5 Methods to Measure the Dimensional Accuracy of Machining Parts

• 1
  Key takeaways
• 2
  What dimensional accuracy of machining parts actually means
• 3
  Calipers and micrometers: fast checks for outside sizes
• 4
  Bore gauges and inside micrometers for holes
• 5
  Height gauges, surface plates and CMMs
• 6
  Five errors that ruin a measurement
• 7
  Step by step: measuring a machined part
• 8
  Which method fits which tolerance

Dimensional accuracy of machining parts means the finished size matches the drawing within the stated tolerance. It is not the same as surface finish, roundness, or position. A part can hit Ø20.000 mm on the caliper and still fail because the bore is oval or the bolt circle is off center.

Three numbers matter on every drawing: the nominal size, the tolerance band, and the datum. Inspectors who skip the datum get repeatable numbers that do not match the customer's report. On a GreatLight job, we hold ±0.005 mm on critical features, so a 0.01 mm setup error is already 100% of the band.

The measurement method has to be at least four times better than the tolerance. For a ±0.01 mm callout, that means a tool good to ±0.0025 mm. A 150 mm caliper rated ±0.02 mm fails that test before the part is even touched.

Before you measure, ask what decision the number drives. Incoming inspection, first article, and in-process checks all need different tools and different sample sizes. One method does not cover all three.

• 1
  Datum firstLocate the part the same way the drawing does.
• 2
  Rule of 4:1Tool accuracy should be four times finer than the tolerance.
• 3
  Same temperatureLet metal and gauge sit at 20 °C for 30 minutes.

Vernier and digital calipers cover 0–150 mm and 0–300 mm. A good 150 mm digital caliper reads to 0.01 mm with an accuracy of about ±0.02 mm. Use them for stock checks, deburring checks, and quick sampling on features with ±0.1 mm tolerance or looser.

Calipers have a weak point: the jaws tilt. On a 25 mm shaft, a 2° tilt reads about 0.02 mm high. Keep the jaws square, use light contact, and measure at three points along the length to catch taper.

Outside micrometers are the next step. A 0–25 mm micrometer with a friction thimble reads to 0.001 mm and holds about ±0.002 mm accuracy. That is fine enough for our ±0.005 mm work on shafts, pins, and plate thickness.

Check the micrometer against a 25 mm setting rod before each session. If it reads 25.003 mm on the rod, the error carries into every part you measure. Clean the anvils with paper, not your fingers.

• 1
  Calipers0–150 mm, ±0.02 mm. Loose tolerances and shop-floor checks.
• 2
  Outside micrometer0–25 mm, ±0.002 mm. Shafts, pins, thickness.
• 3
  AvoidDo not use a caliper on a bore or on a soft plastic part.

Holes need a different tool. An inside micrometer covers 5–30 mm and reads to 0.001 mm, but it measures at one point. A two-point bore gauge with a dial or digital head covers the same range and shows the full length of the bore as you sweep it.

Set the bore gauge to a known ring gauge or to a micrometer-locked size. A 20 mm ring gauge accurate to ±0.002 mm is the reference. Zero the gauge on the ring, then measure the part. Never zero on the part itself.

Sweep the gauge along the bore axis and rotate it slowly. The smallest reading is the true diameter; the largest is the high point of any ovality. A spread over 0.01 mm on a Ø20 H7 bore is a warning sign, not a rounding error.

Bore gauges do not measure position. A hole can be perfect in diameter and still be 0.05 mm off center. Use a height gauge or a CMM for center distance, bolt circles, and wall thickness.

• 1
  Ring gauge firstZero on a certified ring, never on the part.
• 2
  Sweep the axisRotate and slide to find ovality and taper.
• 3
  Ø20 H720.000 to 20.021 mm. A 0.01 mm spread is too much.

A granite surface plate and a height gauge handle steps, heights, and simple layout. A 0–300 mm digital height gauge reads to 0.01 mm and holds about ±0.03 mm over the full travel. That is fine for setup checks and for picking up a datum face.

For anything tighter or more complex, a coordinate measuring machine (CMM) is the standard. A bridge CMM with a touch probe resolves 0.1 μm and reports true position, concentricity, and profile to the drawing. It measures the part in three axes and applies the datum scheme automatically.

CMM time is not free. Probe speed, stylus choice, and the number of points all change the cycle. A 20-point circle takes about 15 seconds; a full first article on a 30-feature bracket can run 45 minutes. Budget for it on first articles, not on every part.

Optical and laser scanners sit between the two. They are fast on free-form surfaces and thin walls, but they struggle on deep bores and on shiny machined faces. Use them when you need a full-field picture, not a single diameter.

• 1
  Height gauge0–300 mm, ±0.03 mm. Steps, heights, layout.
• 2
  CMM0.1 μm resolution. True position, profile, GD&T.
• 3
  ScannerFull-field shape. Weak on deep bores and mirror finishes.

The first error is measuring a hot part. A block that leaves the mill at 35 °C and gets measured at 20 °C shrinks. On a 200 mm aluminium part, that is about 70 μm, which is 14 times our tolerance.

The second is trusting one reading. A single number hides taper, ovality, and burrs. Take three readings and look at the spread before you accept the part.

The third is a dirty gauge. Anvils and probe tips pick up oil film. Wipe them with paper before each setup, and re-zero if the standard reads off.

The fourth is mixing datum schemes. The machine may have used the vice jaw; the drawing may call out the bottom face. Measure from the drawing datum or the number is meaningless.

The fifth is a calibration that lapsed. A micrometer overdue by six months can drift 0.005 mm. Check the sticker before you trust the reading.

• 1
  Hot part200 mm aluminium, 15 °C rise, about 70 μm of false error.
• 2
  One readingHides ovality, taper and burrs.
• 3
  Lapsed calibrationCheck the sticker and the due date.