Why CNC Hole Positions Shift Between Setups and How to Diagnose the Error

CNC hole positions can shift after reclamping when the part no longer seats against the same locating surfaces or the second setup establishes a different work offset. Before correcting the coordinates, compare each hole’s measured X and Y position against the drawing’s datum reference frame. The pattern of errors helps separate translation, rotation, and isolated hole defects.

In secondary machining operations, finish boring can expose an error that was absent at OP10. Check the coordinates before taking another cut.

1. Read the Hole-Pattern Error Before Adjusting Offsets

A hole diameter can pass inspection while the hole pattern is out of position. To diagnose the problem, keep the drawing’s basic dimensions fixed and compare the signed X and Y deviations for every hole.

At this stage, do not reset G54 or G55 to compensate for an unexplained error. First establish whether all holes moved together, whether their displacement changes across the pattern, or whether one hole behaves differently.

A Real EPOC CRAFTER Production Case: Six Holes Shifted After the Second Setup

In a real EPOC CRAFTER production and finished-part inspection case, a six-hole aluminum mounting plate developed a positional error after its second CNC setup.

Batch C03-01 contained 24 mounting plates used to connect controllers to fixed brackets in industrial automation equipment.

Actual production specification

Feature	Actual production specification
Material	6061-T6 aluminum, extruded plate
Part dimensions	120.00 × 80.00 × 12.00 mm
Hole pattern	Six holes, H1 through H6
Finished hole diameter	Ø6.00 +0.015/0.000 mm
Hole position requirement	Ø0.20 mm relative to A, B, and C
Datum A	Bottom mounting face
Datum B	Long locating edge
Datum C	Short end locating edge
Assembly function	Six M5 bolts connecting the mounting plate and bracket

The hole-position requirement had no material condition modifier. The six holes served a bolted mounting interface, not a precision dowel-pin locating system.

During OP10, the bottom face and locating edges were machined, and all six holes were predrilled to Ø5.40 mm. The part was positioned against A, B, and C, with G54 established from those locating features.

The recorded maximum calculated positional deviation of the OP10 preholes was 0.023 mm, within the specified Ø0.20 mm limit.

OP20 required reclamping the plate, finish-boring the six holes to Ø6.00 mm, and chamfering their entrances. The second setup used G55 and a different locating arrangement, including two pins against datum B.

The finished hole pattern showed a common shift toward positive X and negative Y, with a small angular component.

The recorded maximum calculated positional deviation increased to 0.294 mm, exceeding the Ø0.200 mm requirement.

1.1. Separate Translation, Rotation and One-Hole Errors

Compare the sign and magnitude of all six X/Y residuals before relying on a single reported position value.

Table 1 records the actual hole centers of one inspected part from batch C03-01.

Table 1. Actual hole-center measurements before and after reclamping (mm)

Hole	Basic X, Y	OP10 X, Y	OP20 X, Y	OP20 ΔX	OP20 ΔY
H1	25.000, 20.000	25.008, 19.994	25.123, 19.920	+0.123	−0.080
H2	55.000, 20.000	54.995, 20.008	55.120, 19.930	+0.120	−0.070
H3	85.000, 20.000	85.011, 20.004	85.124, 19.935	+0.124	−0.065
H4	25.000, 50.000	24.993, 49.991	25.113, 49.919	+0.113	−0.081
H5	55.000, 50.000	55.004, 50.006	55.116, 49.930	+0.116	−0.070
H6	85.000, 50.000	84.991, 50.007	85.112, 49.936	+0.112	−0.064

Here, ΔX and ΔY are the OP20 measured coordinates minus their corresponding basic coordinates.

All six X deviations are positive, and all six Y deviations are negative. This common direction places the setup location ahead of individual hole operations in the fault check.

A smaller change in direction appears across the pattern. The recorded setup deviation included approximately 0.011° of in-plane angular misalignment. Check coordinate system rotation along with the common X/Y shift.

The common shift directs the check toward seating, fixture location and G55. The angular component adds B/C alignment to that check. The data still need a physical contact test to establish the cause.

For this case, the specified two-dimensional true position calculation was:

P = 2 × √[(Xm − Xb)² + (Ym − Yb)²]

Here P is the diametric positional deviation; Xm and Ym are measured hole-center coordinates, and Xb and Yb are the basic coordinates.

This two-dimensional true position calculation uses the measured center and the basic center in the established datum frame. It suits the coordinate comparison documented in this case. For a drawing that controls the full axis of a through-hole, a single measured circle cannot by itself establish the entire axis position and orientation.

The reported maximum deviation was Ø0.294 mm against a Ø0.200 mm true position tolerance. The position requirement failed even though the process targeted the specified Ø6.00 mm finished hole size.

A different error pattern would require a different investigation. When only one hole shifts, examine that hole’s machining sequence, tool behavior, and measurement before changing the entire work offset. The same distinction matters for reamed holes: an acceptable diameter does not establish an acceptable hole location.

Inspect OP20 datum B contact, the clamp order and G55 before the next cut.

Table 2. How to prioritize the next diagnostic test

Residual pattern	First check	Decision boundary
Similar signed shift across all holes	Datum contact, OP20 work offset and fixture origin	Do not edit six hole coordinates individually
Shift varies across a row or column	Part rotation, B/C location and coordinate rotation	Prove the angular cause before changing an offset
One hole deviates while others agree	Local boring path, cutting condition and hole measurement	Check the affected feature before revising the fixture
CMM report changes with alignment	Specified datum simulation and measured feature construction	Do not best-fit the hole pattern to hide a common shift

2. Trace the Datum and CNC Workholding Across Setups

OP20 failed because the plate did not seat fully against its locating surfaces. In OP10, the plate rested on three points of A, contacted B at two points and C at one, and used G54. OP20 retained the three supports but located B using two pins and C against an end stop. Chips trapped at the B-side contact kept the plate from seating.

Two clamps were applied before the operator confirmed contact with B. That order amplified the mislocation. The G55 origin was established from an incomplete check of the locating surfaces, with a recorded setup displacement of approximately +0.118 mm in X and −0.072 mm in Y. These figures describe the actual setup investigation, not an allowable CNC tolerance.

The datum feature on the part must seat against its intended locating contact. The team checked the OP10 prehole centers, A and C contact, cutter runout and temperature before isolating the seating fault at B.

Table 3. OP20 failure and corrective action

Observed condition	Verified action
B-side locating pin contact obstructed by chips	Clean pins and the workpiece contact area; secure the end stop again
Clamps tightened before side seating was verified	Seat against B and C first, confirm both B contacts, then clamp symmetrically
G55 established without both B contacts confirmed	Probe actual A/B/C seating and rebuild G55 before boring
Common X/Y hole-pattern displacement	Recheck all six centers under the drawing datum reference system

2.1. Test Fixture Repeatability Before Changing G55

To check repeatability with CNC machining fixtures, release and reseat the same part using the same contact and clamp sequence, then compare the hole centers. In this case, the chip at B was identified in the locating contact. Changing G55 before cleaning would have hidden the fault.

Zero point clamping may reduce relocation work, but a supplier repeatability claim does not establish final hole-position conformity. A datum target applies only when the drawing specifies one. This case locates to the A, B and C features, not separately defined datum-target points.

The CNC work offset verification check then confirms that G55 matches the newly seated part. It cannot clean a locating pin or force the plate against the correct stop.

2.2. Do Not Confuse Datum Shift With a Work Offset Error

In GD&T, datum shift can describe permitted movement within a datum feature simulator under applicable material-boundary rules. That is not the same as changing G54 or G55 after a fixture or touch-off error. This drawing specified no material condition modifier on the six-hole position control. Do not use a supposed datum shift to excuse the measured displacement.

3. Rebuild the Drawing Datum Frame for CMM Inspection

A coordinate measuring machine (CMM) inspected the released part after thermal stabilization at 20 ± 1°C. The case records eight sampled points per hole section, or 48 hole points, plus nine points on A and six each on B and C. CMM probing established the reported two-dimensional hole centers using a 3-2-1 alignment to A|B|C. These center coordinates support the case evaluation; a full hole-axis requirement would also call for an appropriate axial sampling and fitting strategy.

Final CMM inspection must use the drawing datum reference frame, not a best-fit alignment of the six hole centers that artificially removes a common shift. The process coordinate system is useful for machining; it is not automatically the inspection reference. The applicable CNC drawing tolerance requirements govern how the part is evaluated.

After the locating contacts were cleaned and G55 rebuilt, the same coordinate basis gave the following finished hole centers. The real case record reports a maximum diametric deviation of 0.030 mm, below the Ø0.200 mm limit.

Table 4. Corrected hole centers and diametric positional deviation (mm)

Hole	Basic X, Y	Corrected X, Y	P, Ø mm
H1	25.000, 20.000	25.011, 19.990	0.030
H2	55.000, 20.000	54.993, 20.012	0.028
H3	85.000, 20.000	85.014, 20.005	0.030
H4	25.000, 50.000	24.989, 49.994	0.025
H5	55.000, 50.000	55.008, 50.010	0.026
H6	85.000, 50.000	84.988, 50.009	0.030

The six M5 bolts passed through the checked mating interface without interference. This does not establish an untested load-life capability. The 24-piece batch followed the revised operation, while the detailed coordinates here document one part.

3.1. Apply Position Callouts and Acceptance Conditions Correctly

Read the feature control frame before choosing a CMM acceptance rule. The case specifies a GD&T position tolerance of Ø0.20 mm relative to A|B|C, without a maximum material condition modifier. An in-size Ø6.00 +0.015/0.000 mm hole does not gain extra positional tolerance from its diameter alone.

With an applicable maximum material condition modifier, bonus tolerance may be available as hole size departs from the maximum-material boundary. No such modifier appears in this case, so the stated Ø0.200 mm position limit applies. Near the decision limit, follow the agreed inspection rule for measurement uncertainty; a 0.001 mm display increment cannot establish accuracy by itself.

4. Confirm the Machine Cause With Probing and Calibration

CNC probing can establish the actual part position before boring and reveal a missing locating contact. In this case, the original OP20 check covered A height and C but not both contacts against B. The revised sequence cleaned the locating pins, seated B and C before clamping, verified both side contacts and probed the part again before machining. Probe calibration and a suitable probing path matter because an uncalibrated probe can corrupt the work offset it is intended to correct.

The record includes checks of cutter runout and thermal stability. The investigation isolated the B-side locating contact as the cause of this event. Machine tool calibration is the next branch when the error persists after contact, probe and offset checks, or varies with machine position or rotary angle.

For parts with opposing-face requirements, multi-face CNC machining choices can reduce the number of transfers. Fewer setups do not guarantee better hole location when the physical datum contact is wrong. In-machine results and an independently established CMM datum frame should answer different questions, not be treated as interchangeable acceptance reports.

5. Specify Setup Transfer in the RFQ and First Article Inspection

An RFQ should identify which holes are finished in OP10 and OP20, the locating surfaces for each setup, the position requirement, and whether inspection takes place after unclamping. Request first article inspection for the entire six-hole pattern with datum references, measured center coordinates and the positional result. FAI inspection must identify the revision, conditions and measurements; a pass label alone is insufficient.

The revised sequence added 0.7 minute to per-part machining checks: 18.5 to 19.2 minutes. Inspection rose from 6.0 to 6.8 minutes after the B-side contact and datum checks were added. The recorded combined time increased from 24.5 to 26.0 minutes per part. The measured benefit was positional conformity and successful assembly. This case does not establish a saving in unit cost or lead time.

Supplier comparison should include the same setup method and inspection scope. Ask suppliers to confirm CNC machining inspection capability against the required A|B|C datum reference system; a quote that excludes second-setup verification is not equivalent to one that includes it.

Table 5. Disposition before releasing the next part

Decision	Required evidence	Action
Hold	Common hole-pattern error with unverified datum contact or CMM alignment	Stop the next cut; inspect seating, offsets and measurement basis
Revise	A specific locating, probing or process defect has been identified	Update and control the setup sequence, then remeasure the corrected part
Go	Corrected geometry and required inspection meet the controlled drawing	Release under the approved checks; continue monitoring the batch

Do not enlarge or relocate a functional hole to rescue a rejected part without design authorization. The broader CNC prototype design verification gate should settle the required datums, process route and acceptance evidence before repeat production.

6. Keep Model-Based Data and Inspection Results Traceable

Model based definition ties geometry to its approved manufacturing annotations. Digital inspection must retain the same revision, H1 through H6 IDs, basic coordinates, datum order and method from CAM through CMM. A report alignment change is not evidence of part movement.

6.1. Verify the PMI-to-Inspection Feature Mapping

PMI inspection maps controlled model features to measured results. Product manufacturing information must preserve feature ID, revision, datum reference and result. Digital records reveal a wrong revision; they cannot correct poor seating.

7. Inspection Questions From Shop-Floor and Metrology Discussions

7.1. How to Measure True Position After Reclamping

Align the released part to the specified drawing datums. Compare each hole center with its basic coordinates, calculate the diametric position error, and check modifiers in the feature control frame. In this case, the corrected maximum was Ø0.030 mm against Ø0.200 mm. Never best-fit the six holes to hide their common shift.

7.2. Why Does the CMM Show Zero Position Error When X Has Changed?

Check which features establish the alignment and which characteristic is being evaluated. A program may remove a translation by using the wrong reference or may report orientation relative to a datum when the user expects location. Rebuild the intended datum simulation and inspect the actual output components before accepting the displayed zero.

7.3. Can One Measured Circle Represent the Whole Hole Axis?

A single circular section gives a center at that section. It does not establish the axis of a tilted or form-affected through-hole. Where the drawing specifies an axial position zone, inspect and evaluate enough sections to construct the axis under the relevant drawing rules. The six-hole case presents a documented two-dimensional center-position comparison.