Dimensional Inspection of Large Frames: Laser Tracker vs Large CMM

A tolerance you cannot agree on measuring is not a tolerance. On a 200 mm bracket that is rarely a problem. On a 6 m machine base it is the central problem, because the instrument, the support condition and the temperature all change the number you get.

The instruments

Laser tracker

A servo-driven head tracks a retroreflector (an SMR) moved over the part by an operator, measuring distance interferometrically and two angles encoder-wise.

• Accuracy: typically in the tens of microns over several metres; manufacturers quote something like 15 µm + 6 µm/m, though real-world accuracy on a shop floor is worse
• Envelope: effectively unlimited, the tracker can be relocated and the measurements bundled
• Speed: fast for surfaces and features; a full frame in hours
• Limits: needs line of sight; angular error grows with distance; the operator’s technique matters

This is the practical default for large frames. It goes to the part rather than the part going to it, which on a 15-tonne weldment is the deciding factor.

Large-envelope CMM

A bridge or gantry coordinate measuring machine in a controlled room.

• Accuracy: the best available, single-digit microns on a good machine
• Envelope: limited by the machine, and large ones are rare and expensive
• Speed: slower, but automated and repeatable
• Limits: the part has to be transported to the machine, lifted onto the table, and allowed to reach temperature

Use it where the tolerance genuinely demands it, and accept the logistics.

Articulated arm

A portable multi-axis arm with a probe or laser scanner.

• Accuracy: tens of microns within its reach
• Envelope: typically 2–4 m, extendable by leapfrogging between reference points, which compounds error
• Use it for: medium parts and local features on large ones

Optical / photogrammetry and structured light

Camera-based, fast over large areas, good for form and deviation maps rather than precise point-to-point dimensions. Increasingly used alongside a tracker to produce a colour deviation map of a whole surface, persuasive in a report, and genuinely useful for showing where a surface is out rather than just by how much.

Traditional methods

Precision level and sweep, straightedge and feeler, optical level, tensioned wire. Slower, cheaper, and still entirely valid for many checks. A precision level on a machined way is a perfectly good flatness measurement and needs no calibration certificate from a metrology lab.

Do not let anyone tell you a frame cannot be inspected because there is no tracker. It can. It just takes longer.

The three things that change the answer

1. Support condition

A 4 m frame clamped rigidly to a machine table will measure flat and spring when released. The same frame on three-point support measures its free shape.

Specify the support. Usually: three-point kinematic support at defined locations, ideally the same points it will sit on in service. Write them on the drawing.

2. Temperature

Steel grows about 12 µm per metre per °C. On a 5 m frame, 10 °C is 0.6 mm of length, larger than most tolerances on this page.

For anything tighter than about 0.05 mm/m, specify:

• Reference temperature, conventionally 20 °C
• A soak period before measurement, typically 12–24 hours for a heavy section
• Whether readings are corrected to 20 °C, and using what coefficient

Without this, you and the shop can measure the same frame on the same day and disagree, and both be right.

3. Datum strategy

Large weldments need datums that actually exist and can be found repeatably. Three principles:

• Use machined features as datums: not as-welded surfaces
• Separate the datums physically: a datum pair 100 mm apart on a 5 m part amplifies angular error enormously
• Mark the datums on the part: stamped or engraved, so the same features are used at the shop, at incoming inspection and at site

An ambiguous datum scheme is the most common cause of a frame that passes at the shop and fails at your plant.

What an FAI report should contain

• Part identification, drawing number and revision
• Instrument used, with serial number and calibration certificate date
• Ambient temperature at measurement, and whether readings are corrected
• The support condition used
• Datum features and how they were established
• Every dimension and tolerance from the drawing, with the measured value and pass/fail
• Out-of-tolerance items flagged, with disposition, repair, rework, or concession request
• Name and signature of the inspector, with qualification
• Photographs of the setup

That last item matters more than it sounds. A photograph of the frame on its support with the tracker in position answers half the questions that would otherwise be raised by email.

Measurement uncertainty

If a tolerance is 0.05 mm and the measurement uncertainty is 0.03 mm, a reading of 0.04 mm does not tell you the part is in tolerance.

The usual convention is that measurement uncertainty should be no more than about a tenth of the tolerance. On large frames with tight tolerances that is not always achievable, in which case the sensible move is to agree a guard band, accept at 0.04 mm on a 0.05 mm tolerance, rather than argue about it later.

Raise this at RFQ, not at inspection.

Where inspection sits in the sequence

1. Fabrication dimensional check, as-welded, before stress relief
2. Post-stress-relief check, has it moved? See stress relief for welded machine bases
3. Post-rough-machining check
4. Final FAI after finish machining
5. Post-shipment check at your site, on your foundation

Step 5 is worth doing on the first piece from a new supplier, because it establishes whether shipping and re-support change the geometry. If it does not, later pieces can be accepted on the shop FAI alone.

Who does the measuring

Three options, in ascending cost and confidence:

• The fabricator’s own inspector. Fine once a relationship is established and the reports have been proven right.
• A third-party inspection agency. A fixed fee per visit. Standard practice on first orders and on anything critical.
• Your own engineer travelling to the shop. Expensive, and occasionally the right answer on a project where the frame is the schedule.

A matched shop can quote the second as a line item, and we will say when we think it is worth the money, which is usually the first order with any shop, and rarely the fourth.

What to write on the drawing

“Final dimensional inspection by laser tracker, part on three-point support at locations shown, soaked to 20 ± 2 °C for minimum 12 hours, all readings corrected to 20 °C. Full FAI report required with instrument calibration certificate.”

That sentence removes most of the arguments that happen at acceptance.

Have a tolerance on a large frame and no confidence it can be verified? Send the drawing and we will tell you how it would be measured and what that adds, within 24 business hours.