque-per-tooth capacity.
For tool changers and rotary indexers, 72 or 96 teeth is the common sweet spot — enough resolution for typical fixturing angles, enough flank area to handle real cutting loads. Go to 360 only when you genuinely need 1° resolution and you can keep the torque budget low.
References & Further Reading
- Wikipedia contributors. Hirth joint. Wikipedia
Building or designing a mechanism like this?
Explore the precision-engineered motion control hardware used by mechanical engineers, makers, and product designers.
