Cross-referencing a motor without requalifying the mechanism

Which parts of a motor change can be closed by comparing data, and which parts force the mechanism back onto a test bench?

A cross-reference is usually treated as one claim. It is three. The replacement fits the mechanism, the replacement drives the mechanism, and the replacement survives the mechanism for as long as the original did. Those three claims are closed by three different kinds of evidence, and only the first can be closed by comparing two data sheets.

Treating them separately is what keeps the requalification small. When the interface is identical, the operating point is inside both machines' continuous capability, and only the parts that touch wear have changed, the test program can be narrowed to the wear question instead of restarting the whole mechanism qualification.

Claim one: the part fits

Fit is geometry, and geometry is the only part of a motor change that a drawing can settle completely. It is also the part most often described in words instead of dimensions. "Same 16 mm motor with a 3 mm shaft" is not a fit statement. The fit statement is a list of controlled features, each with its own tolerance and its own datum.

bolt circle radiusFront view, output endhousing outside diameterpilot diameterSide sectiondatum A: mounting faceshaft extensionshaft dia.body length
The features that decide whether a replacement drops in. Each is dimensioned from the mounting face, not from the housing, because the mounting face is what the mechanism actually locates on.

The mounting face is the datum in almost every mechanism, and the mistake that follows from ignoring that is predictable. Two motors with the same overall length and the same shaft diameter can still put the pinion at different axial positions, because the length that matters is from the mounting face to the shaft end, not from the back of the can. A 0.5 mm shift there moves a spur pinion halfway off its mating gear face width.

Interface features, and what closes each one
Feature What has to match What usually decides it Evidence that closes it
Mounting face and pilotPilot diameter, pilot depth, face flatnessPilot fit class. A slip fit that becomes a light press changes assembly force and concentricityDrawing comparison plus one measured sample
Bolt patternCircle diameter, hole count, thread size and depth, angular clockingThread depth. A shorter blind thread lets the same screw bottom out and distort the endbellDrawing comparison
ShaftDiameter, tolerance class, extension from the mounting face, flat or D-cut position and depthShaft diameter tolerance. A press-on pinion is sized to a specific interference bandMeasured sample, not a nominal dimension
Rotation and lead-outDirection for a given polarity, lead exit position, lead gauge and length, connectorGear stage count. An odd stage count reverses output directionDrawing comparison plus a bench check
EnvelopeMaximum diameter, body length, any local bulge at the terminal or lead exitLocal features, not the nominal diameterMeasured sample against the mechanism model

Claim two: the part drives the mechanism

Performance comparison is where paper cross-references usually go wrong, because the two numbers most often compared are the two numbers least connected to the application: no-load speed and stall torque. Both are endpoints of a line. The mechanism runs somewhere in the middle of it.

n = n0 · (1 − T / TS)

n
shaft speed at the applied load, rpm
n0
no-load speed at the applied voltage, rpm
T
load torque at the shaft, mN·m
TS
stall torque at the applied voltage, mN·m
The linear speed-torque line for a brushed DC motor at fixed voltage. It is accurate enough for selection near the rated point and progressively optimistic near stall, where heating raises winding resistance and the magnetic circuit stops behaving linearly.

Two motors can share a no-load speed and a stall torque and still behave differently in the same mechanism, because those endpoints do not carry the torque constant, the terminal resistance, or the rotor inertia independently. Compare the pair that actually sets behavior under load: the torque constant and the terminal resistance. The torque constant sets how much current a given load draws. The terminal resistance sets how much of the supply is lost heating the winding rather than turning the shaft.

Inertia is the one that gets missed entirely, because it never appears in a speed-torque comparison. In an indexing or dosing mechanism the acceleration time is a real requirement, and it is set by rotor inertia far more than by anything else in the motor.

t = J · ω / Ta

t
time to reach the target speed, s
J
rotor inertia reflected to the motor shaft, kg·m²
ω
target angular velocity, rad/s
Ta
torque available for acceleration, N·m
Illustrative: a rotor of 0.9 g·cm² (9 × 10⁻⁸ kg·m²) reaching 6000 rpm (628 rad/s) with 3 mN·m of accelerating torque takes about 19 ms. A replacement with twice the rotor inertia and the same speed-torque line takes about 38 ms. Both figures assume the accelerating torque is constant, which is a simplification: the real acceleration is faster at the start of the ramp and slower near speed.

For a gearmotor the same comparison has to be done at the output, and the gearhead is not a transparent element. Reduction multiplies torque by the ratio and divides speed by it, but efficiency is a function of stage count, gear type, lubricant and load, and it is lowest exactly where a light mechanism runs: at a small fraction of rated torque. A four-stage planetary and a four-stage spur train with the same nominal ratio can differ by tens of percent in delivered torque at low load, and by more in drag torque.

Claim three: the part survives

The third claim is the one comparison cannot close, and it is the one that fails eight months after the change. Life is set by the parts that wear: the brush and commutator system, the bearing type and its preload, the grease, the axial endplay, and the gear material and its tooth quality. None of these have a specification line that predicts field life in a specific duty cycle.

  • Brush system. A precious-metal commutator system and a carbon commutator system have different contact drop, different current density limits, different wear rates and completely different tolerance for arcing at a reversal. See the article on brush selection.
  • Bearings. Sleeve and ball bearings differ in radial load capacity, drag, noise signature, temperature sensitivity and life under a side load. A sleeve-bearing motor into a belt-loaded mechanism is a wear change, not a fit change.
  • Grease. Base oil viscosity and thickener set drag torque at cold start, migration under vibration, and evaporation rate at temperature.
  • Endplay and preload. Axial float that is acceptable in continuous rotation becomes an audible knock in a reversing duty cycle.
  • Gear material and quality grade. Two gears with the same module, tooth count and material can have different tooth quality, different runout and different noise, and none of that appears on a catalog page.

The requalification matrix

The useful output of a cross-reference is not the candidate part number. It is the short list of tests that the specific differences make necessary. If nothing in a row changed, that row's test does not have to be repeated.

What changed, and what that forces back onto the bench
What changed What it can affect Test that closes it
Nothing but the labelNothing, if the drawing, the source process and the revision are identicalIncoming dimensional and electrical characterization of one lot
Winding, at the same envelopeCurrent draw, speed under load, thermal rise, controller current limit interactionTorque-speed verification and a thermal run at the application duty cycle
Gear stage count or gear typeOutput direction, backlash, drag torque, noise, efficiency at low loadBacklash measurement, noise measurement, and output torque at the application point
Bearing type or endplayNoise under reversal, radial load capacity, drag at cold startEndplay measurement, noise under the real motion profile, and a cold-start current check
Brush or commutator systemLife, electrical noise, contact drop, behavior at low dutyLife test at the application duty cycle. Nothing shorter substitutes for it
LubricantCold-start torque, drag, noise, migration, long-term efficiencyCold-start current and torque at the temperature extremes of the application

The rows are ordered deliberately. The tests near the top are hours of bench work. The test near the bottom is weeks of running, and it is the only one that answers the question the field will ask. Planning a cross-reference means deciding early whether the schedule can carry that test, because a program that cannot wait for it is a program that has decided to accept the risk rather than close it.

What to send

The information that makes a cross-reference fast is rarely the part number. A part number identifies a family, and families contain windings and ratios that behave nothing alike. What narrows the answer is evidence about the operating point.

  • The physical part, a nameplate photograph, or a drawing. A failed unit is more informative than a new one.
  • Supply voltage, and whether it is regulated, a battery, or a PWM drive.
  • Measured current in the running mechanism, if a meter has ever been put on it. This single number does more than any catalog comparison.
  • The motion profile: on time, off time, reversals per hour, and whether the mechanism ever stalls on purpose.
  • What the mechanism does when the motor is wrong: too slow, too hot, too loud, tripping the drive, or failing at a known hour count.

Measured current is worth explaining. It converts directly to torque through the torque constant, which means one meter reading tells you where on the speed-torque line the mechanism actually sits. Most cross-reference requests describe the load in terms of what the drawing intended. The current reading describes what the load is.

Article details

Author, technical reviewer, and last review date.

Written by
Micro Motion application engineering
Reviewed by
Micro Motion manufacturing engineering
First published
Last reviewed
Reading time
9 minutes

Corrections

No corrections have been issued for this article. When a figure, a number or a statement here is found to be wrong, the correction is recorded in this block with its date rather than edited silently into the text.

Report an error in this article