WORK ORDER — Maintenance and Improvement Record
Work order no: WO-A-2026-05 Date issued: 2026-09-11 Category: Pallet changer
Machine / cell: Rotary pallet changer on Machining Centre MC-4
Status: Closed Engineer: L. Brandt (machine tool) Total labour: 8.0 hours Parts cost: EUR 210.00
Reason for work: parallelogram error growing on machined faces of parts from the B-pallet
1. Fault description as logged by the shift team
Friday 09:30, operator record: parts machined from the changer B-pallet come out with a growing out-of-square error, the milled face sits 0.06 mm higher at one corner than the opposite corner over a 250 mm face, while parts from the A-pallet stay square within 0.02 mm, and the finish pass has started leaving witness marks because the tool tries to cut a twisted plane. The changer is a rotary pallet interchange, a round platen that carries two pallets, A and B, and pivots 180 degrees on a vertical bearing so the operator can load one pallet while the machine cuts the other, and the pallets register into a clamping station that does the fine positioning.
The geometric fault, a twisted plane on the B-pallet only, is the classic signature of a platen that is not stopping at the same angular position every cycle, because a platen that over- or under-shoots its 180 degree index lands the pallet at a slightly different orientation each time, and the clamp station then forces the pallet flat against a wedge error, twisting the part plane before the first cut starts. The machine operator set the B-zero twice in a month, assuming the pallet surface shifted, but the surface was fixed, the twist was in how the platen parked the pallet. I carried a dial indicator and a clinometer to the machine rather than another set of offsets, because a growing squareness error that lives on one pallet only is a changer geometry fault, not a tool or a thermal thing.
Evening note to the file: a rotary changer that parks two millimetres short in angle turns every B-pallet part into a twisted plane that no tool offset can nurse, and the clinometer is the instrument that points at the parking, not the process.
2. On-site inspection record (Day 1)
I set the dial indicator tip on the platen edge at the 300 mm radius and ran the changer through twenty full 180 degree interchange cycles, logging the platen stop position in the machine coordinate readout, and the record showed the platen parking anywhere between the nominal angle and 0.14 degrees short, with the short parking clustering in the last five cycles before I ran the test. The clinometer across the B-pallet surface confirmed the consequence, the pallet plane tilted 0.05 mm over its 250 mm span when parked short, which is exactly the 0.06 mm out-of-square the operator had measured on the parts, so the angle error and the part twist were the same number seen from two ends.
I took the changer cover off to look at the index mechanism and found the drive belt stretched: the platen is driven by a toothed belt through a 12-to-1 reduction, and the belt had a visible 2.1 mm stretch on its loaded run plus a row of worn teeth where it had been slipping over the drive sprocket. The belt stretch alone is a 0.11 degree index loss at the platen, close to the worst parking we logged, and the worn teeth explain why the error grew over the shift, because a slipping belt grips at first and then lets go as the transition torque spikes when the platen crosses its dwell detent. The platen brake, a spring-applied caliper on the driven shaft, was the second finding: its brake shoe had worn to 6.2 mm against a 5.0 mm minimum wear limit, so the brake no longer held the platen solidly at the docked angle, letting the platen back-drive a little after parking and turn a clean index into a wandering one.
Evening note: the changer had been telling us the truth for a while, a stretched belt and a thin brake shoe both rob the platen of its ability to park where it is told, and every short parking showed up as one more micron of twist on the B-pallet parts.
3. Corrective action taken
I replaced the drive belt with the maker’s toothed belt and tensioned it to the specified 260 Hz by the belt tension gauge, then checked the belt seating on both sprockets so the teeth meshed clean over the full wrap. I replaced the brake shoe set on the platen caliper, the old shoes were below the 5.0 mm wear limit and could no longer hold the docked platen, and I set the brake torque to the maker’s docking value so the platen parks and stays parked. With the belt and brake done I commissioned the index: twenty five full interchange cycles, and the platen now stops within 0.01 degrees of the nominal 180 every time, with the clinometer across the B-pallet reading the same 0.005 mm flatness it shows on A, so the changer is parking honestly again.
While the cover was off I put the platen bearing on the inspection list, the vertical bearing takes the whole interchange load and its preload had relaxed a little, so I re-torqued the bearing collar to the drawing setting to keep the platen running true in the same plane rather than wobbling in elevation. I also cleaned the detent pocket and re-greased it, because the detent wedge parks the platen at the last micron and a dirty pocket lets it rock. The duty sheet now includes the belt tension and the brake shoe thickness as a quarterly check, so the interchange cannot stretch its way back into twisted planes without an audit catching it first.
| Item | Before | After | Spec |
|---|---|---|---|
| Platen index parking range | 0 to -0.14° | ±0.01° | ±0.02° |
| Drive belt | 2.1 mm stretch, worn teeth | new, 260 Hz tension | no stretch |
| Brake shoe thickness | 6.2 mm (below 5.0 limit wearing) | new set | ≥7.0 mm |
| B-pallet plane µm/250mm | 0.05 mm twist | 0.005 mm | ≤0.01 |
The geometry note for the file: a rotary changer is a machine tool in miniature, its index accuracy writes itself onto every pallet face, so the belt and brake are not consumables to run to destruction, they are the repeatability of the whole cell dressed as spare parts.
4. Acceptance result after the repair
I ran the changer through a full production day, 36 interchange cycles with parts on both pallets, and the machined faces now hold squareness within 0.02 mm over the 250 mm face on both pallets, with no witness marks on the finish pass, because the tool no longer has to cut a twisted plane. The machine coordinate readout locked the platen at the same index every cycle, the B pallet registered flat against the clamp station, and quality accepted a full batch of 40 parts without a single re-work. I left the machine running the standard programme with the original offsets untouched, proof that the geometry fix, not an offset fight, was the cure, and after a week of production the squareness log stays flat on both pallets, so I closed the order and released the changer.
5. Hours and parts list
- Labour — 8.0 h total: index angle and clinometer inspection 2.5 h, belt and brake replacement 2.0 h, bearing collar re-torque and detent service 1.5 h, commissioning and production day verify 2.0 h
- Parts — toothed drive belt EUR 45.00, brake shoe set EUR 38.00, bearing collar set and grease EUR 47.00, detent pocket parts EUR 30.00, misc EUR 50.00; total EUR 210.00
- Rejected in the drift window — 6 parts re-machined for squareness, none scrapped, all recovered by a finish pass after the geometry was restored
6. Retrospective notes for the next engineer
A squareness error that lives on one pallet of a rotary changer is almost never the tool and almost always the parking, because the platen parks a fraction short in angle and the clamp station then twists the pallet flat against that wedge, machining a twisted plane into every part. The dial indicator on the platen edge and the clinometer across the pallet turned a vague quality complaint into two numbers in twenty minutes, and the stretched belt and the thin brake shoe were exactly the two parts that rob a changer of its index honesty. Check the belt for stretch, the brake shoe for the wear limit, and the bearing collar for preload, then commission with a twenty cycle parking log before you let it cut, because a changer that parks within 0.01 degrees gives the machinist his geometry for free, and a changer that parks short sells it back as twisted parts.
Attachments: index parking log, clinometer readings, belt and brake measurement sheets, one-day production acceptance record (filed with WO-A-2026-05).
Glossary — pallet changer: mechanism swapping pallets in and out of a machine; platen: the rotating plate carrying the pallets; index angle: the angular position a rotary mechanism stops at; detent: a pocket that parks the platen at the last micron; brake shoe: the friction element stopping and holding the platen.
Field note — the geometry that turns an angle error into a twisted plane
It helps the next engineer to see the arithmetic before it becomes a part failure, because a 0.14 degree error sounds too small to matter until you multiply it out. The platen parks short by an angle, say 0.14 degrees, and the B-pallet clamp station then pulls the pallet into its registers, forcing the far edge of the pallet to rotate down by the sine of that angle over the pallet length, so over a 250 mm pallet span a 0.14 degree parking slip makes the far corner sit 0.14 times pi over 180 times 250 mm below the near edge, roughly 0.61 mm if the platen never corrected, but the clamp spring compliance and the register geometry share that error, which is why the measured plane twist settled around 0.05 mm instead of the full arithmetic value. Either way the finish pass sees a plane that is not a plane, it is a gentle screw surface, and the tool, set to a flat plane, cuts high at one corner and bites deep at the other, leaving the witness marks and the twisted face the quality office measured. Two maintenance truths come out of the arithmetic. First, index accuracy is on the same order as the part tolerance budget, so a changer that parks within 0.01 degrees contributes nothing and a changer that drifts 0.14 degrees contributes the whole defect. Second, the belt and the brake wear in opposite directions, the belt stretches and parks the platen short, the shoe wears and lets the platen rock, so they hide each other on a poorly kept sheet: a fresh belt on a worn brake still parks short after docking, and a fresh brake on a stretched belt never reaches the dock cleanly. Keep both on the quarterly list, log the parking angle with the dial indicator on the platen edge, and you will catch the interchange before it writes the twist into a single part.
