GreatLight CNC Machining Factory logo
CNC Machining
Rapid Prototyping
Materials
Industries
News
About GL

Get Instant Quote

CNC troubleshooting

Nine Major Machining Errors and How to Trace Them

This page lists nine major machining errors that show up on the inspection report, then pairs each one with the likely cause and the adjustment that clears it. It is written for engineers and shop leads who need to decide fast whether to re-cut, re-fixture or re-program. No theory dump, just symptom, cause, action.

±0.005 mm capability127 CNC machines100% inspection
CNC Knowledge: nine major machining errors, have you encountered them?
Symptom map

Nine major machining errors: symptom, cause, action

Read down the left column from your inspection report. The middle column is the most common cause, not the only one.

SymptomMost likely causeFirst action
Hole sits off nominal positionFixture moved or thermal driftRe-datum, warm up spindle 15 min
Bore out of roundWeak boring bar, too much overhangShorten holder, reduce depth 0.2 mm
Flat face not flatWorkpiece lift during face millingAdd clamps, reduce feed 20%
Slot width drifts along lengthTool deflection on deep passesStep down, use 4-flute cutter
Surface torn or smearedWrong speed for the alloyRaise surface speed 30–50 m/min
Thread galls on assemblyPitch diameter cut too tightRe-cut to 6H, check with ring gauge
Part measures short after coolingCutting heat, no dwellCool in air 30 min, then measure
Corner radius oversizedCutter runout above 0.01 mmIndicate holder, replace collet
Repeat parts scatter over a runTool wear not compensatedSet wear offset, re-check every 20 parts
Definition

What counts as a machining error

A machining error is the gap between the geometry you asked for and the geometry the part actually has. That gap has three parts: size, shape and position. A hole can be the right diameter and still be in the wrong place. A face can be in the right place and still not be flat. When we talk about nine major machining errors, we mean the faults that keep showing up across those three categories on real inspection reports.

Accuracy is how close the group of parts sits to nominal. Repeatability is how close the parts sit to each other. They are not the same problem and they do not have the same fix. If ten parts all measure 0.03 mm oversize, your offset is wrong and one number fixes it. If ten parts measure anywhere between ±0.04 mm, you have a stability problem: fixture, spindle, tool or temperature. Diagnose the pattern before you touch the offset.

Most shops quote ±0.005 mm capability, and that is achievable on a rigid setup in aluminium with a warm machine. It is not achievable on a thin-walled part held in a three-jaw chuck with a long tool. The tolerance is a property of the whole setup, not the machine nameplate.

So before you chase any single error, classify it. Is it constant across parts, drifting across parts, or random across parts? Constant points to the program or the offset. Drifting points to heat or tool wear. Random points to clamping, chip evacuation or spindle condition. The table above is organised that way.

Size and position

Errors one to three: size, position and datum

Error one is a consistent size offset. Every part is 0.02 mm large on the same bore. The cause is usually a worn or mis-set tool offset, or thermal growth that was not measured when the offset was taken. Fix: measure the part at shop temperature, adjust the wear offset by the measured deviation, and run three parts to confirm. Do not adjust the geometry offset unless the deviation is larger than 0.1 mm.

Error two is position error. The hole is the right size but sits 0.05 mm off nominal. Check the datum first, because position errors usually trace back to how the part was located, not how it was cut. If the fixture has a 0.02 mm burr under a locating pad, every part inherits that shift. Clean the pads, re-indicate the fixture, and re-cut one part before you touch the program.

Error three is datum shift between operations. Op 1 is fine, op 2 is off. This happens when the second setup references a surface that moved after the first cut, for example a face that sprang back after unclamping. The fix is to reference the same physical feature in both operations and to check that the first face is flat before you clamp on it. On thin parts, rough, unclamp, let the part settle, then finish.

All three of these are constant errors. They repeat. That is good news, because a constant error is a number you can correct. Write it down before you change anything, so you can tell whether your fix worked.

Geometry

Errors four to six: roundness, flatness and wall thickness

Error four is out-of-round bores. A boring bar with a 4:1 overhang will flex under load, and the bore comes out lobed. Measure with a bore gauge at three depths and two axes. If the reading changes with depth, the bar is the problem. Shorten the overhang, increase the bar diameter, or reduce depth of cut to 0.2 mm and add a spring pass. A Ø400 mm rotary table does not help here; stiffness does.

Error five is a face that is not flat. On a face mill, the usual cause is workpiece lift: the cut pulls the part up off the parallels and it springs back. Listen for chatter at the exit edge. Add clamps near the unsupported corner, reduce feed per tooth by 20 percent, and check that the inserts are all at the same height. A face that is 0.01 mm concave across 200 mm is usually lift, not a machine level issue.

Error six is wall thickness variation on thin parts. The wall measures 1.98 mm at one end and 2.04 mm at the other. The part moved during cutting because residual stress released as material came off. Rough to within 0.5 mm, stress-relieve if the alloy allows, then take light finishing passes of 0.15 mm on both sides to balance the load. Do not finish one wall completely before starting the other.

These three are shape errors. They scale with load and with how the part is held, so they respond to setup changes more than to offset changes.

Surface

Errors seven to nine: finish, threads and heat

Error seven is a torn or smeared finish. Aluminium smears when the surface speed is too low or the chip load is too light, so the edge rubs instead of cutting. For 6061, run 300–500 m/min with a 0.05–0.10 mm feed per tooth and plenty of coolant. Stainless 316 wants lower speed, 120–180 m/min, but a heavier feed so the edge stays under the work. A Ra 0.8–1.6 μm finish is a reasonable target for a normal milling pass; Ra 0.2–0.8 μm needs a finishing strategy, not just a slower feed.

Error eight is thread problems. Galled threads on assembly almost always mean the pitch diameter was cut tight, or the thread was cut with a dull tool and the flanks tore. Check with a ring gauge or a thread micrometer, not with a nut. If the gauge goes on with two fingers, the thread is fine. If it needs a wrench, the tool is dull or the minor diameter is undersize. Re-cut to 6H and inspect 100 percent of the batch.

Error nine is heat. A part that measures on size at the machine and 0.03 mm small two hours later was measured hot. Titanium and stainless hold heat in the part, not in the chip. Let parts cool in still air for 30 minutes before final inspection. If you cannot wait, measure at a fixed temperature and apply a known correction, but write that correction down or the next shift will not use it.

Finish, thread and heat errors are the ones that pass at the machine and fail at the customer. That is why final inspection belongs after cooling, not during the cut.

Trace it

How to trace a machining error in seven steps

Work in this order. Skipping to the offset is the most common mistake.

  • 1
    1. Read the whole report firstCollect every measurement, not just the one that failed. Look for a pattern: one feature, one axis, one operation, or all features. A single-feature error points at a tool or a program block. An all-feature error points at the fixture or the machine.
  • 2
    2. Classify constant, drift or randomCut three parts and measure the same feature on all three. Same deviation on all three is constant. Growing deviation is drift, usually heat or wear. Scattered values inside a band of 0.02–0.04 mm are random, usually clamping or chips.
  • 3
    3. Check the datum and the fixtureWipe locating pads, look for burrs and chips, and indicate the fixture before touching the program. A 0.02 mm burr under a pad shifts every part in the same direction. This check takes ten minutes and clears a large share of position errors.
  • 4
    4. Verify tool condition and runoutIndicate the cutter in the holder. Runout above 0.01 mm will oversize slots and corners. Check flank wear on the inserts; a dull edge raises cutting force, which bends thin walls and pushes bores out of round.
  • 5
    5. Confirm the machine is thermally stableRun the spindle 15–20 minutes before taking offsets. A cold spindle can move 0.02–0.05 mm in the first hour. If the first part of the shift is always the worst one, this is your cause.
  • 6
    6. Change one variable at a timeAdjust the wear offset, the speed, or the clamping, but not all three. Cut one part and measure. If the error halves, you are on the right track. If it does not move, undo the change and pick a different cause.
  • 7
    7. Re-qualify the process before the runOnce the part is good, run five more and measure all of them. If the spread stays inside the tolerance band, release the run. If it does not, the fix was a coincidence, and you need to go back to step two.
FAQs

Questions engineers ask about machining errors

Which of the nine major machining errors is the most expensive?

Position and datum errors, because they usually reach the customer. A size offset is caught by a gauge at the machine. A hole that is the right size but 0.05 mm off nominal passes every simple check and fails at assembly.

Catching it needs either a CMM report or a functional gauge that locates from the real datum. Ask for the inspection method before the run starts, not after.

Can a machining error be fixed by adjusting the offset?

Only a constant size error. If every part is 0.02 mm large, the offset is the right tool. If the parts scatter, an offset will just move the scatter to a different place.

Shape, position and finish errors respond to setup, tooling and speed changes. Treat them as process problems, not number problems.

How do we keep thin-wall parts from moving during cutting?

Rough with 0.5 mm of stock, unclamp, let the part settle, then finish with light passes of about 0.15 mm on alternating sides. Support the wall where you can and keep the tool short.

If the alloy has high residual stress, a stress-relief step between roughing and finishing removes most of the movement.

What surface finish should we specify for a machined part?

Ra 1.6–3.2 μm is a standard as-machined finish and costs nothing extra. Ra 0.8–1.6 μm is a normal finishing pass on most alloys. Ra 0.2–0.8 μm needs a dedicated finishing strategy, sometimes a second operation.

Specify the number only where it matters. A blanket fine-finish call on every face adds cost without adding function.

Do we need to inspect 100 percent of a batch?

For tight-tolerance features, yes. Tool wear drifts across a run, so the first and last part are the ones most likely to fail. Sampling can miss a slow drift that starts midway.

In-process monitoring plus a final inspection catches both the drift and any single bad part.

When should we send a part back to the machine rather than scrap it?

If the error is a size error and there is stock left to remove, re-cut it. If the error is a position error, re-cutting usually will not help because the feature is already in the wrong place.

Decide on the measurement, not on the schedule. A scrapped part costs less than a returned assembly.

Send us the drawing and the failing dimension

We quote in 12 hours with a free DFM review, and we will tell you which of the nine major machining errors is likely to bite on your part before the first chip is cut.

12-hour quote100% inspection±0.005 mm capability

Follow

More machining notes

We publish setup notes, tooling trials and inspection data from the factory floor.

FacebookTikTokYouTubeLinkedInInstagramThreadsPinterest

Trusted by engineers and manufacturers worldwide

Tesla Ford Motor Company BYD Auto Denso Magna International Boeing Airbus Medtronic KUKA FANUC