Gears & Gearboxes: The Complete Engineering Guide to Involute Geometry, Load Rating, Gear Manufacturing, and Reduction Design

A dense equation-driven engineering deep-dive into gears and gearboxes — the involute math with module, pressure angle, undercut and profile-shift rules, a fully worked contact-ratio calculation for a 20/40 pair, Lewis and ISO 6336 bending stress with a worked 3 kW pinion rating (637 N tangential load, 53 MPa root stress), Hertzian contact stress with the Z_E derivation and a 475 MPa case-hardening verdict, the material-and-treatment table from EN8 through 20MnCr5 to POM and sintered steel, failure-mode taxonomy (pitting, scuffing, micropitting, bending fatigue), worm-gear efficiency math with a worked 40:1 self-locking check, backlash and thermal-growth math, eight gear manufacturing routes with tolerances and Indian costs, 3D-printed gear design rules, planetary-gear Willis kinematics with a worked 4:1 stage, a fully worked 10:1 two-stage reducer design, Indian sourcing economics by gear cluster, and a failure-mode troubleshooting matrix.

·

Gears & Gearboxes: The Complete Engineering Guide to Involute Geometry, Load Rating, Gear Manufacturing, and Reduction Design

A gear pair is the highest-torque-density rotary transmission in the machine designer's toolkit: 97–99% efficiency per stage, zero slip, a ratio locked by tooth count, and power densities that belts and chains cannot touch. A single 25 mm-diameter, 5 mm-face-width steel gear can transmit several kilowatts continuously, while a planetary stage packs a 3:1–12:1 reduction into a coaxial envelope the size of the motor flange itself. That is why every serious machine — CNC spindles, robot joints, extruder drives, tractor final drives, watch movements — carries gears somewhere, and why gear design is the highest-leverage skill a mechanical builder can learn.

The mathematics is remarkably compact. One curve (the involute), one pitch rule (module), and two stress equations (root bending, flank contact) predict whether a gear pair will survive 10⁸ cycles or shed teeth in a week. This guide works through that math with fully worked examples, covers materials and heat treatment, manufacturing routes from hobbing to SLS printing, gearbox staging and planetary kinematics, Indian sourcing economics, and a failure-mode matrix keyed to physical root causes.

1. The Gear Taxonomy: What Runs Where

Every gear type is a trade between cost, noise, efficiency, ratio capacity, and axis geometry. The decision table:

Type · Typical ratio per stage · Efficiency · Axis layout · Signature trade-off

Spur · 1:1 – 1:6 · 98–99% · Parallel · Cheapest; noisy above ~10 m/s; zero axial thrust

Helical · 1:1 – 1:10 · 98–99% · Parallel · Quiet, higher contact ratio; axial thrust needs bearings

Straight bevel · 1:1 – 1:5 · 95–98% · 90° intersecting · Simple right-angle drive; noisy

Spiral bevel / hypoid · 1:1 – 1:10 · 94–98% · 90°, offset · Smooth; hypoid offset adds sliding → needs EP oil

Worm · 1:5 – 1:100 · 50–90% · 90° non-intersecting · Huge ratio, possibly self-locking; efficiency collapses at high ratio

Planetary · 1:3 – 1:12 per stage · 97–99% per mesh · Coaxial · Compact, load-shared; tight tolerance and bearing cost

Harmonic (strain wave) · 1:30 – 1:320 · 70–90% · Coaxial · Near-zero backlash; limited shock tolerance

Cycloidal · 1:10 – 1:100 · 80–93% · Coaxial · High shock capacity; backlash from bearings only

Rack & pinion · Linear · 90–95% · Rotation ↔ translation · Gantry and steering; open mesh wears

Selection logic for machine builders. A 3D-printer extruder wants a 3:1 dual-drive with a POM-on-steel spur stage — cheap, quiet, self-lubricating, and the backlash is set by the gear mesh, not the drive. A CNC router's fourth axis at 100 rpm output from a 3000 rpm servo wants a 30:1 planetary or harmonic drive — coaxial, stiff, sub-arc-minute lost motion if you pay for it. A conveyor or hopper drive that must hold position with power off wants a worm stage, because a worm with lead angle below the friction angle is self-locking. And a robot arm shoulder wanting 10 N·m at 300 rpm from a 1 N·m motor wants a 10:1 spur or planetary reducer — the worked design in Section 7.

2. Involute Geometry: The Math That Makes Gears Work

2.1 Module, pitch, and the defining numbers

The entire gear is scaled by one number: the module m (mm). The pitch diameter — the imaginary circle where teeth mesh — is

where z is the tooth count. Circular pitch (tooth-to-tooth spacing along the pitch circle) and diametral pitch (the imperial reciprocal) follow:

Standard modules follow ISO 54 preferred series — 0.5, 0.8, 1, 1.25, 1.5, 2, 2.5, 3, 4, 5 mm — and you should never design to a non-standard module unless a mating part forces it. A non-standard module means a custom hob or cutter and a lifetime of expensive replacements.

The tooth height splits around the pitch circle: addendum h_a = m above, dedendum h_f = 1.25\,m below (full-depth standard), for a working depth of 2m and a clearance of 0.25m at the root. The pressure angle \phi is the angle of the tooth flank at the pitch point. It is the single most consequential standard choice:

For a 20° full-depth pair on standard center distance, the center distance is simply

The base circle — the circle from which the involute is generated — sits inside the pitch circle:

2.2 The involute curve and why it wins

An involute is the path traced by the end of a taut string unwinding from the base circle. Its polar equation is defined through the involute function:

The genius of the involute is conjugate action with center-distance insensitivity: two involute flanks roll against each other with a constant angular-velocity ratio i = z_2/z_1 even if the center distance drifts by a few tenths of a millimetre. The pressure angle changes slightly, the backlash changes slightly — but the velocity ratio does not. That is why involute gearing displaced cycloidal gearing for power transmission (cycloidal remains in watch mechanisms and rotary lobe pumps, where its lower friction matters more than its intolerance to center-distance error). In practice, center-distance drift manifests as measurable backlash change, not ratio error:

A 0.1 mm center-distance error on a 20° pair opens about 2 \times 0.1 \times 0.364 = 0.073 mm of backlash — a full precision-class budget. Housing bore positions are not free.

2.3 Undercut and profile shift

With 20° full-depth geometry, any pinion below 17 teeth has its root undercut — the cutter removes material inside the involute near the root, thinning the tooth exactly where bending stress peaks:

The fix is profile shift: cut the tooth with the cutter pulled back by x\,m, thickening the root. The minimum shift to eliminate undercut:

A 12-tooth 20° pinion therefore needs x = +0.294 \approx +0.3. Profile shift changes the center distance by (x_1 + x_2)\,m, so a shifted pair needs a custom center distance — or equal-and-opposite shifts (x_1 = -x_2) to keep it standard. Note the practical ceiling: x > +0.5 produces a needle-pointed tip with drastically reduced addendum contact.

2.4 Contact ratio

The contact ratio \varepsilon is the average number of teeth in mesh. For a spur pair:

where r_a is the tip radius and r_b the base radius of each gear. A pair must keep \varepsilon > 1.0 (ideally \geq 1.2) or the load is never carried by two teeth — the gear "clicks" from tooth to tooth.

Worked example — the ubiquitous 20/40 pair, module 2: d_1 = 40 mm, d_2 = 80 mm, a = 60 mm, 20° pressure angle. Tip radii: r_{a1} = 22 mm, r_{a2} = 42 mm. Base radii: r_{b1} = 20\cos 20° = 18.79 mm, r_{b2} = 37.59 mm.

\varepsilon = 1.63 means load is shared by two teeth 63% of the time — the standard result, and the reason 20/40 is a sane default ratio. Helical gears add an overlap (face) contact ratio \varepsilon_\beta = b \sin\beta / (\pi m_n) that can push total contact above 2.5, which is why helicals run quieter and rate higher than spurs of equal size.

One more geometry rule that costs nothing and extends life: choose hunting tooth counts — \gcd(z_1, z_2) = 1. A 20/40 pair has each pinion tooth meeting the same two wheel teeth every revolution (gcd = 20); a 21/40 pair cycles every tooth against every tooth, spreading wear and erasing periodic errors. The "prime tooth count" folklore in 3D-printing extruder design is this exact principle.

3. Tooth Loads and Strength: Lewis, ISO 6336, and a Worked Rating

3.1 The tangential load

All gear rating starts from the tangential force at the pitch circle:

Worked example — a 3 kW, 1500 rpm pinion, module 3, 20 teeth: d_1 = 60 mm, \omega_1 = 2\pi \cdot 1500/60 = 157.1 rad/s, so T_1 = 3000/157.1 = 19.1 N·m and

Pitch-line velocity v = \omega d/2 = 4.7 m/s — the boundary between oil-bath and splash lubrication, as Section 5 discusses. This 637 N is the load every equation below consumes.

3.2 The Lewis bending check

Wilfred Lewis's 1892 model treats a tooth as a cantilever beam of parabolic (uniform-stress) profile, failing at the root:

where b is face width and Y the Lewis form factor — a function of tooth count and pressure angle only. For 20° full-depth teeth, Y ranges from ~0.29 (12 teeth) through 0.322 (20 teeth) to ~0.48 (rack). Taking b = 18 mm (the rule-of-thumb 6m to 16m window, here 6m):

Against an EN8 (C45) through-hardened gear with root fatigue limit \sigma_{F\,lim} \approx 250 MPa, that looks like a factor of ~7 — and it is misleadingly optimistic, because Lewis's parabolic beam ignores the root fillet stress concentration, which typically doubles the real peak stress. The modern standard factors it in properly.

3.3 ISO 6336 bending — the real factor stack

ISO 6336 (the international equivalent of AGMA 2001-D04; the ISO Y_F plays the role of AGMA's J geometry factor) rates root bending as:

\sigma_F = 53 MPa versus the Lewis quick-check's 37 MPa — the 45% difference is the fillet stress concentration Lewis ignores. Acceptance test: \sigma_F \leq \sigma_{F\,lim}/S_{F\,min}, with S_{F\,min} = 1.25–1.5. EN8 at 53 MPa against 250 MPa → passes with ~4.7 margin. A sintered-steel or POM gear would need a re-check; a 12-tooth pinion would not.

3.4 Contact (Hertzian) stress — the pitting limit

Flank pitting, not root breakage, is the usual lifetime limiter of lubricated steel gears. The flank contact is Hertzian line contact, and ISO 6336 rates it as:

with u = z_2/z_1 (always \geq 1). The elasticity factor comes straight from Hertz theory:

for steel-on-steel (E = 206 GPa, \nu = 0.3) — the number appears in every gear standard's tables. Z_H = 2.49 for 20° (the geometry factor Z_H = \sqrt{2/\sin 2\phi}), Z_\varepsilon = \sqrt{(4-\varepsilon)/3} = 0.89, Z_\beta = 1 for spur.

Continuing the 3 kW pinion (u = 2 for a 40-tooth mate, K_{H\beta} = 1.1):

The verdict: EN8 through-hardened at 250–300 HB has \sigma_{H\,lim} \approx 620 MPa — passing with a thin ~1.3 margin that ignores every real-world excursion (shock, contamination, alignment). Case-carburized 20MnCr5 at 58–62 HRC has \sigma_{H\,lim} \approx 1450–1500 MPa — a margin of 3. The standard engineering conclusion: this gear pair deserves case hardening or nitriding, which is exactly why Section 4 spends a whole table on heat treatment. Contact stress is what makes the difference between a gear that dies in 500 hours and one that runs a decade.

4. Materials, Heat Treatment, and Failure Modes

4.1 The material-and-treatment ladder

Material (IS/EN grade) · Treatment · Surface / core · \sigma_{H\,lim} · \sigma_{F\,lim} · Use case

EN8 / C45 (1045) · Through-hardened 250–300 HB, or induction-hardened flanks · — · 620 MPa · 250 MPa · General machinery, machine tools, extruders

EN24 / 4340 · Through-hardened 30–38 HRC · — · 750–800 MPa · 300 MPa · High-strength shafts & gears, aerospace

EN36A / 20MnCr5 · Case-carburized, 58–62 HRC, core 30–40 HRC · 0.8–1.2 mm case · 1450–1500 MPa · 450–500 MPa · Automotive transmissions, industrial reducers

EN19 / 4140 · Nitrided (60–65 HRC, 0.2–0.5 mm case) · Thin, no quench distortion · 1100–1250 MPa · 350 MPa · Precision gears post-machining, low distortion

SG iron (ductile) · As-cast / normalized · — · 500–650 MPa · 180 MPa · Large low-speed gears, damping

POM (acetal) · Molded / machined · — · ~40 MPa* · 25–35 MPa · Light drives, one plastic + one steel mesh

PA66 + 30% GF · Molded · — · — · 35–50 MPa · Higher temp than POM, abrasive to steel mate

PEEK · Molded / machined · — · — · 45–60 MPa · 250°C, chemical, high-purity environments

Sintered Fe-Cu-Ni · PM, 6.8–7.4 g/cm³ · — · ~500 MPa · 200–280 MPa · Millions/year: power tools, seat adjusters

*Plastic gear contact rating is usually replaced by wear-rate and PV-limit analysis rather than Hertz stress.

The case-depth rule ties back to Hertz: for line contact, the maximum subsurface shear stress sits at z \approx 0.78 b_H below the surface (where b_H is the contact half-width). If the hardened case is thinner than that depth, the stress peaks in the soft core and the case spalls off — the classic failure of a shallow induction-hardened flank. The rule of thumb: case depth \geq 3–5× the depth of maximum shear. Carburizing to 0.8–1.2 mm covers typical module-3 contacts; a module-8 mill gear needs 2+ mm.

4.2 Failure-mode taxonomy

  1. Bending fatigue (tooth breakage) — crack initiates at the root fillet on the loaded flank (tensile side), grows over millions of cycles, ends in a characteristic beach-marked fracture. Fixes: larger root radius, shot peening (adds 20–30% to \sigma_{F\,lim}), profile shift, bigger module.
  2. Pitting (contact fatigue) — subsurface cracks at the shear-stress peak grow to the surface and eject flake-shaped craters. Fixes: higher \sigma_{H\,lim} (case hardening), lower K_{H\beta} (better alignment), smoother flanks.
  3. Micropitting (grey staining) — surface-initiated micro-craters at ~10 µm scale, typical of case-hardened gears when the specific film thickness \lambda = h_{min}/\sqrt{Ra_1^2 + Ra_2^2} falls below 1.0. Fixes: better oil film (viscosity, speed), superfinished flanks (Ra < 0.2 µm), EP additives.
  4. Scuffing — instantaneous welding and tearing of asperities at high sliding speed; the failure of hypoids and high-speed spurs with inadequate oil. Flash-temperature analysis (the AGMA/ISO scuffing criteria) predicts it; EP oil prevents it.
  5. Abrasive wear — contaminant-driven polishing that thins teeth into knife edges. Fixes: sealing, filtration, hardened flanks.
  6. Plastic yielding / rolling — overload on soft teeth deforms the flank profile. Fix: harder material, bigger module.
  7. Fretting corrosion — micro-sliding at splines, shrink-fit hubs, and idle gears under vibration. Fix: preload, lubrication, larger interference.

5. Efficiency, Backlash, and Lubrication

5.1 Efficiency by type

Spur and helical meshes lose 1–2% per stage (rolling contact dominates; sliding loss is ~1% per mesh). Bevels lose 2–5%. Planetaries lose 1–3% per mesh but stack two meshes per stage, so a 3-stage planetary box lands around 90–94% overall. Harmonic and cycloidal drives lose 10–30% — real heat to dissipate.

Worm gears are the special case, because the mesh is pure sliding. Efficiency follows:

with lead angle \gamma (for module worms, \tan\gamma = m\, z_1 / d_{w1}, where z_1 is the number of starts) and friction angle \rho = \arctan\mu, with \mu = 0.03–0.08 depending on lubricant and sliding speed.

Worked example — a single-start, 40-tooth wheel (40:1) module-2 worm, worm pitch diameter 25 mm: \tan\gamma = 2/25 = 0.08, so \gamma = 4.57°. With \mu = 0.05, \rho = 2.86°:

61% efficiency — and the other 39% of input power becomes heat inside the box. This is why 40:1 worm drives on conveyors run hot and why a 5:1 worm (two-start, larger lead angle) is dramatically more efficient than a 100:1. Self-locking requires \gamma < \rho statically — our example's 4.57° > 2.86° means it is not reliably self-locking; that property needs a finer lead angle, and it comes with efficiency below 50%. Design worm drives for the efficiency, never rely on self-locking for safety (brakes exist for that).

5.2 Backlash

Backlash j_t is the free rotation of the output with the input locked — the sum of tooth-thickness tolerance, center-distance error, and thermal growth. Budgets:

Thermal worked example: a 60 mm center-distance steel pair in an aluminum housing, 40°C temperature rise. Aluminum grows faster than steel (\alpha_{Al} = 23\times10^{-6}/K vs 12\times10^{-6}/K), opening the center distance by

which becomes \Delta j_t = 2\,\Delta a \tan\phi = 2 \times 0.026 \times 0.364 = 0.019 mm of extra backlash — nearly 40% of a precision budget, purely from thermal mismatch. Aluminum gearboxes need either matched-growth design or a roomier backlash allowance.

Anti-backlash techniques, in increasing cost: spring-loaded scissor gears (split gear, two halves torsionally preloaded — the standard for instrument drives), duplex worms, adjustable-center-distance mounts, and the harmonic/cycloidal route where lost motion is set by bearing preload instead of tooth clearance. Note that anti-backlash only buys static position — under reversing load the spring preload must exceed the load, or the "zero-backlash" claim evaporates.

5.3 Lubrication

6. How Gears Are Made: Eight Manufacturing Routes

Route · Accuracy (ISO 1328) · Cost driver · Best for

Form milling (disc cutter) · 10–12 · Low, one-off · Prototypes, repair, large gears

Hobbing · 7–9 (CNC: 6–7) · Machine + hob · 1–10,000 pcs, the workhorse

Gear shaping · 7–9 · Machine + cutter · Internal gears, clusters

Broaching · 8–9 · Broach tooling · Internal splines, high volume

Skiving (power skiving) · 6–7 · CNC lathe/mill + software · Hard finishing, near-net

Grinding (generating) · 3–6 · Expensive machines · Post-heat-treat precision, automotive

Powder metallurgy · 8–10 · Die set · Net-shape millions: power tools, pumps

Injection molding (POM/PA) · DIN 8–11 · Mold (₹3–15 lakh) · 10k+ pcs, light loads, quiet drives

Hobbing is the reference process: a worm-shaped hob and the gear blank rotate in a synchronized generating motion, cutting any tooth count of the same module with one tool. A modern CNC hobber (₹25–80 lakh new, ₹3–10 lakh for used gear hobbers at Indian machinery dealers) turns a module-2, 40-tooth steel gear in 30–90 seconds. Through-hardening or carburizing follows, and the distortion of heat treatment is corrected by grinding or skiving — which is why case-hardened precision gears cost 3–5× their soft-hobbed equivalents, and why nitriding (minimal distortion, no post-grind) is the cost-effective precision route up to moderate loads.

Powder metallurgy skips cutting entirely: Fe-Cu-Ni powder compacts in a die at 600–800 MPa, sinters at ~1120°C, and emerges at net shape with 92–96% density — gear teeth included. At tooling volumes it drops cost to ₹20–80 per gear and dominates power tools, seat adjusters, and pump gears. Porosity caps \sigma_{F\,lim} at ~250 MPa and makes PM gears slightly quieter (porosity damps vibration).

6.1 3D-printed gears — the honest engineering

Printed gears are genuinely useful, within declared limits:

The realistic verdict: printed gears serve prototypes, jigs, and light automation; machined or molded gears serve production. The crossover is volume and load, not enthusiasm.

7. Gearbox Design: Staging, Planetaries, and a Worked 10:1 Reducer

7.1 Staging rules

Total ratio multiplies across stages: i_{total} = i_1 \times i_2 \times i_3. Three rules govern the split:

  1. Keep each spur/helical stage between 3 and 5 (6 max). Beyond that the wheel grows huge and the pinion teeth get thin.
  2. Split near-equally — i_{stage} \approx i_{total}^{1/n} minimizes overall size, inertia, and maximum tooth stress. A 10:1 box should be ~3.16×3.16, or 2.5×4 as in our worked design, not 8×1.25.
  3. Reverted (coaxial) trains — if input and output shafts must be coaxial, the two stages must satisfy z_1 + z_2 = z_3 + z_4 so both center distances are equal.

7.2 Planetary kinematics — the Willis equation

A planetary stage's motion is governed by the Willis equation relating sun, ring, and carrier speeds:

Worked example — sun 18, planets 18 (×3), ring 54, ring fixed (n_r = 0):

A 4:1 reduction in one coaxial stage. The geometry constraints are exact: z_r = z_s + 2 z_p (54 = 18 + 36) and (z_s + z_r)/N must be an integer for N evenly spaced planets — (18+54)/3 = 24 ✓. Three planets split the torque three ways, which is why planetary stages pack 2–3× the torque density of a comparable spur stage. Stacking two 4:1 planetaries gives 16:1 at ~95% overall efficiency.

7.3 Worked design — a 10:1 two-stage reducer

Requirement: BLDC motor, 3000 rpm, 1.06 N·m continuous → output 300 rpm, target 10 N·m, for a robot-arm joint or rotary axis. Two stages: stage 1 = 20/50 teeth, m = 1.5 (i = 2.5); stage 2 = 20/80 teeth, m = 2 (i = 4). Total i = 10.

Torque flow: T_2 (intermediate) = 1.06 \times 2.5 \times 0.98 = 2.60 N·m; T_{out} = 2.60 \times 4 \times 0.98 = 10.2 N·m ✓.

Bending checks (Lewis, 20°):

Both stages are stress-trivial — even POM would pass the stage-1 bending check. The contact check on stage 2 confirms it:

— comfortably inside EN8's 620 MPa allowable. The engineering lesson: a 10 N·m gearbox fails through bearings, housing stiffness, alignment (K_{F\beta}), and lubrication, not through gear teeth. Spend the design budget on stiff shafts (deflection under 0.01 mm at the mesh), dowel-pinned housing bores, and proper bearing preload — then the tooth math takes care of itself.

Practical notes: a POM-on-steel stage 1 (the BMG extruder pattern) cuts noise and lubrication needs at this load level; keep stages backdrivable (spur/planetary are; high-ratio worms are not) if the joint must be manually teachable; and if 300 rpm output still needs sub-arc-minute positioning, replace the whole train with a 10:1 precision planetary (~₹3,000–8,000 imported) or accept the extra stage count.

8. Indian Sourcing & Economics

India has deep gear-manufacturing clusters and prices that make prototyping cheap — indicative 2026 numbers:

Item · Indicative cost · Notes

Prototype spur gear, m2, 30T, EN8, milled/wire-EDM + hardened · ₹800–1,800/pc · 1–5 pcs, 3–7 day lead

CNC-hobbed small steel gear, batch 10–50 · ₹250–600/pc · Setup ₹500–1,500 amortized over batch

Case-carburized ground gear (20MnCr5), small batch · ₹1,500–5,000/pc · Automotive-grade process chain

PM sintered gear @ 10,000 pcs · ₹20–80/pc · Tooling ₹2–6 lakh, amortized

Injection-molded POM gear @ 50,000 pcs · ₹10–40/pc · Mold ₹4–10 lakh

SLS PA12 printed gear, m2, 30T · ₹400–900/pc · Same-week turnaround

10:1 planetary gearbox, ~10 N·m · ₹3,000–8,000 (import) / ₹8,000–20,000 (custom-built) · Backlash spec drives price

Used gear hobber · ₹3–10 lakh · Auctions, machinery dealers

New CNC gear hobber · ₹25–80 lakh · + cutter inventory

Cluster geography: Rajkot (Gujarat) is the power-transmission hub — gears, gearboxes, tractor and machinery components at aggressive prices. Coimbatore pairs gearboxes with its motor industry; Pune and Chennai serve automotive-precision demand; Faridabad and the NCR belt cover general engineering. For a FabFlow-style sourcing flow: specify module, teeth, face width, material grade, heat treatment, and quality class (ISO 7 hobbing is the default ask) — a drawing that says only "40-tooth gear" will get you a 40-tooth gear and a headache.

Lead-time reality: prototype hobbed or wire-EDM gears ship in 3–7 days; carburize-and-grind chains add 2–3 weeks; PM and molded tooling add 4–8 weeks before the first production part.

9. Design Checklist & Failure-Mode Matrix

Before releasing a gear design:

  1. 20° pressure angle, standard module from ISO 54 preferred series.
  2. Pinion \geq 17 teeth; profile-shift if below (x_{min} = (17-z)/17).
  3. Contact ratio \varepsilon \geq 1.2 (spur); hunt for coprime tooth counts.
  4. Face width 8m–16m; wider buys nothing but edge loading.
  5. Maximum root fillet radius; shot peen if K_A > 1.5 or life > 10^8 cycles.
  6. Rating: Lewis for the quick pass, ISO 6336 with real K factors for the release check, \sigma_{H\,lim} against case depth \geq max-shear depth.
  7. Backlash budget including thermal mismatch (\Delta j = 2\Delta a \tan\phi).
  8. Lubrication mode matches pitch-line velocity; EP oil for hypoids and high sliding.
  9. Housing bores pinned and toleranced — alignment is a K_{F\beta} multiplier on every stress.
  10. Prototype first: blue-marking contact pattern should cover 60–80% of the flank, centered; a dial-gauge backlash check verifies the budget before load.

Failure-mode matrix:

Symptom · Root cause · Fix

Tooth snapped at root, beach marks · Bending fatigue · Larger module, profile shift, shot peening, bigger fillet

Flake-shaped craters on flanks · Pitting (contact fatigue) · Case hardening, better alignment, smoother flanks

Grey matte staining, micro-craters · Micropitting (\lambda < 1) · Higher viscosity, superfinish, EP additives

Torn, smeared flanks, blued tips · Scuffing (asperity welding) · EP oil, lower temperature, better cooling

Knife-edge thin teeth · Abrasive wear · Sealing, filtration, hardened flanks

Flank plastic flow, rolled edges · Overload / soft material · Harder grade, bigger module

Clicking at reversal, position error · Excessive backlash · Tighter class, anti-backlash design, thermal-matched housing

Noise at one-per-rev frequency · Eccentricity / periodic error · Grind, hunting tooth counts, better mounting

Hot box, low output torque (worm) · High sliding loss · Fewer ratio per stage, forced cooling, better lubricant

10. Conclusion

Gears reward the engineer who respects the three numbers that govern them: the module that scales the tooth, the pressure angle that shapes the flank, and the contact stress that sets the lifetime. The worked examples here — a 637 N pinion rating out at 53 MPa root / 475 MPa flank, a 40:1 worm at 61% efficiency, a 4:1 planetary from one Willis equation, and a 10 N·m reducer whose teeth are the least stressed parts in the box — are the everyday arithmetic of machine building, and none of it requires software fancier than a spreadsheet.

The sourcing math is equally plain: India's gear clusters turn a napkin sketch into a hobbed, hardened gear in under a week for under ₹2,000, and into a molded or sintered production part for under ₹80 at volume. Whether you're prototyping a robot joint with SLS nylon, spec'ing a case-carburized pinion for a production reducer, or designing a 10:1 box for a rotary axis, the constraint is never the gear math — it's the alignment, the bearings, and the thermal budget around it.

When you need gears, gearboxes, or the machines to make them, FabFlow connects you with verified Indian manufacturers — specify your module, tooth count, material, and quality class, and get quotes from shops that cut, sinter, mold, and print teeth every day.

More FabFlow blog posts