Metrology Guide

Measuring CNC Machined Housing Parts: A Practical Method

Housings are the hardest parts to measure. Bore-to-bore distance, flatness, wall thickness and angular faces all reference each other, so a single caliper reading tells you almost nothing. This guide is for engineers and inspectors who need a repeatable method, not a catalog of instruments. Read it and you can pick the right setup for a prototype, a pilot run or a production lot.

Get Instant Quote Talk to an Engineer

±0.005 mm capability100% inspectionCMM reports on requestDFM in 12 hours

Measuring CNC machined housing parts on a five-axis machined engine housing

On this page

1
Key takeaways
2
Why measuring CNC machined housing parts is harder than it looks
3
Datum setup: the step that decides everything
4
Choosing a method for each feature type
5
Temperature, clamping and probe force: the hidden error sources
6
What to put in the inspection report
7
Step by step: how to measure a machined housing
8
Which measurement method fits which feature

Quick answer

Key takeaways

Fix the datum firstA housing measured from three different datums gives three different answers. Lock the datum scheme before you touch a probe.

Match the tool to the toleranceCalipers for ±0.05 mm, micrometers and bore gauges for ±0.01 mm, CMM for ±0.005 mm and form.

Watch the temperatureAluminum grows about 23 μm per meter per °C. A 5 °C shop swing moves a 300 mm housing by 0.035 mm.

Report form, not just sizeRoundness, flatness and perpendicularity drive assembly fit more often than nominal diameter does.

Measure in the clamped state you shipA housing that reads true on the bench can spring 0.02 mm once bolted to a fixture.

Section 1

Why measuring CNC machined housing parts is harder than it looks

A housing is a network of features that reference each other. A bearing bore sits at a defined distance from a mounting face. That face sits at a defined angle to a second face. Wall thickness ties the whole thing together. Change one reference and every derived dimension shifts with it.

That is why two inspectors can measure the same housing and disagree by 0.03 mm without either of them making a mistake. They picked different datums, different probe tips, or different clamping. The part did not change. The measurement setup did.

The fix is not a better instrument. It is a written method: which datum, which probe, which clamping force, which temperature. Once that method is fixed, the numbers become comparable across shifts, machines and suppliers.

For housings machined to ±0.005 mm, measurement uncertainty should stay under 20% of the tolerance band. That means your method needs a realistic uncertainty near ±0.001 mm. Few hand tools reach that. Plan the method before you plan the inspection queue.

Section 2

Datum setup: the step that decides everything

On the drawing, the datum callouts already tell you the answer. The primary datum is usually the largest mounting face or the main bore axis. The secondary is the next most stable feature. The tertiary locks rotation. If the drawing is ambiguous, ask the designer before you measure, not after.

On the machine or CMM, establish a datum reference frame from those same features. For a housing with a large flat base, level the base on three points and probe at least six to eight points to build the plane. Three points define a plane mathematically, but they also hide waviness. More points give you a real flatness number.

For a bore-based datum, measure the bore at two heights and fit an axis. A single circle at one height captures diameter but not axis tilt. If the bore is 40 mm deep, two circles 30 mm apart will expose a tilt that one circle misses entirely.

Common error: using a machined surface as datum because it is convenient to reach. If the drawing calls a cast pad as primary datum, measure from the pad. Convenience datums create false rejections and false acceptances in equal measure.

1
Primary datumLargest stable face or main bore axis; probe 6–8 points minimum
2
Secondary datumNext most repeatable feature; usually a side face or second bore
3
Tertiary datumLocks rotation; often a slot, pin hole or chamfer edge

Section 3

Choosing a method for each feature type

Not every feature deserves the same instrument. Spending CMM time on a clearance hole is waste. Measuring a bearing bore with calipers is risk. Sort the features by tolerance band first.

Wide-tolerance features, typically ±0.1 mm or looser, are fine with calipers, height gauges or pin gauges. These are clearance holes, chamfers, outside profiles and non-critical slots. Record them, but do not let them slow the queue.

Mid-tolerance features, roughly ±0.01 to ±0.05 mm, need micrometers, bore gauges, depth micrometers or a height gauge on a granite plate. This covers most mounting hole positions, step depths and shoulder diameters. Use a bore gauge with a setting ring, not a telescoping gauge alone.

Tight-tolerance and form features, ±0.005 mm and below, belong on a CMM or a dedicated form tester. This includes bearing bores, seal faces, perpendicularity, coaxiality, flatness and any true position callout. Optical systems help when the feature is small, thin-walled or too soft to touch.

Wall thickness on thin housings is a special case. Ultrasonic thickness gauges read without contact force, which matters on 1.5 mm walls where a micrometer would deflect the part. Pair the ultrasonic reading with a CMM scan of the outer profile when both surfaces matter.

Section 4

Temperature, clamping and probe force: the hidden error sources

Aluminum expands about 23 μm per meter per °C. Steel is closer to 11 μm. A 300 mm aluminum housing measured at 25 °C but inspected against a 20 °C drawing is already 0.035 mm off before any instrument error. In a shop that swings 5 °C between morning and afternoon, that alone can consume a ±0.05 mm band.

Let parts soak. A housing straight off the machine carries cutting heat. Twenty minutes on a granite plate in still air is a reasonable minimum for a 300 mm part. For tight work, soak longer and log the part temperature alongside the reading.

Clamping force is the next silent error. A housing held in a vise at 2 kN will deform. Release it and the bore springs back. If the part is assembled under bolt load, measure it under the same load, or measure it free and note the free-state number.

Probe force matters on thin sections. A touch-trigger CMM probe deflects 0.3–0.5 mm before it triggers, and the stylus bends slightly. On a 2 mm wall this can push the reading. Use a scanning head with low measuring force, or switch to optical for those features.

Finally, calibrate against a known artifact before each session. A setting ring or gauge block that reads 0.002 mm off will bias every subsequent number in the same direction.

Section 5

What to put in the inspection report

A report that lists only nominal and actual is half useful. The engineer reading it needs to know the method, the datum, the temperature and the uncertainty. Without those four items, the number cannot be compared to anything.

For each critical feature, report the measured value, the tolerance, the instrument used and the datum reference frame. Where a form callout applies, include the fitted value: flatness, roundness, perpendicularity or true position. These are the numbers that predict assembly success.

If a feature is out of tolerance, report the deviation direction and the amount. A bore that is 0.008 mm oversize is actionable. A bore marked simply 'fail' is not. Direction tells the machinist which way to adjust the offset.

Attach the raw point data when the customer asks. On a first article, a full CMM point cloud with the alignment report is standard practice. On production lots, a summary table with a few representative values is usually enough.

Procedure

Step by step: how to measure a machined housing

Follow this order. Skipping a step is the most common cause of a rejected first article.

1
1. Read the drawing and list critical featuresSort every dimension into three bands: loose (±0.1 mm or wider), mid (±0.01–0.05 mm) and tight (≤±0.005 mm). Mark which features carry a form or position callout. This list decides your instrument plan.
2
2. Clean and deburr the partWipe with lint-free cloth and solvent. Burrs at bore edges read as size error and can deflect a probe. A 0.02 mm burr on a 10 mm bore looks like 0.04 mm of diameter error.
3
3. Soak the part to room temperatureMinimum 20 minutes on a granite plate for a 300 mm housing. Log the part temperature. Do not measure a part that is still warm from the machine.
4
4. Establish the datum reference frameProbe 6–8 points on the primary face, 4–6 on the secondary, 2–3 on the tertiary. Fit the plane and axis in the software before measuring any feature dimension.
5
5. Measure tight features firstDo bearing bores, seal faces and position callouts while the setup is fresh. Use a scanning head with low force on thin walls. Record diameter at two heights to catch axis tilt.
6
6. Then measure mid-band featuresUse a bore gauge set against a ring, or a micrometer on a granite plate. Take three readings 120° apart on each bore and average. Note the spread; a spread over 0.01 mm means the bore is not round.
7
7. Finish with loose featuresCalipers and pin gauges are fine here. Keep them fast. Do not let clearance holes block the queue behind tight work.
8
8. Verify before you releaseRe-measure one tight feature at the end of the session. If it moved more than 0.002 mm, your setup drifted. Recheck temperature and clamping, then re-run the alignment.

Method selection

Which measurement method fits which feature

Pick by tolerance band and feature geometry. Uncertainty is the practical floor for a trained operator in a temperature-controlled shop.

Method	Typical uncertainty	Best for	Avoid when
Calipers, height gauge	±0.02–0.05 mm	Clearance holes, profiles, loose steps	Any tolerance under ±0.05 mm
Micrometer, bore gauge	±0.002–0.005 mm	Bore diameters, wall thickness, step depth	Thin walls that deflect under contact
CMM touch probe	±0.001–0.003 mm	Position, form, coaxiality, tight bores	Soft or thin parts that flex
CMM scanning head	±0.001 mm	Freeform surfaces, thin walls, full profiles	Simple prismatic parts, cost not justified
Optical / vision system	±0.002 mm	Small features, edges, soft materials	Deep bores the camera cannot see
Ultrasonic thickness gauge	±0.01 mm	Thin walls, closed sections, no contact	Curved surfaces under 10 mm radius

The method matters more than the machine

A ±0.005 mm CMM with a sloppy datum scheme produces worse data than calipers used against a written method. Fix the datum, soak the part, log the temperature, and pick the instrument by tolerance band. That sequence decides whether your numbers mean anything.

Get Instant Quote Talk to an Engineer

FAQs

Frequently asked questions

Can I measure a housing with calipers alone?

For clearance holes and overall length, yes. For anything called out at ±0.05 mm or tighter, no. Caliper jaw pressure varies between operators, and the reading changes with how hard you squeeze.

A practical rule: if the tolerance band is wider than 0.1 mm, calipers are fine. Below that, move to a micrometer, bore gauge or CMM.

How many points do I need for a reliable flatness reading?

Six to eight points on a mounting face is a reasonable minimum for a housing. Three points define a plane but hide waviness, so a three-point reading will look flat even when the face is bowed.

On a large face over 200 mm across, probe 12 points in a grid. That catches both bow and twist, which a few points along one line will miss.

Does part temperature really matter at ±0.005 mm?

Yes. Aluminum moves about 23 μm per meter per °C. A 300 mm housing measured 5 °C above the reference temperature reads 0.035 mm large. That is seven times the tolerance band.

Soak parts on a granite plate in still air for at least 20 minutes, log the temperature, and if your shop swings more than 3 °C, correct the reading mathematically.

What is the difference between free-state and clamped-state measurement?

Free-state means the part rests on its own, unclamped. Clamped-state means it is bolted or fixtured the way it will be in service. Thin housings can differ by 0.02 mm or more between the two.

Measure both when the housing is thin-walled. Report which state each number refers to, otherwise the assembly team will see a mismatch they cannot explain.

When should I switch from a touch probe to optical measurement?

Switch when the feature is too small, too thin or too soft for contact. A 0.5 mm wide slot or a 1.5 mm wall is a good candidate. The camera reads the edge without touching it.

Stay with touch probing for deep bores and any feature where the line of sight is blocked. Optical systems cannot see into a 60 mm deep bore.

How often should I re-calibrate the CMM during a job?

Check against a setting ring or gauge block at the start of each session, and again after any probe change. A probe swapped mid-job without re-qualification is a common source of a 0.003 mm bias.

For long production runs, re-check every four hours or after 50 parts, whichever comes first.