How center distance and gear quality affect current draw
Why does one gearmotor from a lot draw noticeably more no-load current than the next, when both pass every dimensional check on the drawing?
No-load current is the most useful single measurement on a gearmotor production line, because it is a direct reading of internal drag. The motor converts current to torque through its torque constant. With nothing on the output shaft, every millinewton meter the motor produces is being spent inside the assembly.
ΔI = ΔTdrag / kT
- ΔI
- change in no-load current, A
- ΔTdrag
- change in internal drag torque referred to the motor shaft, N·m
- kT
- torque constant, N·m/A
That last point decides where to look. A first-stage pinion running tight is far more visible in no-load current than a final-stage problem, because first-stage drag is not divided down. It is also the stage that turns fastest, so it dominates the churning and windage terms as well.
Center distance sets backlash, by geometry
For a standard involute mesh, a change in center distance changes backlash by a fixed factor. Nothing about the gears has to change for the assembly to lose or gain clearance. The relationship is one of the few exact statements in the whole subject.
Δj = 2 · Δa · tan α
- Δj
- change in backlash, measured normal to the flank
- Δa
- change in center distance
- α
- pressure angle, typically 20 degrees
The important thing about that plot is how short the horizontal axis is. Tens of micrometers of bore-to-bore variation are ordinary in a small molded or die-cast housing, and tens of micrometers of center distance is the entire backlash budget. This is why housing bore position, not gear tooth thickness, is usually the variable that separates a smooth unit from a tight one.
Temperature moves the same variable
Center distance is not fixed once the parts are assembled. Gears and housings expand at different rates, and in a compact gearmotor the parts are frequently made from materials whose expansion coefficients differ by a factor of four or more.
Take an illustrative case: a polymer gear with a 10 mm pitch diameter and a linear expansion coefficient near 90 µm/m/K, running in an aluminum housing with a 20 mm center distance and a coefficient near 23 µm/m/K. Warming the assembly by 40 K grows each gear's pitch radius by about 18 µm, while the housing's center distance grows by about 18 µm as well, but that housing growth has to cover the growth of both gear radii. The mesh tightens by roughly 18 µm of effective center distance, which by the relationship above removes about 13 µm of backlash. An assembly built at 20 µm of cold backlash is close to zero when hot.
Hygroscopic polymers add a second, slower shift in the same direction. Several common polyamides absorb moisture from the air and grow as they do, over days and weeks rather than minutes. A gear train that measures correctly at final assembly can be tighter after storage in a humid warehouse, and the change is not reversible on the same time scale as a temperature change.
What else raises drag
| Source | Where it acts | Visibility in no-load current | How to confirm it |
|---|---|---|---|
| First-stage center distance | Motor shaft, undivided | High. A tight first mesh is the classic outlier | Measure bore-to-bore on the housing, and backlash at the first stage |
| Grease quantity and base viscosity | Every stage, strongly at low temperature | High when cold, moderate when warm | Compare current at cold start and after a warm-up run |
| Bearing preload or misalignment | Motor shaft and output shaft | High at the motor, low at the output | Spin the bare motor before assembly and record its own no-load current |
| Tooth surface finish and profile deviation | Every mesh, load dependent | Moderate, and it falls as the unit runs in | Repeat the reading after a run-in period and compare the change |
| Seal or shaft wiper | Output shaft | Low. Output drag is divided by the ratio | Measure output breakaway torque separately |
| Debris or a burr | Wherever it landed | Erratic. The reading is not repeatable over one revolution | Log current against shaft angle rather than as a single value |
The last row is worth doing routinely. A single averaged current number hides the difference between a uniformly tight assembly and one that is fine for most of a revolution and binds once. Those two conditions have completely different causes and completely different consequences in the field.
The tradeoff, stated plainly
Backlash is not a defect. It is the clearance that lets a mesh run without binding when the parts are at their tolerance limits and the assembly is at temperature. Removing it costs current, heat and wear. Adding it costs positional accuracy and, in a reversing duty cycle, audible rattle.
- Positioning mechanisms that always approach from one direction can tolerate more backlash than the specification suggests, because the error is repeatable and can be handled in the motion profile.
- Reversing mechanisms pay for backlash twice: as lost motion, and as an impact each time the drive crosses the gap.
- A design that specifies near-zero backlash has to state the temperature range it is measured over, or it is specifying an interference at one end of the range.
The practical resolution is to hold center distance rather than to chase tooth thickness. Center distance is set by two bores in one part, and it is measurable, controllable and correctable. Tooth thickness is distributed across a supply chain.
Reading a no-load current distribution
Because no-load current sums every internal drag term, it is the natural end-of-line screen. It is also easy to use badly. A single upper limit taken from a catalog value passes a population that has drifted and rejects a population that was always fine, because the limit was never derived from this process.
- Set the limit from the measured distribution of a stable process, not from a published figure. The published figure describes a design; the distribution describes what this line makes.
- Watch the shape, not only the mean. A bimodal distribution says two things are happening, usually one fixture or one operator or one cavity behaving differently from the rest.
- Record the reading against shaft angle where the equipment allows it. A unit that binds once per revolution has a specific, findable cause; a uniformly high unit has a different one.
- Take the reading after a defined run-in, at a defined temperature, at a defined voltage. Drag falls during the first minutes of running and it falls further as the unit warms, so an undefined procedure produces an undefined limit.
- Keep the data. A slow upward drift in mean no-load current across weeks is one of the earliest visible signs of tool wear or a material change, and it is only visible if the numbers were stored.
Used this way the measurement is diagnostic rather than a pass or fail gate. It tells you which of the drag sources in the table above is moving, and it tells you before the change is large enough to be a complaint.