Taper, Chatter and a Step on a 250 mm Shaft

Fault 1 - taper from 35.98 to 35.90–35.93 mm over 250 mm. The work is held in the chuck only, with a length-to-diameter ratio of about 7, at 1.5 mm depth and 0.25 mm/rev on a 1.5 kW lathe. Radial cutting force deflects the bar away from the tool, and the deflection is greatest at the free end, which therefore cuts smaller. That is the dominant cause. Chuck bell-mouth or a worn scroll would show as part-to-part variation and runout at the chuck end rather than a consistent taper along the part.

Fault 2 - spiral chatter, 0.02 mm deep. This is a stiffness and resonance problem. A 55 mm tool overhang on a small toolpost, combined with the unsupported work overhang and a 0.25 mm/rev feed, puts the tool and work in a vibration loop. The worn chuck jaws also allow the work to move.

Fault 3 - the 0.05 mm step at 150 mm. This is a re-clamping artefact, not a machine fault: swarf or a burr trapped between jaw and work, jaws bearing on a previously machined diameter, or the part simply repositioned.

Order of checks - measurement first.

1. Verify the measurement. A vernier caliper on a warm part is not adequate for a 0.05 mm tolerance. Check with a micrometer, three positions along the length, two diameters at 90°, after the part has cooled. Do not chase a machine fault that may be a measurement artefact.
2. Tool condition. Insert wear, tool on centre, overhang, shim stack and toolpost clamping - quick, cheap and the most likely chatter source.
3. Workpiece support. Whether the free end was centre-drilled and running on a centre, and whether a steady was used.
4. Chuck condition. Dial-test a ground bar in the jaws to measure bell-mouth and runout, and inspect jaw wear.
5. Machine alignment last. Bed level and twist, tailstock alignment with a test bar. Only relevant once support is re-established, since the taper was produced without a centre.

Taper.

• Centre-drill the bar and run a revolving centre in the tailstock, or use a fixed steady at about two-thirds of the overhang. This supports the free end and removes the deflection that produces the taper - it is the single biggest improvement available on this job.
• Reduce the depth of cut for the finishing passes to 0.5–1.0 mm and take a spring pass (no infeed) at the same setting. The spring pass removes the material left by deflection on the previous cut.
• Measure both ends after the pass, and only adjust tailstock set-over if a taper persists with the centre in place. Adjusting the tailstock to correct a deflection taper simply cuts a deliberate taper into every part.

Chatter.

• Reduce tool overhang to 25–30 mm (roughly 1.5–2 times the shank height) or use a larger-section tool; clamp with the minimum number of shims and set the tool exactly on centre.
• Support the work as above - most of the compliance in the system is in the slender bar, not the tool.
• Change the cutting conditions: once rigidity is restored, raise the speed for the finishing cut (carbide in mild steel wants 120–180 m/min, about 1,100–1,600 rpm at Ø36) and reduce the feed for finish to 0.1–0.15 mm/rev. Increasing speed on an unsupported bar will make chatter worse, so restore stiffness first, then tune speed and depth.
• Check the insert and the chuck jaws; a chipped insert or a loose jaw will chatter regardless of parameters.

Step.

• Finish the diameter in one setting: machine the full 250 mm without re-chucking, using the cross-slide reading and a carriage stop.
• If the part must be re-chucked, use soft jaws or a collet, clean the jaws and register face, fit a chuck stop, and mark the part and jaw positions. Never let jaws grip a finished diameter without protection.
• Keep coolant on and check the insert before the finishing pass, so a worn edge does not force a second setting.

Instrument. An outside micrometer, 25–50 mm range, resolution 0.01 mm, zeroed on a setting gauge before the run. A vernier caliper is not acceptable for a 0.05 mm total tolerance: its jaws and scale introduce errors of the same order as the tolerance itself. Use a surface finish comparator or a portable Ra tester for the 1.6 µm journal requirement, and a dial test indicator on V-blocks on a surface plate to quantify the step and taper along the length.

Temperature. Steel expands about 11.5 µm/m/°C, so a Ø36 part measured 10 °C above ambient reads roughly 4 µm oversize - nearly a tenth of the tolerance. Let each part cool to room temperature on a plate, record the part temperature, and measure at a consistent 20 °C; correct the reading if the part is still warm.

Sampling and records. Measure every part for the next five, then three per shift. On each part measure two diameters at 90° at three positions - 0, 125 and 250 mm - so taper and ovality are both captured. Record on a mean-and-range chart so drift is visible before a part goes out of tolerance.

Acceptance criteria.

• Diameter 35.95–36.00 mm at every position.
• Taper and ovality no more than 0.01 mm over the 250 mm length - tight enough to leave margin inside the tolerance.
• Ra no greater than 1.6 µm on the journal, with no visible chatter marking; a 0.02 mm chatter mark is 12 times the Ra limit and will fail.
• No step detectable with a dial indicator at any point along the machined length.

Capability. With a 0.05 mm total tolerance, ±0.025 mm each side, a process capability index of 1.33 needs a standard deviation no greater than about 0.006 mm. Aim the process mean at 35.975 mm and hold the spread well inside that, otherwise the process will produce out-of-tolerance parts even when it looks acceptable on a few samples.

Gate. Hold a first-off inspection with the supervisor - dimensions, taper, finish - before releasing the remaining nine parts, and stop the run again if the chart shows a trend rather than waiting for a reject.