A planetary gearbox rated for 400 Nm at 20:1 ratio was installed on an indexing turret. The turret stalled under a 250 Nm load. The customer sized the gearbox from the catalog — 400 Nm rated, 250 Nm applied, 60% margin. The gearbox stalled anyway. The customer called the gearbox manufacturer. The manufacturer asked about the shock load, the duty cycle, and the output bearing. The customer hadn’t considered any of them.
The catalog rating assumes even load sharing
A planetary gearbox shares torque across three planet gears. The catalog rating assumes the three planets share the load equally — one-third each. In reality, the planets share load evenly only when the carrier, ring, and sun gears are perfectly manufactured. Manufacturing tolerances (gear tooth spacing, carrier pin position, planet bore position) mean the planets don’t share equally. One planet carries 40% of the load, the others 30% each. The gearbox fails at 60% of its catalog rating.
The turret’s gearbox was a standard commercial unit. The load-sharing tolerance was typical — about ±15%. The effective rating was 340 Nm, not 400 Nm. The applied load was 250 Nm steady, but the indexing motion added a shock factor of 1.5 (the turret accelerated and decelerated 10 times a minute). The effective load was 375 Nm. That’s over the 340 Nm effective rating. The gearbox stalled and the teeth on the sun gear started to fret.
The calculation that should have been done
The sizing calculation for a planetary gearbox on an indexing table:
Torque = 250 Nm static.
Service factor for indexing (10 starts/min) = 1.5.
Effective torque = 375 Nm.
Required rating = 375 / 0.85 (load-sharing allowance) = 440 Nm.
The correct catalog selection was 500 Nm rated, not 400 Nm. The bigger gearbox cost 20% more and fixed the stall. The customer’s “60% margin” was actually a 15% deficit once the real service conditions were included.
The output bearing is a separate rating
The second issue: the turret’s overhung load. The turret was 300 mm in diameter, mounted directly on the gearbox output shaft. The 250 Nm torque plus the 40 kg turret weight created an overhung load on the output bearing. The catalog’s output bearing rating assumed a pure torque load with a small axial component. The actual overhung moment was 3x the catalog’s limit. The output bearing was the weak point — it deflected, the sun gear lost mesh, and the gearbox stalled.
The fix: a support bearing on the turret, mounted outside the gearbox. The turret was supported by a separate bearing at the base. The gearbox output shaft now carries torque only, no overhung moment. The gearbox’s output bearing life went from months to years.
The planetary gearbox selection checklist
| Item | Rule |
|---|---|
| Nominal torque | Catalog rating at the actual speed, not the peak |
| Service factor | 1.25 steady, 1.5 indexing, 2.0 shock |
| Load sharing | Derate commercial gearboxes by 15% |
| Overhung load | Check the output bearing rating against the actual moment |
| Backlash class | Indexing needs backlash < 0.1° (use preloaded planetary) |
| Thermal | Check the oil temperature at the actual duty cycle |
Why it stalled instead of grinding
The stall happened because the gearbox was at the edge of its rating. The motor (a servo with torque limit) hit its limit and stopped. The teeth were fretting — micro-wear on the tooth flanks from the load sharing imbalance. If the motor had no torque limit, the gearbox would have sheared teeth. The servo’s torque limit saved the gearbox and made the stall look like a control problem. It wasn’t. It was a sizing problem all along.
Planetary gearboxes share load unevenly. Derate commercial units by 15%, apply the service factor, and check the output bearing. The 250 Nm load on a 400 Nm gearbox stalled because the real demand was 440 Nm and the bearing couldn’t carry the overhung moment. Size the gearbox for the shock, the sharing, and the moment — the catalog number is the start, not the answer.
