The Gearbox That Got Too Hot After an Hour
We mounted a planetary gearbox on a servo axis. The motor was sized for the load torque. After an hour of continuous running, the gearbox housing hit 70°C. Too hot to touch. The bearing grease was breaking down. The problem: we calculated the gear ratio from the load torque, but forgot the gearbox efficiency. The gearbox absorbs about 3–5% of the power as heat. At continuous high torque, that heat builds up. We added a thermal pad (mounted the gearbox to the aluminum extrusion frame, which acts as a heat sink) and slowed the duty cycle slightly. The temperature dropped to 50°C. The mistake was treating the gearbox as a ratio-only device, ignoring its efficiency and heat.
Gearbox efficiency and heat matter for continuous-duty axes. The gearbox isn’t free — it loses power as heat. This article covers the numbers.
Gearbox Efficiency
The gearbox’s efficiency (η) is the fraction of input power that reaches the output. The rest becomes heat.
- Planetary gearbox: η = 95–97% per stage. A 2-stage gearbox loses about 5–8%.
- Worm gearbox: η = 40–70% (much lower). Lots of heat. Not for continuous servo duty.
- Helical gearbox: η = 97–98%.
The lost power becomes heat:
P_heat = P_input × (1 − η)
For a 400 W servo motor (0.4 kW) through a 2-stage planetary (η = 0.93): P_heat = 0.4 × (1 − 0.93) = 0.028 kW = 28 W. That heat must go somewhere.
Step 1: Continuous vs Intermittent Duty
A servo axis that moves occasionally (pick-and-place, 10 seconds on, 50 seconds off) has low heat. The gearbox cools between cycles.
A continuous-duty axis (conveyor, rotary indexer running 24/7) builds heat. The gearbox housing temperature rises until it reaches equilibrium (heat generated = heat dissipated to the air).
For continuous duty, check the gearbox’s rated continuous torque (not just the peak). The gearbox datasheet lists a “permissible torque” for continuous operation. If you exceed it, the gearbox overheats.
Step 2: Heat Dissipation
The gearbox dissipates heat through its housing (natural convection). A bare gearbox dissipates maybe 10–30 W. If it generates more, it gets hot.
Ways to improve cooling:
- Mount to a metal frame: The aluminum extrusion or steel plate acts as a heat sink. Bolt the gearbox housing to the frame (use thermal paste between them).
- Add a cooling fan: For high-power gearboxes, a fan blows air over the housing.
- Derate: Reduce the duty cycle (let it cool between bursts).
The gearbox heat rule: Calculate P_heat = P × (1−η). For continuous duty, mount the gearbox to a metal frame (heat sink). The gearbox that hit 70°C had no heat sinking — it was mounted on a thin bracket. Bolt it to the frame, and the temperature drops.
Step 3: Backlash and Accuracy
Besides heat, the gearbox’s backlash affects the axis accuracy (article 68). For a positioning axis, pick a low-backlash gearbox (<1 arc-minute for planetary, <5 arc-min for standard). For a conveyor (no precision), standard backlash is fine.
Low-backlash gearboxes cost more and have more friction (slightly lower efficiency). Match the backlash to the application.
Step 4: Gear Ratio Selection
The gear ratio (article 55) is chosen to match the motor speed to the load speed. But it also affects the reflected inertia and torque:
- Reflected inertia: The load inertia divided by the square of the ratio. A higher ratio reduces the reflected inertia (good for motor sizing).
- Torque multiplication: The output torque = motor torque × ratio × η.
Don’t over-ratio. A 10:1 gearbox on a light load wastes torque and adds heat. Pick the ratio that matches the motor’s speed to the load.
| Gearbox Type | Efficiency | Backlash | Best For |
|---|---|---|---|
| Planetary (standard) | 95–97% | 3–8 arc-min | General servo, conveyors |
| Planetary (low-backlash) | 93–96% | <1 arc-min | Precision positioning |
| Worm gear | 40–70% | High | Low speed, self-locking (not continuous servo) |
| Helical bevel | 97–98% | 2–5 arc-min | High-power, right-angle |
A Gearbox Selection Checklist
- What is the load torque? (Continuous and peak.)
- What ratio? (Motor speed to load speed.)
- Efficiency at that ratio? (Planetary ~95%.)
- P_heat = P × (1−η)?
- Is the duty continuous or intermittent?
- Is the gearbox mounted to a heat-sinking frame?
- Backlash needed? (Precision or conveyor?)
- Is the reflected inertia within 5× motor inertia?
- Is the output torque within the gearbox’s continuous rating?
- Will the temperature be measured at FAT? (Baseline?)
The Bottom Line
Gearbox efficiency and heat are not free. The gearbox that hit 70°C had no heat sinking and ignored its efficiency loss. Calculate P_heat = P × (1−η), mount the gearbox to a metal frame for cooling, and check the continuous torque rating. Planetary gearboxes (95% efficient) are standard for servo; worm gearboxes (50% efficient) are not for continuous duty. The gearbox that runs cool after 8 hours wasn’t the biggest one — it had proper heat sinking.