What Analyzes Elements With CNC Machines?

• 1
  What Analyzes Elements With CNC Machines on the Dimensional Side
• 2
  Geometric Elements: Flatness, Position and Perpendicularity
• 3
  Surface Finish and Edge Condition After Machining
• 4
  Material Integrity and the Tests That Sit Outside the Machine
• 5
  In-Process Monitoring: Measuring While the Tool Is Still Cutting
• 6
  When Analysis Methods Stop Being Useful
• 7
  Which Method Analyzes Which Element

Start with the obvious layer: size. A coordinate measuring machine touches discrete points and reconstructs a feature from them. Bore diameter, slot width, boss height, hole position, wall thickness. That data is compared against the CAD model and the drawing, and the deviation is reported per feature. The CMM is the reference instrument in most shops because its uncertainty budget is well documented and it does not care what the material is.

Typical CMM accuracy on our floor sits well inside the ±0.005 mm tolerance band we hold on tight features, which means the measurement itself is not the limiting factor. The part is. When a feature is marked ±0.005 mm, the machine, the fixture and the thermal state of the part all matter more than the probe.

Hand tools still have a place. Micrometers, bore gauges, pin gauges and height stands are fast and shop-floor friendly. For a 10,000-piece run of a simple turned part, a digital micrometer checked every 30 minutes catches drift long before a CMM queue does. Hand tools do not give you a full geometry picture, but they give you trend.

The real distinction is not good versus bad instruments. It is whether the instrument matches the tolerance. A caliper reading ±0.02 mm against a ±0.005 mm requirement tells you almost nothing useful. Same idea for any analysis step: resolution has to be at least four times tighter than the tolerance you are trying to prove, otherwise you are measuring noise.

• 1
  CMMPoint-by-point contact measurement of size and position; the reference method for ±0.005 mm work.
• 2
  Hand toolsMicrometers, bore and pin gauges for fast in-process trend checks.
• 3
  Rule of thumbInstrument resolution should be at least 4× tighter than the feature tolerance.

Size alone passes parts that do not assemble. A hole can sit at the right diameter and still be 0.08 mm off nominal position, and the mating pin will not go in. Geometric dimensioning and tolerancing exists for exactly this reason, and it is where most CNC analysis time is actually spent.

Flatness and parallelism usually drive the fixturing discussion. A 300 mm plate with a 0.02 mm flatness callout will move when you clamp it, and it will move again when you release it. Measuring flatness on a constrained part is close to meaningless. Good practice is to measure free-state, then measure again on the fixture, then compare.

Position tolerance and true position get checked with the CMM or with functional gauges. On high-volume automotive and EV parts, a go/no-go gauge is often faster than a CMM and just as decisive. The gauge asks one question: does the feature fit the mating condition? That is the question the customer actually cares about.

Concentricity and runout matter on anything that spins. Shafts, impellers, motor housings, spindle components. A roundness tester or a CMM with a rotary table handles this well. For a shaft with a Ø400 mm rotary table setup, runout is often the single number that decides whether the part ships.

• 1
  Free-state firstMeasure flatness unclamped, then clamped, then compare the two.
• 2
  Functional gaugesGo/no-go checks beat CMM throughput on high-volume position tolerances.
• 3
  RunoutThe deciding number for shafts, housings and anything that rotates.

Surface finish is analyzed with a profilometer, usually reported as Ra. The number is a roughness average across a sampling length, and it is easy to game. A surface can read Ra 0.8 μm and still have a directional lay that leaks. If sealing is the function, the lay direction matters as much as the average.

Our standard bands run from Ra 0.2–0.8 μm on fine finishes, Ra 0.8–1.6 μm on high-quality machined surfaces, and Ra 1.6–3.2 μm as-machined. Which band you need is a function question, not a prestige question. A bracket does not need Ra 0.4 μm. A hydraulic sealing face often does.

Edges get analyzed too, and they are the most frequently skipped item. A burr on a medical or aerospace part is a reject, not a cosmetic issue. Deburring, edge breaks and controlled radii are checked visually under magnification and with profile tracing where the drawing specifies a radius.

One boundary worth stating: surface analysis before and after finishing are different measurements. Anodizing adds a few micrometres and will change the Ra reading. Bead blasting will change it more. Measure the finish the customer receives, not the one straight off the cutter, unless the drawing explicitly calls for pre-finish condition.

• 1
  Ra is an averageIt hides lay direction; sealing faces need both numbers.
• 2
  Measure after finishingAnodizing and blasting shift the Ra reading.
• 3
  Edge conditionBurrs and sharp corners are dimensional rejects on safety-critical parts.

Not everything can be measured with a probe. Material integrity analysis covers what is inside the part: voids, porosity, inclusions, cracks and internal stress. These defects are invisible to a CMM and to the eye, and they fail under load rather than on the bench.

X-ray inspection and industrial CT are the methods for internal structure. CT gives a three-dimensional density map and can resolve porosity down to tens of micrometres on a small casting. It is slow and expensive, so it is reserved for mission-critical components: aerospace structural parts, implantable medical devices, high-pressure housings.

Ultrasonic testing is the cheaper alternative for detecting internal cracks and delamination in thicker sections. Dye penetrant inspection finds surface-breaking cracks on non-porous materials. Magnetic particle inspection does the same job for ferrous parts and can find slightly subsurface indications.

Hardness testing rounds out the material layer. Rockwell or Vickers indents confirm that heat treatment did what the certificate says. For Inconel, titanium and 17-4PH work, hardness is often the fastest way to catch a heat-treat mix-up before the part reaches a customer. It is destructive at the indent point, so it lives on test coupons or on non-critical surfaces.

• 1
  X-ray and CTInternal voids and porosity; slow and costly, so used on critical parts.
• 2
  Ultrasonic and penetrantCracks and delamination at lower cost than CT.
• 3
  HardnessFast confirmation that heat treatment matches the certificate.

Post-process inspection catches bad parts. In-process monitoring tries to prevent them. The distinction matters on long cycle times, where a 4,000 mm part might be six hours into a cut before anyone measures it.

Spindle load and power monitoring is the most widely used signal. A dull tool raises cutting force, which raises spindle load. The controller sees the trend and either alarms or compensates with a feed override. It is a coarse signal, but it catches tool wear and broken tools before the scrap value climbs.

Touch probes on the machine measure the part on the fixture between operations. This is not a replacement for a CMM, because machine thermal drift and probe calibration add uncertainty. It is a fast way to confirm a datum before a critical finishing pass, and to catch a locating error before you cut the whole batch.

Laser tool setters measure tool length and diameter on the machine, which keeps the offset table honest. Combined with a warm-up cycle, this is often the difference between holding ±0.005 mm across a shift and drifting out of it by mid-afternoon.

• 1
  Spindle loadCatches tool wear and breakage from the cutting force trend.
• 2
  On-machine probingConfirms datums between operations; not a CMM substitute.
• 3
  Tool settingKeeps offsets current so tight tolerances hold across a shift.

Every method has a limit, and knowing the limit is part of the engineering. CMM measurement uncertainty grows with part size. On a 4,000 mm part, the uncertainty budget is dominated by thermal effects and by the machine geometry, not by the probe. A ±0.005 mm claim on a feature 3,000 mm away from the datum is not realistic without temperature control and a large-capacity CMM.

Optical and vision systems are fast and non-contact, which makes them ideal for soft materials, thin walls and large arrays of small features. They struggle with steep vertical walls, shiny surfaces and deep bores. If the feature is a blind hole with a 5:1 depth-to-diameter ratio, a vision system will not see the bottom. A touch probe or a bore gauge will.

CT is powerful but not free. Scan time, data processing and the interpretation skill required all scale with part size. On a 500 mm aluminium housing, a full CT scan can take longer than the machining cycle did. That is fine for a first article. It is not a production control method.

The practical rule: pick the cheapest method that has enough resolution to prove the tolerance, and use it at the frequency the risk justifies. A one-off prototype gets a full first-article inspection report. A mature 10,000-piece run gets a documented sampling plan plus 100% visual and functional checks. Both are legitimate. Pretending they are the same is not.

• 1
  Size limitCMM uncertainty grows with distance from datum; thermal control is essential.
• 2
  Vision limitsDeep bores, steep walls and shiny surfaces defeat optical systems.
• 3
  CT costScan time can exceed machining time on large parts.