Spiral, Zerol and straight bevel gears cut, hardened and ground in-house. We hold ±0.005 mm and check tooth contact before anything ships. One prototype or a 10,000-piece run.
Bevel sets fail for the same handful of reasons. Each one shows up on the bench, not on the drawing.
The set was cut without a contact check. Load sits on one end of the flank, the pattern walks under torque, and the pair starts whining within a few hundred hours. No amount of lapping saves a pattern that was set wrong at the start.
Distortion moves the teeth a few hundredths. A set that measured fine green comes back tight and seizes at operating temperature. This is a process planning problem, not a grinding problem.
A 0.02 mm runout at the bore throws the whole mesh off. The gearbox runs hot on one side, lubrication breaks down, and you replace a bearing that was never the cause.
Hobbing is only half the route. Without a grinder and a lapping setup, the flank finish stays at Ra 3.2 μm and the set never gets quiet at speed.
Turning, gear cutting, heat treatment, grinding and inspection under one roof in Dongguan.
Spiral bevel sets carry high loads quietly because the teeth engage gradually instead of all at once. The spiral angle and the hand of the spiral have to match the mating pinion, so we cut pinion and gear as a pair, never as two separate jobs.
Zerol bevel gears give you the same smooth engagement with a zero spiral angle. That matters when the gear has to run in either direction, or when you want spiral-bevel behavior on a shaft that does not tolerate axial thrust. We cut both on 16 simultaneous 5-axis machining centers, with a Ø400 mm rotary table for the larger ring gears.
Case hardening is where bevel gears usually lose their geometry. We plan the allowance for it and grind after, so the tooth thickness you specified is the tooth thickness that ships. Ground flanks reach Ra 0.8–1.6 μm as a normal production finish, and Ra 0.2–0.8 μm when the application calls for it.
Lapping comes last. We run the pair together with marking compound and read the contact pattern across the flank, then adjust until it sits in the middle of the tooth. If the pattern cannot be brought in, the part does not leave the building.
Pick by load, speed and whether the shaft can take axial thrust.
| Type | Best for | Watch out for |
|---|---|---|
| Straight bevel | Low-speed drives, hand feed, differentials | Noisy above moderate rpm; load hits the tooth all at once |
| Spiral bevel | High load, quiet running, vehicle axles | Produces axial and radial thrust; needs matched pinion |
| Zerol bevel | Reversing drives, spiral behavior without thrust | Slightly lower load rating than a true spiral set |
| Hypoid | Offset axes, packaging tight around the axle | Requires special lubricant; sliding action at the flank |
| Ground spiral | Servo and spindle drives needing low backlash | Grinding adds cost; justified only on tight backlash specs |
A bevel gear is rarely the only part on the drawing.
Simultaneous 5-axis cutting for gear blanks, housings and the mating shafts in the same tolerance band.
Shafts, hubs and ring gear blanks turned and milled to the bore and face the gear will be cut against.
Case hardening and through hardening planned into the routing so grinding allowance is correct.
Black oxide, electroless nickel, zinc and hardcoat anodizing for corrosion and wear.
Fit-check prototypes before you commit to a hardened gear set and its tooling.
Runout, backlash, tooth thickness and contact pattern measured on the finished pair.
Numbers only from our own machine list.
| Item | Capability | Notes |
|---|---|---|
| Achievable tolerance | ±0.005 mm | ±0.0002 in on request |
| Fine finish | Ra 0.2–0.8 μm | Ground flanks |
| Standard ground finish | Ra 0.8–1.6 μm | Normal production finish |
| As-machined finish | Ra 1.6–3.2 μm | Pre-grind condition |
| Max processing size | 4,000 mm | Across the shop |
| Rotary table | Ø400 mm | For larger ring gears |
| Order quantity | 1 to 10,000+ parts | No minimum order quantity |
| First article | 3–5 days | From released drawing |
Data we can point to on the shop floor, not adjectives.
Fifteen years of gear and machined part production across three wholly-owned plants in Dongguan and Singapore.
16 simultaneous 5-axis centers, 16 mill-turn centers, 12 four-axis mills and 27 three-axis machines.
Tolerance on checked features after hardening, not on the green blank.
150 technicians in total, with process planning done by the people running the machines.
ISO 9001:2015, IATF 16949:2016, ISO 13485:2016 and ISO 27001:2022.
Drawing review with manufacturability notes returned within 12 hours.
Differential and reduction gears where backlash control decides cabin noise.
Right-angle drives and index tables that run continuously for years.
Servo axes that need low backlash and repeatable positioning.
Low-volume, fully documented gear sets with material traceability.
Bevel gears suit right-angle drives where efficiency matters. A worm set gives you high reduction in one stage and self-locking, but it loses more energy to sliding friction.
If your input speed is above a few hundred rpm and efficiency is on the spec sheet, use a bevel pair. If you need 50:1 in one stage and the drive can run hot, a worm is the cheaper answer.
Under roughly 300 rpm with moderate load, straight bevel teeth are cheaper to cut and easier to inspect. They get noisy as speed rises because the whole tooth engages at once.
Spiral bevel sets cost more but run quieter and carry more load at the same size. If the gearbox will be heard by an operator, spiral is usually worth it.
Backlash is set from the drawing and verified on the assembled pair. It depends on module, center distance and how the housing controls the shaft positions.
Send the housing bore positions along with the gear drawing. Backlash that looks achievable on the gear alone can disappear if the bearing seats are not held tight.
Yes. Cutting the pair from the same setup is the only way to guarantee the contact pattern lands correctly under load.
We record the pattern on marking compound and keep the pair together through heat treatment, grinding and lapping so the two parts never get mixed with another set.
Case-hardening alloy steels such as 4140 and 4340 are the usual choice for loaded drives. We also machine 1045, 1018 and 4130 where hardening is not required.
Stainless grades 303, 304, 316, 17-4PH and 440C cover food, medical and corrosive environments. Bronze and brass work for low-load, low-speed sets.
We leave grinding allowance based on the section thickness and the hardening cycle, then grind after treatment to bring tooth thickness and runout back to spec.
For tight sets we plan a second stress relief before final grinding. That adds a step but it is cheaper than scrapping a hardened batch.
Yes, if we can recover the module, pressure angle, spiral angle and hand from the sample. Heavily worn flanks are harder to measure, so a drawing or a mating part helps.
We will tell you up front what can and cannot be determined from the sample before you commit to a production run.
A drawing with module, number of teeth, pressure angle, spiral angle and hand, plus material, heat treatment and backlash. A 3D model helps but does not replace the tooth data.
Send files through the quote page and you get a price plus DFM notes within 12 hours. NDA is available before you send anything.
Upload the tooth data and we will come back with a price and manufacturability notes within 12 hours.
12-hour quote100% inspectionNo minimum order quantityNDA on request
Trusted by engineers and manufacturers worldwide
Upload your 3D model or 2D drawing and get a quotation with a free DFM analysis. Maximum processing size 4,000 mm.
CNC Metals 13 grades
CNC Plastics 10 grades
Machines & processes 12 options
Surface & post-processing 10 options
Material: not selected Machine: not selected Post-process: not selected
Fill in this short form and we'll open WhatsApp with your message ready to send. We only use these details to reply to you.
Nothing is sent until you press send inside WhatsApp.
