Selecting a gearmotor for optical positioning

On an optical stage the torque is trivial and the error budget is everything. Lost motion and the motor's own dissipation are usually the two largest terms in a micron-level budget.

Wedge
Industrial instrumentation and automation
Catalog reviewed
2026-08-18
Configurations screened
32 published

A filter slide, a variable aperture, a mirror tip, a spectrometer grating drive, a focus stage. A gearmotor turns a fine-lead screw or a lever, and an optic moves a few millimetres or a few degrees. The forces are small: a light carriage on preloaded guides, sometimes a return spring. The mechanism exists to place an optical surface, and it is judged by how well it comes back to the same place.

The torque model is the ordinary power-screw one, T = F·L / (2π·η), with F being guide friction plus any spring preload. It is included here because it has to be checked, not because it will be close. The model that decides the design is the error budget, and it has four terms that combine as a root sum of squares: lost motion in the geartrain referred through the screw lead, axial play at the screw thrust bearing, encoder and control resolution, and thermal expansion of the structure between the datum and the optic. On a micron-level stage the first and the last routinely dominate, and the last one includes the heat the motor itself puts into the bracket.

The error budget, and specifically whether the axis can approach unidirectionally.

Approaching from one direction removes the geartrain lost motion from the budget completely and usually turns an impossible specification into a straightforward one. If bidirectional positioning is required, lost motion referred through a fine screw lead is typically several microns and there is no motor selection that fixes it. The second governing factor, once the mechanical terms are handled, is dissipation: a motor left energized to hold position heats the bracket it is bolted to, and on an aluminium structure a couple of degrees over 60 mm is comparable to the whole positioning tolerance.

Continuous, peak, starting and abnormal

Sizing on one of them is how a mechanism gets a part that works on the bench and fails in the field.

Row Continuous Sizes the thermal case Peak Sizes the torque case Starting Sizes the transient Abnormal Sizes the protection
What the mechanism is doing The move itself, at guide friction plus spring preload. On a light carriage this is under a millinewton-metre and is not a selection input.Breakaway from rest with a preloaded guide, and the end of a move against a spring at full extension. Both small.The reversal. On a bidirectional move the drive traverses backlash before the optic moves, and any control loop closing on a motor-shaft encoder reports motion the optic is not making.The axis driven into a mechanical end stop, or a jammed slide. Small optical stages have very little mechanical strength between the screw and the optic mount, and the stall torque available through a 31:1 reduction is far more than a fine screw and a light carriage are designed to carry.

Failure modes and misleading specifications

What follows is what actually fails, and which published number sends people the wrong way.

Common failure modes
Failure mode Physical cause What it looks like in the field
Repeatability passes one way and fails the otherGeartrain lost motion referred through the screw lead. At 1.5 degrees of output backlash and a 1 mm lead this is 4.2 microns of dead travel on reversal.A fixed offset between approach directions, constant per unit, unaffected by speed, that disappears entirely if the test only ever approaches from one side.
Position drifts over the first tens of minutesThermal expansion. The motor is dissipating into the bracket, or the instrument is warming from cold.Monotonic drift that correlates with elapsed time from power-on and stabilizes, then repeats after every cold start.
The stage will not hold position with power removedThe screw lead angle exceeds the friction angle, so the mechanism back-drives, and the geartrain back-drives with it at roughly 2 − 1/η efficiency.Position lost on every power cycle, worse on a vertical axis, worse when the mechanism is well lubricated.
Bent screw or damaged mount after a homing faultThe axis driven into an end stop at full command. The reduction multiplies motor stall torque into far more axial force than the mechanism was designed for.A single event, usually traceable to a homing sequence or a lost limit signal, with damage at the weakest element rather than at the motor.
Specifications that mislead here
Specification Why it misleads in this application
Resolution quoted in microns per countIt is arithmetic, not performance. A 48 count per revolution encoder behind 31:1 and a 1 mm lead gives 0.67 microns per count while the same train has 4.2 microns of lost motion.
Continuous output torqueThe mechanism uses under a fifth of a millinewton-metre. Every candidate has one to two orders of magnitude of margin and the number does not distinguish between them.
Rated speedOnly relevant as a gate. Optical stages move slowly, but a trapezoidal profile peaks at 1.5 times its mean speed and that peak has to fit inside the rated output speed.
Backlash quoted as typicalOn an error budget you need a bound, not a mean. The typical figure will be exceeded by half the population.

Three decisions carry most of the risk, and each one costs something real.

Spur or planetary reduction

Row Spur Planetary
When it is right Where the shortest possible train is wanted, because fewer stages means less accumulated lost motion, and torque is not a constraint.Higher ratios in a 10 to 13 mm envelope, and where the end-stop case has to be carried by the gearhead.
What it costs Lower torque ratings, which matter only for the end-stop case rather than for the move.Four stages of lost motion in series, which is the term the whole error budget turns on.

Brushed coreless or brushless

Row Coreless brushed Brushless
When it is right Low duty, and where the very low dissipation of a lightly loaded coreless motor keeps the thermal term small. At the operating point below the motor dissipates about 0.1 W.Continuous scanning duty, or where no wear debris is acceptable near the optics.
What it costs Commutation ripple appears as velocity ripple during a scan, and brush debris is a source close to an optical surface.Driver switching next to a sensitive detector, and a 24.2 mm minimum body diameter in the published range, which most optical stages cannot accept.

How position is held

Row Self-locking screw, motor de-energized Motor energized at holding current Detent or kinematic stop at each position
When it is right The lead angle is below the friction angle. At a 1 mm lead on a 4 mm screw the lead angle is 4.55° against a friction angle of 8.53°, so it locks.The mechanism cannot self-lock, or the axis has to be repositioned continuously.The mechanism has a small number of discrete positions, such as a filter slide.
What it costs Release torque is higher than drive torque, and the locking margin falls as the thread is lubricated or wears.Continuous dissipation into the structure, which is the term most likely to blow a micron-level budget.The stop sets the repeatability, so the drive only has to reach it. Continuous positioning is given up entirely.

Worked example

Build the error budget for a filter slide specified at ±5 microns, then check the drive against it.

Positioning tolerance
±5 µm
Travel
12 mm in 4 s
Screw lead
1 mm, on a 4 mm nominal diameter
Carriage mass
0.05 kg, horizontal
Return spring
1.0 N
Calculation
Step Expression Result
Axial force0.1 × 0.05 kg × 9.81 m/s² + 1.0 N1.05 N
Screw torqueT = 1.05 N × 0.001 m ÷ (2π × 0.25)0.668 mN·m
Mean and peak output speed12 mm ÷ 1 mm per rev ÷ 4 s, then × 1.5180 rpm mean, 270 rpm peak
Motor torque at the operating point0.668 mN·m ÷ (31.12 × 0.66)0.0325 mN·m
Current and dissipation, 3 V winding0.0325 ÷ 2.343 mN·m/A + 19.2 mA, then × 3 V33.1 mA, about 0.099 W
Resolution with the published encoder1 000 µm ÷ (48 counts × 31.12)0.669 µm per count
Lost motion referred to the slide1.5° ÷ 360° × 1 000 µm4.17 µm
Thermal term for a 2 K rise60 mm × 23.1 × 10⁻⁶ /K × 2 K2.77 µm
Screw lead angle and friction anglearctan(1 mm ÷ (π × 4 mm)) against arctan(0.15)4.55° against 8.53°, so the screw self-locks
Budget, bidirectional, root sum of squares√(4.17² + 2.77² + 0.67²)5.05 µm, over the ±5 µm tolerance
Budget, unidirectional approach√(2.77² + 0.67²)2.85 µm, inside the tolerance with room for the terms not yet counted

Result

MM-C1025-P00311-030B satisfies the mechanical requirement easily: 10 mm coreless brushed, 31.12:1 planetary, 3 V, rated 334.2 rpm at the output against the 270 rpm peak demand, and 10.1 mN·m continuous against a 0.668 mN·m load, which is a factor of 15. It is not chosen for torque and no candidate would be. The budget is what decides the design. Approaching bidirectionally, the assumed lost motion alone is 4.17 microns and the total reaches 5.05 microns, which fails a ±5 micron tolerance before screw axial play, mounting compliance and the optic's own mount are counted. Approaching unidirectionally the budget falls to 2.85 microns and the dominant term becomes thermal, at 2.77 microns for a 2 kelvin rise across 60 mm of aluminium. The screw self-locks at a 4.55 degree lead angle, so the motor can be de-energized between moves and its 0.099 W stops heating the bracket. Unidirectional approach plus a de-energized hold is the design, and neither of those is a motor selection.

What would change it

A finer screw lead scales lost motion and resolution down together: at a 0.5 mm lead the lost-motion term halves to 2.08 microns and the bidirectional budget closes, at the cost of doubling the required output speed to 540 rpm peak, which exceeds every published rated output speed. That trade is the whole design. A measured lost-motion figure replaces the largest assumed number in the budget and should be obtained first. Where the lowest possible dissipation matters, MM-C1017-P00311-060A in the same 10 mm envelope runs at about 0.11 W and has a stall torque essentially equal to its gearhead rating, which also protects the screw in a homing fault.

A short qualification plan

Each step names the measurement, not the intention.

  1. Measure lost motion, do not assume it

    Bidirectional positioning error at the optic with an external displacement reference, on at least ten assemblies, reported as a distribution.

  2. Separate thermal drift from mechanical error

    Position held constant while the mechanism warms from cold, with a thermocouple on the bracket. Drift against temperature gives the coefficient directly.

  3. Verify self-locking as built

    Applied axial force at the carriage with the motor de-energized and leads open, after the lubricant the assembly will ship with, at the temperature extremes.

  4. Prove the end-stop case

    Deliberate drive into the mechanical stop at full command, then inspection of the screw, nut and mount. Confirm the current limit or the stall margin protects them.

  5. Repeat after life

    The whole budget re-measured at end of life cycling. Lost motion grows and the locking margin falls.

Send the stage and its tolerance for an application review

Send the travel, the lead, the tolerance and whether the approach can be unidirectional. You get back an error budget with each term separated, and the candidates that fit inside it.

Send the stage and its tolerance for an application review