Analysis and Maintenance of CNC Roll Mill Defects

A roll mill rarely fails without a signature. The shape of the error tells you which axis or which transmission stage is at fault. Measure the roll after grinding and record roundness, taper and surface finish as separate numbers. A single bar chart of those three values across ten rolls is more useful than any vibration log taken at idle.

Cylindricity error that repeats at one angular position on every roll points to the workhead or the wheelhead, not to the roll itself. An error that drifts along the roll length usually comes from bed twist, guideway wear or thermal growth in the headstock. Errors that change with spindle speed belong to the drive train: belts, gears, bearings.

Record the cutting or grinding parameters next to the dimensional result. Depth of cut, feed, wheel or insert grade, and coolant condition all shift the outcome. Without them, two measurements taken a week apart cannot be compared, and the analysis becomes guesswork.

• 1
  Repeating at one angleSuspect the workhead spindle, chuck or drive key.
• 2
  Drifting along the axisSuspect bed level, guideway wear or thermal growth.
• 3
  Changing with speedSuspect belts, gear mesh or bearing preload.

Chatter on a roll surface is a resonance problem, but the frequency band tells you where it enters. Low-frequency marks spaced 20–60 mm apart usually come from the belt drive or a loose motor mount. Fine marks below 2 mm spacing come from the grinding wheel or the workpiece support. Count the marks over a known length and convert to a frequency using the surface speed; the number will match one of the rotating components.

Taper on a roll body has three common causes. The tailstock is out of alignment with the headstock centerline. The bed is twisted because the leveling pads moved after foundation settling. Or the roll itself deflects under the grinding force because the steady rest is set too far from the wheel. Check alignment first, then level, then support position.

Transmission step design matters on older roll mills. A layout with a large single reduction from motor to roll leaves a tall belt ratio and long belt spans, which slip under start-up load and print marks on the work. Splitting the reduction into two or three shorter stages keeps belt length manageable and raises the torque actually delivered to the roll. This is a rebuild decision, not a daily adjustment.

Bearings on a roll grinder work under a mix of radial load and slow rotation, which is the worst case for oil film formation. Grease that has sat for a year will have separated. Pull the bearing housings on the schedule, not after the finish degrades. Check clearance with a dial indicator on the spindle nose, then check the housing bore for fretting marks where the outer ring has crept.

Gear backlash on the feed and transmission stages should be measured with the axis at both ends of travel. Backlash that is constant along the travel suggests uniform wear. Backlash that changes near one end suggests a bent shaft or a worn key. Both need correction before the machine can hold a roll tolerance, because backlash shows up as a step in the surface when the feed direction reverses.

Guideway wear is the slowest defect to appear and the most expensive to ignore. Measure straightness of the bed with a level and a straightedge over the full travel once a year. A bed that is out of level by more than the manufacturer's limit will transfer that error into every roll ground on it. Re-leveling is far cheaper than a bed regrind.

• 1
  QuarterlySpindle runout, belt tension, coolant condition.
• 2
  AnnuallyBed level, guideway wear, gear backlash at both ends.
• 3
  On finish driftBearing clearance and headstock oil temperature.

Adjustment is the right answer when the error is a setup condition: alignment, level, belt tension, coolant flow, dressing depth. These are reversible and cheap. Remachining is the right answer when the error is built into a worn surface: a scored roll neck, a fretted bearing seat, a galled housing bore.

A scored roll journal will not clean up with a stone. If the damage is deeper than a few hundredths of a millimeter, the journal has to be turned and then finished to size, or the roll replaced. Welding and re-turning a journal can work if the heat is controlled and the part is stress-relieved afterwards, but the metallurgy near the weld changes and that must be accepted before the repair is approved.

For housings and bearing seats, the practical limit is roundness and fit. A bore that has gone oval by more than the bearing clearance budget will never hold preload. In that case the housing is remachined oversize and a sleeve is fitted, or the part is replaced. Both paths need the same measurement first: the actual bore diameter at several angles and depths.

Downtime on a roll mill is driven by how long the machine sits waiting for a part. A small spare kit changes the recovery time from days to hours. Keep belts sized to the actual drive layout, not to the original drawing, because rebuilt stages often need a different length. Keep a matched set of bearings for the workhead, since mixing brands in one housing changes preload behavior.

For machined spares, the parts that wear fastest are the ones that should be stocked: bearing housings, steady rest pads, chuck jaws, drive keys, and the end caps that carry the spindle seals. These are simple turned and milled parts with a short process route. A shop that can turn them to your drawing in a few days is worth more than one that quotes the lowest unit price.

Record the as-built dimensions of every spare when it arrives. When the worn part is pulled, compare it to the as-built record, not to the drawing. The difference between those two numbers is the actual wear, and it is the only reliable input for the next interval decision.