The Coupling That Failed in 30 Days: Misalignment, Not Torque
A servo motor driving a ball screw through a jaw coupling (aluminum hubs, polyurethane spider). After 30 days, the spider cracked. The motor lost torque transmission. The axis stopped. The coupling was rated for 20 Nm. The actual torque was 5 Nm. Torque wasn’t the problem. Misalignment was.
What the failed spider looked like
The spider (the polyurethane element between the two jaw hubs) had cracks radiating from the center. One of the three legs was broken off. The rest were deformed. The aluminum hubs were fine. The set screws were tight. The shaft keys were intact.
The spider didn’t fail from overload. It failed from angular misalignment. The motor and screw weren’t perfectly coaxial. Every revolution, the spider flexed to compensate. At 3000 rpm, that’s 50 flexes per second. The polyurethane fatigue limit was reached in 30 days.
The three misalignment types
| Misalignment type | Description | Typical tolerance for jaw coupling |
|---|---|---|
| Parallel (offset) | Shafts are parallel but offset | 0.1–0.2 mm |
| Angular | Shafts meet at an angle | 0.5–1.0° |
| Axial (end float) | Shafts move in/out | 0.2–0.5 mm |
Our coupling had about 0.3 mm parallel offset. That’s 3× the rated tolerance. The spider was working overtime to compensate. At 3000 rpm, the fatigue life is measured in days, not years.
The alignment process
We re-aligned the motor and screw using a dial indicator. We loosened the motor mount bolts, adjusted the motor position until the offset was under 0.05 mm (measured at the coupling OD), and re-tightened. The spider still needed replacing — it was already cracked. We installed a new spider. The coupling ran for 18 months without another failure.
When a jaw coupling isn’t enough
A jaw coupling handles small misalignment. If the machine has thermal growth (the screw expands when it warms up), or if the mount is a bolted plate that shifts, a jaw coupling won’t last. Use a bellows coupling or a beam coupling for higher misalignment tolerance. A bellows coupling handles 0.3–0.5 mm parallel and 1–2° angular — without the spider fatigue.
| Coupling type | Parallel tolerance | Angular tolerance | Best for |
|---|---|---|---|
| Jaw (elastomer spider) | 0.1 mm | 0.5° | Low misalignment, shock loads |
| Bellows | 0.3 mm | 1.5° | Servo, precision positioning |
| Beam (helical) | 0.2 mm | 2° | General servo, compact |
| Disc | 0.5 mm | 0.5° | High torque, high speed |
| Oldham | 0.5 mm | 0.5° | Large parallel offset, low speed |
The calculation that matters
Coupling selection isn’t just torque. The speed matters. At high speed, misalignment causes reaction forces on the bearings. The ball screw bearing and the motor bearing see side load from a misaligned coupling. That side load shortens the bearing life.
F_reaction ≈ M_misalign · k / L
Where M_misalign is the misalignment moment (a function of coupling type and offset), k is the coupling stiffness, and L is the shaft separation. For our 0.3 mm offset jaw coupling, the side load on the screw bearing was about 15 N. Small, but it adds to the bearing’s radial load. The bearing was rated for 50 N dynamic. At 15 N side load plus 20 N belt load, it was running at 70% capacity. That’s why the bearing was also noisy.
What we changed
Re-aligned to 0.05 mm offset. Replaced the cracked spider. Switched to a bellows coupling (for future builds) because the machine’s mount shifts by 0.2 mm during warm-up. The bellows absorbs that shift without fatiguing. The coupling that failed in 30 days is now running for years.
The alignment rule: A coupling that fails quickly isn’t undersized on torque. It’s misaligned. Measure the offset with a dial indicator. Get under 0.05 mm for a jaw coupling. If the machine shifts (thermal, vibration), use a bellows coupling instead. The spider is the sacrifice component — it tells you when the alignment is bad.
