How axial endplay creates intermittent noise

Why does a motor tick, knock or clatter only on reversals, only at certain speeds, or only in one mounting orientation, when nothing about the gear mesh has changed?

Endplay is the axial clearance that lets a rotor and its shaft move along their own axis between the two surfaces that constrain them. Every practical small motor has some. It absorbs thermal growth, tolerance stack-up in the bearing seats, and assembly variation. It is a design necessity that becomes a noise source the moment the axial force on the rotor changes sign.

rotor stackcommutator and brushesrear bearingfront bearingoutput shaftaxial force on the rotordetail: endplay gapendplay sshoulderbearing face
The clearance is between a shaft shoulder, retaining feature or bearing inner race and the face that stops it. In the assembled motor it is a few hundredths of a millimeter, and it is set by a stack of parts rather than by any one dimension.

What pushes the rotor along its axis

The rotor sits against one thrust face for as long as the net axial force holds it there, and it is quiet while it does. The noise comes from the transition. So the question in an endplay investigation is never whether the assembly has endplay, it is what makes the axial force reverse.

  • Gear thrust. A helical gear or a worm generates an axial force proportional to the transmitted load, and that force reverses with the direction of rotation. A spur mesh is nominally free of axial force, and generates one anyway when the axes are not parallel.
  • Magnetic centering. The rotor is drawn toward the axial position of minimum reluctance in the magnetic circuit. It is a restoring force, so a rotor displaced from magnetic center is pushed back, and a rotor near center has very little force holding it anywhere.
  • External axial load. Anything the mechanism pushes onto the output shaft, including a preloaded coupling, a lead screw, or a user pressing on a mechanism.
  • Commutation and brush geometry. In some constructions the brush contact contributes a small axial component that changes as the commutator wears.
  • Orientation and gravity. On a small rotor, gravity is not negligible against the other forces here, which is why a shaft-up unit and a shaft-down unit can behave differently.

This list explains why the noise is intermittent. Continuous rotation in one direction under steady load holds the rotor firmly against one face and is silent. Reversal, a load that crosses zero, or an operating point where magnetic centering and gear thrust nearly cancel lets the rotor cross the gap.

The size of the impact

The rotor crosses the gap under a roughly constant axial force, so the energy delivered at the far end is the work that force does across the gap. That gives an unusually clean result: the impact energy does not depend on the rotor mass, only on the force and the clearance.

E = F · s, v = √(2 · F · s / m)

E
kinetic energy at impact, J
F
net axial force on the rotor, N
s
axial endplay, m
m
moving mass, kg
v
impact velocity, m/s
Illustrative: 0.3 N across 0.15 mm of endplay delivers 45 µJ. On a 15 g rotor that is an impact velocity of 78 mm/s. The assumptions are a constant axial force across the gap and no viscous damping from lubricant in the joint, both of which make the figure an upper bound.
00.10.20.30.40100200300400Axial endplay (mm)Impact energy (µJ)1.0 N0.3 N0.1 N0.3 N across 0.15 mm: 45 µJ
Impact energy against endplay, from the relation above. Halving the endplay halves the energy delivered at each strike, and reducing the axial force does the same thing.

The practical consequence is that endplay reduction is a proportional fix, not a threshold fix. There is no clearance below which the knock disappears and above which it appears. Each reduction buys a proportional reduction in impact energy, and the audible result depends on how efficiently the housing radiates that energy.

How to tell endplay noise from gear noise

Separating the two most common mechanical noise complaints
Observation Endplay Gear mesh
CharacterImpulsive: a tick, knock or clatterTonal: a whine or growl
TimingTied to reversals, starts, stops and load changesContinuous while running
SpectrumBroadband transient, no fixed orderMesh frequency and its harmonics, with speed
Load dependenceLoudest where the axial force crosses zeroRises steadily with transmitted load
OrientationFrequently changes with shaft up, down or horizontalLargely unchanged by orientation
Response to a thrust loadDisappears or drops sharply when a light axial preload is applied to the shaft by handUnchanged

Fixing it, and what each fix costs

  • Reduce the endplay in the stack. Effective and proportional. Costs tolerance capability, and an assembly built with too little endplay binds when it warms up or when the tolerances stack the other way.
  • Add an axial preload with a wave washer or spring. Removes the free gap without removing the clearance, so thermal growth is still accommodated. Costs a small amount of continuous drag, which shows as no-load current.
  • Add a compliant thrust washer. Turns a metal-on-metal impact into a damped one. Cuts the radiated noise without changing the mechanics, and adds a wear item.
  • Reduce the axial force. Choose a spur mesh instead of helical where the thrust is coming from the gear, or correct the parallelism that generates thrust in a nominally axial-force-free mesh.
  • Change how the impact radiates. Isolating the motor from a large flat panel usually does more for the product's perceived noise than any change inside the motor.

The order matters. Preload is usually the cheapest effective fix, because it removes the free travel without asking the tolerance stack to be tighter. Reducing endplay by tightening tolerances is the most expensive, and it is the one that most often produces a second problem at the temperature extremes.

What to measure

  1. Measure endplay, do not infer it

    A dial indicator on the shaft end while a known light axial load is reversed. Record the load used, because the reading grows with the load applied.

  2. Record it against the noise result

    Endplay measured on the same units that were sorted by noise is what shows whether the two are related in this assembly.

  3. Repeat at temperature

    Endplay changes as the stack expands. A unit that is quiet at 25 °C can be quiet for that reason alone.

  4. Test in the real motion profile

    Endplay noise appears at reversals. A continuous-rotation bench test will not find it.

Where the endplay in the stack comes from

Endplay is not a dimension on one part. It is the sum of a stack: bearing seat depth in each endbell, the axial position of a retaining feature, washer thickness, bearing internal clearance and the length of the shaft between its shoulders. Each contributes its own tolerance, and the assembled clearance is the accumulation.

Temperature moves the same stack. A steel shaft in an aluminum housing is the common case, and the two materials expand at roughly 11 and 23 µm per meter per kelvin. Over a 40 mm span between the thrust faces, a 40 K rise grows the housing about 19 µm more than the shaft, so the endplay grows by that amount. An assembly with 60 µm of endplay at room temperature has about 80 µm at 65 °C, and the impact energy grows with it.

Who wrote it, who checked it, and what has been corrected since.

Every article on this site carries this block. A correction is recorded here with its date rather than edited silently into the text.

Written by
Micro Motion application engineering
Reviewed by
Micro Motion manufacturing engineering
First published
Last reviewed
Reading time
7 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