Composite Profile: Passing the Upper Segment and Failing the Lower
A die-cast gearbox cover has a curved sealing land that compresses a rubber gasket. The drawing uses a composite profile-of-a-surface frame: the upper segment is 0.8 to A|B|C and the lower segment is 0.3 to A. Datum A is the machined mounting face, B and C are the two Ø10 dowel holes, and the land is 6 mm wide and 240 mm long. CMM results, all evaluated in the same datum reference frame: Part 1 - form within 0.18 mm, the whole land displaced 0.35 mm in X, orientation to A good; Part 2 - form within 0.26 mm, displacement 0.05 mm; Part 3 - form within 0.10 mm, no displacement, but the land leans 0.32 mm relative to A over its length. The vendor's CMM software reports only the upper segment and passes all three parts.
1. Explain what each segment of the composite frame controls and what it does not. Why must the lower value be smaller than the upper?
Read the worked answer
Upper segment, 0.8 to A|B|C: a single profile zone 0.8 wide (0.4 each side of the true profile by default) that is fully located and oriented by the complete datum reference frame. It controls location in X and Y relative to the dowel holes, orientation to A, and form - all in one 0.8 budget.
Lower segment, 0.3 to A: a zone 0.3 wide that is oriented to A only. Its location relative to B and C is not controlled, and it may translate and rotate about A's normal to best contain the surface. It refines orientation to A and form. It says nothing about where the land sits relative to the dowels.
So the pair reads: 'put the land within 0.8 of where the datums say, and regardless of exactly where it lands, keep its shape and its squareness to the mounting face within 0.3.' That is a sensible gasket requirement - sealing depends on form and contact, not on X-Y placement.
Why lower < upper: the lower segment is a refinement sharing the same true profile. It must be a subset of the upper budget, otherwise it could never govern anything. Some drawing systems express the same intent as two single-segment feature control frames, one locating (A|B|C) and one refining (A); the interpretation is identical and both segments must be verified.
2. Evaluate Parts 1, 2 and 3 against both segments. Remember the profile zone is bilateral and total width equals the stated tolerance.
Read the worked answer
Upper budget: 0.8 total = 0.4 per side. A displacement d plus a form spread f gives a worst-case normal deviation of d + f/2.
Part 1 - d = 0.35, f = 0.18, worst case 0.35 + 0.09 = 0.44 > 0.40 → fails the upper segment. Lower segment: form 0.18 ≤ 0.30 and orientation to A good → passes lower. Net: reject. The part is in the right shape but in the wrong place.
Part 2 - d = 0.05, f = 0.26, worst case 0.05 + 0.13 = 0.18 ≤ 0.40 → passes upper. Lower: 0.26 ≤ 0.30 → passes, but only 0.04 of margin. Net: accept, with a note to the process - it is one small drift away from failing the lower segment while still looking fine to the upper.
Part 3 - d = 0, f = 0.10, lean 0.32 relative to A. Upper: 0.32 + 0.05 = 0.37 ≤ 0.40 → passes upper. Lower: the lean of 0.32 alone exceeds the 0.30 zone, because the lower zone is still oriented to A and cannot tilt with the surface → fails lower. Net: reject.
Note the two distinct failure modes: Part 1 fails on location, Part 3 fails on orientation. Neither can be diagnosed from a single number.
3. The vendor reports only the upper segment. What is the risk, and what should change in the inspection plan?
Read the worked answer
The risk: the upper segment is the permissive one. Because the lower segment is a refinement, any part that fails the lower can still pass the upper - Part 3 is a live example, and Part 2 shows a part drifting toward the same condition. Reporting only the upper segment is therefore a systematic false-accept bias, not a random error, and it will pass a whole population of parts whose sealing lands are skewed relative to the mounting face. The failure mode appears in service as a leak, not as a dimensional rejection, which is exactly the expensive kind.
What to change:
- Require a separate reported result for each segment. The software must evaluate the lower segment with the zone free to translate and rotate about A - a common programming error is to lock the lower-segment zone to the upper-segment datum reference frame, which makes the lower segment duplicate the upper and hides the very condition it exists to catch.
- Recognise that the lower segment is the function-critical one for a gasket: it is the check to put on the control plan and on the SPC chart, not the upper.
- Verify the orientation contribution explicitly. A 0.32 mm lean over 240 mm is only 0.0013 rad and easy to miss in a summary number, so report profile deviation as mapped colour rather than a single max value.
- If the two-segment frame is being converted to single-segment frames under a newer revision of the standard, keep both frames on the drawing and keep both in the inspection plan - the interpretation is unchanged.