Almost every product with a moving part contains a transmission, even if nobody on the team calls it that. A motor spins fast and weak; the job usually needs something slow and strong, or rotary motion converted into linear travel. The gearbox is what bridges the two. Get it wrong and you get a product that is too loud for a bedroom, too slow for the marketing claim, or one that strips a gear the third time a user forces it. Get it right and it disappears — which is the whole point.
Start from the load, not the motor
The most common sequence error is picking a motor first and then hunting for a gearbox to rescue it. Work the other direction. Define the output the mechanism has to deliver: peak torque, continuous torque, output speed, duty cycle, and how many cycles over the product's life. From there the reduction ratio and the motor almost choose themselves.
Peak and continuous are different numbers and both matter. A powered window shade might need 4 N·m to break a stuck roller free but only 0.8 N·m to keep it moving. Sizing everything to the peak makes the product heavy and expensive; sizing to the continuous number makes it fail on the first sticky unit. Running the arithmetic properly — inertia, friction, gravity, acceleration — is covered in our guide to calculating the motor torque your product needs, and it is worth doing on paper before anything gets modeled.
Ratio follows directly. If the motor's efficient operating band is 3,000 rpm and the output should turn at 60 rpm, you need roughly 50:1, plus margin for efficiency losses. Whether that reduction is best delivered by gears, a belt, or a lead screw depends on packaging and noise as much as on math, and the tradeoffs between the first two are laid out in our comparison of belt drives and gear drives.
The gear types and what each one buys you
| Type | Typical ratio per stage | Efficiency | Notes |
|---|---|---|---|
| Spur | 1:1 to 6:1 | 96–99% | Cheapest, easiest to mold, noisiest at speed |
| Helical | 1:1 to 6:1 | 94–98% | Quieter than spur, generates axial thrust the bearings must take |
| Planetary | 3:1 to 10:1 | 90–97% | High torque density, coaxial input and output, more parts |
| Worm | 5:1 to 100:1 | 40–85% | Right-angle, often self-locking, runs warm |
| Bevel | 1:1 to 5:1 | 93–97% | Turns a corner; alignment sensitive |
| Harmonic / cycloidal | 30:1 to 160:1 | 70–90% | Near-zero backlash, expensive, used in robotics joints |
Two properties decide more designs than efficiency does. The first is self-locking: a worm drive with a low lead angle will not back-drive, so a lifted load stays lifted with the motor off. That is free safety in a lift or an actuator, and a trap in anything a user must be able to move by hand in a power failure. The second is backlash — the free play when you reverse direction. Positioning mechanisms, camera gimbals, and anything with a closed control loop care intensely; a fan or a pump does not.
Buy the gearbox or design one
For most products the honest answer is buy. A catalog gearmotor from an established supplier arrives with published torque ratings, life data, and a price that a custom design will not beat below several thousand units per year. Design custom when the packaging genuinely will not accept a cylinder, when the ratio is unusual, when unit volume is high enough that tooled gears pay back, or when the motion path is not a simple reduction. The general framework for that call is in our piece on off-the-shelf components versus custom design.
A middle path works surprisingly often: buy a standard gearmotor and design only the last stage — the output gear, the crank, the linkage — as a custom molded or machined part. You keep the vendor's reliability data on the hard part and control the interface that matters.
Material, lubrication, and noise
Metal gears carry more load and tolerate heat; molded plastic gears in acetal, nylon, or PEEK are quieter, self-lubricating to a degree, and dramatically cheaper at volume. The limits on the plastic option are real but often misunderstood — see when plastic gears can handle the load before ruling them out of a consumer product. A common hybrid pairs a steel pinion with a plastic wheel, which cuts noise and lets the softer part act as the sacrificial element.
Lubrication is a specification, not an afterthought. Pick the grease grade, state where it goes and how much, and confirm it is compatible with the plastics it touches — some greases attack polycarbonate and ABS. Then design the housing to keep it in and contamination out.
Noise is where consumer products win or lose. Gear mesh frequency equals tooth count times shaft speed, and it will excite whatever panel resonates near it. Helical teeth, higher tooth counts with a finer module, tighter center-distance control, and isolating the gearbox from the enclosure all help. Test it early: a mechanism that measures fine and sounds like a coffee grinder will be sent back by the industrial designer.
Bearings, shafts, and the parts that actually fail
Gears rarely fail alone. They fail because a shaft deflected, a bearing wore, or a housing flexed and threw the mesh out of alignment. Support every gear on two bearings where you can, size them against the radial and axial loads the gear type generates, and treat bearing selection as part of the transmission design rather than a detail. Shaft deflection under peak torque should be checked, not assumed.
Then check life. A mechanism that cycles 20 times a day for seven years sees roughly 51,000 cycles; a printer feed mechanism may see millions. Tooth root stress that is fine for one is a fatigue failure for the other, which is exactly the pattern described in why parts break after a year in service. Apply a service factor for shock loading and jams, and be explicit about the factor of safety you are designing to rather than leaving it implied.
Projects House designs mechanisms, gearboxes, and drive trains for products that have to survive years of real use — from concept kinematics through tolerance analysis, prototyping, and production drawings. If you have a mechanism that needs to be quieter, stronger, or cheaper, describe it through our contact form.