A 5M-15 timing belt on a labeling machine broke after nine months. The belt was 15 mm wide, 360 teeth, running over a 28-tooth drive pulley and a 56-tooth driven pulley. The load was a rotary labeling head that indexes and holds position. The belt snapped at a tooth. The customer ordered the same belt again. It broke again after eleven months. The engineer was ready to blame the belt manufacturer until someone measured the actual tooth load and found the belt was being asked to carry three times its rated tooth shear capacity. The belt was not defective. It was undersized by design.
Timing belts fail in two distinct ways. Tensile failure: the belt cord (fiberglass or steel) snaps from being stretched beyond its breaking strength. Tooth shear: a tooth tears off the belt body because the torque on the pulley exceeds the tooth’s grip capacity. The two failures look different, are caused by different things, and are fixed differently. The labeling belt broke at a tooth, which is the shear mode. The fix is not a stronger belt of the same size — it is a wider belt or a larger pulley, because tooth shear capacity scales with width and with the number of teeth in mesh.
Doing the tooth shear check
The torque on the drive pulley: the labeling head indexes at 30 cycles per minute with a peak torque of 6 N·m at the drive. The 28-tooth pulley has a pitch diameter of D = z × p / π = 28 × 5 / π = 44.6 mm. The tangential force on the belt: F = 2 × T / D = 2 × 6 / 0.0446 = 269 N.
The tooth shear capacity of a 5M belt, 15 mm wide, with a 28-tooth pulley at 180 degrees of wrap: the number of teeth in mesh is about half the pulley teeth, or 14. Each tooth of a 5M belt on a 28-tooth pulley carries about 30 N at its rated capacity for a 15 mm wide belt. Total capacity: 14 × 30 = 420 N. That is above the 269 N load, so the belt should survive — on paper. The margin is 1.5x. In practice, the labeling head has a 2 N·m overload at each index start, and the belt has a 25% derate for the small pulley (28 teeth on 5M is below the recommended 30-tooth minimum). The real capacity: 420 × 0.75 = 315 N against a peak of 2 × (6+2)/0.0446 = 359 N. The belt is overloaded by 14%. It fails at the tooth, nine months in.
The numbers are worth repeating because they show how the margin evaporates: catalog capacity 420 N, derated 315 N, actual peak 359 N. The belt was designed with a comfortable-looking 1.5x margin that disappeared under the derate and the overload. No belt manufacturer’s warranty covers that gap. The failure was not a defect — it was an application error that looked like one.
Why the small pulley kills the belt
The small pulley is the single biggest factor in tooth shear. A 28-tooth pulley on a 5M belt means each tooth wraps around a tight radius. The belt tooth is bent sharply. The rubber at the tooth root is pre-stressed by the bending. The shear capacity drops. The manufacturer’s rating assumes a pulley with at least 30-36 teeth. Below that, the derate applies — typically 5% per tooth below the recommended minimum.
The fix: increase the drive pulley from 28 to 36 teeth. The pitch diameter rises from 44.6 to 57.3 mm. The tangential force drops from 269 N to 209 N (for the same 6 N·m torque). The teeth in mesh rise from 14 to 18. The capacity rises from 420 to 540 N, derated for the now-acceptable pulley: 540 × 0.95 = 513 N. The load is 209-279 N against a capacity of 513 N — a 2x margin that survives the overload. The belt that failed in nine months now runs three years and is replaced on a schedule.
The pulley change cost $60. The customer also upsized the belt width from 15 to 25 mm for good measure, which doubled the capacity again. The belt now outlasts the machine’s other wear parts. The original failure analysis took one afternoon of calculation and one look at the failed belt’s break location.
Reading the failure mode from the broken belt
The broken belt tells you which failure mode it was, and the fix follows from that.
- Tooth torn off, rubber at the tooth root, cord intact: tooth shear. Widen the belt or enlarge the pulley.
- Cord snapped, clean break across the belt width: tensile. The belt is stretched beyond capacity; reduce the load or use a stronger cord (steel instead of fiberglass).
- Teeth worn smooth, belt still whole: abrasion or wrong pulley profile. Check the pulley tooth wear and the belt tension.
- Edge fraying: misalignment. The belt is rubbing against a flange or an adjacent component.
- Backside cracked, teeth fine: backside idler over-tensioned or running over a too-small backside radius.
The labeling belt broke with a clean tooth tear and the cord intact. That is the signature of tooth shear. The customer’s first response — order the same belt — treated a sizing problem as a part quality problem. The second failure proved it was not the part.
The tension trap
Timing belts are often over-tensioned “to be safe.” Over-tension stretches the belt, reduces belt life, and increases bearing loads. The correct tension for a 5M-15 belt is measured by deflection: a mid-span force that deflects the belt 1.6 mm per 100 mm of span. The labeling machine’s belt was tensioned by feel — pulled until it “felt tight.” That was about 30% over the correct tension. The over-tension added to the tooth load at the small pulley. The combination of under-sized pulley and over-tensioned belt is the standard way to kill a timing belt in under a year.
The tension check takes two minutes with a force gauge or a fish scale: measure the span length, divide by 64, apply that deflection, and read the force. Compare to the manufacturer’s spec. If the machine has an eccentric tensioner, set it with the gauge, not by feel.
The drive selection checklist
For any new timing belt drive, before ordering:
- Calculate the torque at the drive pulley and convert to tangential force.
- Check the teeth in mesh: at least 6, ideally 10+.
- Check the pulley size against the manufacturer’s minimum for that pitch.
- Apply the small-pulley derate.
- Apply the service factor (1.3-2.0 for indexing, 1.5-2.5 for reversing or shock loads).
- Compare the derated capacity to the peak load, not the average.
- Set the tension by deflection measurement, not by feel.
The labeling machine’s drive failed steps 3, 4, 5, and 7. It passed steps 1 and 2 by a margin that did not survive the derates. The one calculation that would have caught it — capacity after derate against peak load — takes five minutes and requires only the pulley tooth counts and the motor torque.
A timing belt that breaks at a tooth is telling you the tooth shear capacity is exceeded, not that the belt is bad. Check the pulley size, the teeth in mesh, and the derates before reordering the same belt. The labeling belt was overloaded 14% after the small-pulley derate and the index overload. A 36-tooth pulley and a wider belt fixed it for $60.
