Holes miss coordinates after flipping the part

What changes after flipping the part

When you flip a part, it ends up referenced in a different coordinate system. On the first setup you clamped it on certain surfaces; on the second setup you use different surfaces. So holes can be off not because of the toolpath or the cutter, but because of the new datum.

Don’t jump to complex causes. First look at how the holes shifted: along X, along Y, both axes, or as a rotation. That pattern usually shows you where to look.

If an entire row of holes shifted by the same amount, the most likely culprits are the second-setup zero or the base transfer between setups. If one hole is correct and another is off, the cause could be the program, the part geometry, or how the part was clamped after flipping.

For the first part it’s useful to record four things:

• which direction the shift went — X, Y or both
• the size of the shift in millimetres
• whether all holes shifted by the same amount
• whether the same dimension differs between the first and second sides

That is usually enough to separate a programming error from a fixturing error quickly. A program error gives a stable and repeatable pattern: the first and fifth parts shift the same way. A fixturing error behaves messier: one part shifts 0.03 mm, another 0.07 mm, and after re-clamping you get a different result.

What to compare first

You don’t need to measure every dimension at once. Pick one dimension that links the first and second sides — for example, the distance from the primary end face processed in the first setup to the axis of the hole that’s drilled after flipping.

If that dimension holds and dimensions inside the second side are also good, the program probably isn’t at fault. If the dimension between sides varies, look for an error in the transferred datum.

On a simple plate this is obvious. On the first side you machine a reference face and two holes. Then you flip the part and drill the matching holes. If every hole on the second side is shifted 0.05 mm in the same direction, the second-setup zero is usually off. If the shift varies, check the clamp, the contact surfaces and the fixturing scheme on the CNC.

The practical question is: where does this displacement still sit inside tolerance and where does it already cause rejection? A clearance hole may tolerate extra hundredths; a fit or threaded hole or paired holes for pins usually tolerate almost nothing. Once you separate those, it’s clear what to fix first: the program, the second-setup zero, or how the part is supported.

Where the error accumulates

When holes don’t land in the right coordinates after flipping, the shift rarely comes from a single source. It usually adds up from several small errors. Each one might seem acceptable, but together they produce a miss visible in inspection.

The first trap is the first-setup datum. If the zero was taken from a surface different from the drawing’s datum, the whole chain is already offset. On the screen everything may look correct, but in metal the part is referenced to the wrong geometry.

The tooling often adds error. Jaws, soft jaws, a V-block or stops must repeat the seating consistently. If one jaw is tightened harder, the V-block is worn, or the part contacts different points, a new effective zero appears even though the setup looks the same.

Chips and burrs are a simple but unpleasant source of shift. A thin spiral of chip under a support or a burr left after the first operation can lift or tilt the part. On a long part this quickly becomes a noticeable displacement of holes along an axis or between centers.

Clamping also pulls the part more often than you expect. This is very visible on thin plates, housings and long shafts with grooves. The operator places the part correctly, but tightening pulls it toward the jaw, rotates it in the V-block, or changes the angle by fractions of a degree.

Check the points where the probe finds zero. If on the first setup zero was taken from a clean surface, but on the second the probe touched a machined edge, a chamfer or a locally heated area, the coordinate system changes. The error may be small here, but it easily adds to fixturing errors.

Don’t forget temperature. After the first operation the part can be warm, especially after removing a large allowance or running without pauses. A warm part seats slightly differently in the fixture, and as it cools the size and hole position shift again.

Most often the error builds up from:

• the first-setup datum not matching the drawing datum
• the tooling not repeating the seating reliably
• chips or burrs under the support
• the clamp shifting or tilting the part
• zero being taken from different points than in the first setup

A short example. On a plate you drill two holes in the first setup, then flip and machine the reverse side. If a 0.03 mm burr remained under a support after flipping, the clamp pulls the part another 0.02 mm, and zero was taken from a chamfer instead of the plane, that can already be enough to cause scrap. So don’t look for one single cause — follow the whole base-transfer chain.

Order of base transfer between setups

If holes miss coordinates after flipping, the problem is often not the drill or the program. Usually the fixturing logic is off. On the first and second setups the part is referenced from different surfaces and the error accumulates step by step.

Start from the drawing. Choose the single primary datum from which hole coordinates are defined — it may be a plane, an axis, an end face or a machined center. That datum should control both setups. A convenient raw edge of a blank is only good for rough orientation. For accurate transfer it’s a poor reference: stock allowance, curvature and burrs immediately create drift.

On the first setup bind the zero to that datum or to geometry that unambiguously leads to it. If the drawing defines holes from the part axis and end A, don’t take some random outer edge just because it’s easier to reach with the probe. It’s fine at the start, but you lose your link to the drawing.

After the first operation the part should have clean, reliable supports for flipping — a machined end face, a plane, a boss or a bored reference. The idea is simple: the second setup must sit on surfaces processed in the first operation, not on the raw blank.

A typical working sequence looks like this:

1. Find the primary datum on the drawing and identify what the hole coordinates reference.
2. On the first setup create the datum following that logic, not from the most convenient edge.
3. In the first operation prepare the supports for flipping so the second setup can seat repeatably.
4. On the second setup place the part on those processed bases and transfer zero with a probe, gauge or a control dimension.
5. On the first part compare the calculated shift with the actual shift.

People often skip the last step, and that’s a mistake. For example, if the calculated zero shift after flipping is 126.40 mm but on the first part you get 126.32 mm, the 0.08 mm difference already points to play, tilt, chips under the support or a wrong chosen datum. Before the run this is easy to catch.

A correct base transfer is simple to recognise: remove and refit the part, then recheck zero — the deviation stays small and consistent. If it’s different every time, don’t touch the program. First check exactly how the part sits after flipping and whether that logic matches the drawing.

How to test the second setup on the first part

After flipping, don’t immediately run the whole drilling cycle. First confirm the new datum truly landed where you expected.

Start with basic dimensions you can measure before machining all holes. If the dimension is already off at this stage, the problem is in the base transfer, not the drill or tool offsets.

A practical sequence:

• flip the part and clamp it as you will for the run
• check that it seats on the base surfaces and check runout after final clamping
• make one control hole where the shift will be obvious
• immediately measure the distance from the datum to that hole
• compare the measured value with the coordinate shown at the CNC zero

Check runout after the final clamp, not before. Often the part sits flat on a soft clamp, but moves 0.03–0.08 mm when the operator tightens the clamp or changes a stop. For the second setup that’s enough to make holes miss coordinates, especially when holes pair with features on the other side.

One control hole is often more useful than running a full cycle and producing many bad parts. It quickly shows whether the part moved in X, Y, height or rotated. If the hole shifts the same way on several first parts, the cause is not the tool but a repeatable shift in the tooling, stop or clamping order.

Compare with the CNC display without guessing. The machine shows the center it thinks it will cut; the part shows where the hole actually landed. That difference is the error, which you then tie to a concrete cause: stop, V-block, jaws, pin, fixture plate or clamp.

Record not only the correction size but the condition when it appeared. For example: “after flipping, clamped from the right stop, X shifts +0.04 mm.” Such a record tells more than just “added an offset.” After a few parts you’ll know whether the error is random or repeats the same way.

If the shift repeats, don’t treat it only by correcting the zero. A correction helps release the first batch, but it doesn’t explain why the part keeps moving the same way. Fix the root cause first.

Example on a simple part

Take a simple housing block: a rectangular bar with two datum faces and a row of four holes on the second side. On such parts you can quickly see why holes miss coordinates after flipping.

If the pitch between holes is correct but the whole row moved, the error is almost always in the base transfer between setups, not the drill or the program.

On the first setup don’t try to close every dimension at once. First machine one supporting plane, then two end faces that will become the lateral and longitudinal datums. After that the part has a clear set of surfaces from which to build the second setup. If after the first operation only one clean plane remains and the second stop is taken from a raw edge, drift is almost inevitable.

On the second setup seat the part on the machined bases. Use the 3-2-1 principle: plane, lateral base, end face. A raw edge is poor because it can have scale, burrs, uneven casting or variable stock. Even 0.05–0.10 mm on such an edge easily becomes a track-out of the whole row of holes.

Work sequence:

1. On the first setup machine plane A and ends B and C.
2. Check that bases are free of chips and burrs.
3. On the second setup place the part on A, clamp to B and stop on C.
4. Only after that drill or bore the row of holes.
5. Immediately measure one control dimension from base B to the center of the first hole.

A strong side clamp often spoils the picture. The part may look seated correctly, but tightening pulls it along the contact. On an indicator this may be nearly invisible while the row of holes shows a stable offset. This often happens on low housings with small contact areas where the clamp is overtightened.

One control size usually points to the cause fastest. For example, if the distance from face B to the center of the first hole should be 40.00 mm but you measure 40.12 mm, and all other holes are shifted by ~0.12 mm, the fault is the base transfer or clamp-induced shift. If the first dimension is correct but the error builds across the row, then check the tool, thermal growth or toolpath.

What most often moves holes off

In practice the cause is often simple: after flipping the part rests on a different surface — the machine goes to the coordinates, but the part sits differently. That’s why holes miss coordinates even with a correct program and proper tool.

The datum must transfer between setups by the same rule. If on the first setup you used a machined plane and end face, but on the second you rely on another end face just because it’s easier, the shift is already introduced. Sometimes small, but visible across several holes.

Another frequent mistake is taking zero from an edge with a burr, chip or mark from previous operations. A probe or indicator will record a point, but it’s not the part geometry — it’s a random defect. A single burr can give tens of microns, which is enough to fail a tight tolerance.

Jaws and soft jaws also confuse people. It seems that if the part fitted once it will seat the same way again, but seating changes due to chips, clamp force, wear, slight misalignment or different stop depth. The repeatability of a chuck doesn’t remove the need to check the actual part position.

Confusing the program datum and the fixture datum happens all the time. For example, CAM zero can be at the part center while the operator references from a fixture stop and inputs offsets by chaining dimensions. It works until the first sign or direction mistake.

A less obvious cause is a small rotation of the part. Linear displacement may be nearly invisible, but on a 100 mm arm a 0.1° rotation gives about 0.17 mm shift. For holes on a circle this is clearly scrap.

Usually the shift is a combination of factors:

• the part rests on a different surface after flipping
• zero was taken from a defective edge
• clamping repeated the seating worse than expected
• program coordinates and fixture coordinates were mixed
• a first successful correction was found but not recorded or locked for the series

People often underestimate that last point. If you found the real shift on the first part but didn’t document it by axes and didn’t fix the working steps, the next part will “float” again. From the outside it looks random, while the actual reason is the same: no rigid rule for base transfer between setups.

Quick checklist before a production run

Before starting a run spend a few minutes on routine checks. This often decides whether the second setup will hit size or holes will miss coordinates from the first batch.

The most boring mistake is also the most dangerous: chips left between the part and supports, a burr on a datum edge, or the operator tightening the clamp a little more or less than during setup. One part may not show it, but in a series the shift becomes obvious quickly.

A short pre-run routine:

• wipe supports, jaws, clamps and the part
• remove burrs from datum surfaces
• clamp the part with the same force and in the same sequence used during setup
• confirm zero by the same method every time
• after the first part check two coordinates (not just one) to catch linear shift and rotation

Practically: you flip the part, take zero, and X may be nearly in tolerance. Many stop there. But if you check a second coordinate — Y or a distance to another datum — you can catch a rotation of a few hundredths in time. Otherwise the drift becomes serial.

Watch the repeatability of the clamp. If on the first part you used a short lever and later started tightening harder, a soft or long part changes position. On both turning and milling the effect is the same: the datum chain looks identical but the part actually sits differently.

If the shift grows after the first and second parts, stop the run immediately. Don’t wait for the size to “settle.” Recheck cleanliness of bases, condition of jaws and stops, the second-setup zero and the clamping sequence. Five minutes of stop are usually cheaper than ten parts scrapped.

What to do next on your shop floor

If holes miss coordinates after flipping, don’t start with guesses. First lock a single consistent work procedure so the error stops wandering between shifts.

Start with a simple base-transfer diagram for both setups. Often a single sheet showing where the part contacts, how it’s clamped, which surface defines zero and which dimensions transfer to the second setup is far more useful than verbal explanations.

The setup technician, operator and inspector should share the same logic. If the paper says one datum but the fixture makes the part stop elsewhere, a shift is almost certain.

Rules to make standard

On the shop floor it helps to adopt a few steady rules:

• always measure the first part in the same sequence
• record which surfaces were used to take dimensions
• keep a log of actual shifts by batch and material
• note which fixture and chuck were used for the run
• record how zero was established: probe, indicator or touch-off

This is not needless paperwork. After two or three batches the records often reveal patterns better than memory. For example, one material holds size but another consistently shows 0.03–0.05 mm drift on one axis after flipping.

Another useful step is to formalise one inspection route for the first part: check bases and the second-setup reference, then inter-center dimensions, and only then start the run. If you change the inspection order you may catch the symptom rather than the cause.

When to look at hardware, not the program

If the shift repeats, don’t only hunt in offsets. Inspect the machine, the tooling and the probe as a single chain. Small play in the fixture, a tilted clamp or a probe error combined can create the drift that later looks mysterious.

A simple teardown helps: mount one part, recheck the base plane, repeat the probe touch, verify clamping repeatability and only then touch the program. Very often the cause is found before examining the code.

If you lack time or experience for that analysis, call service specialists. For setup, commissioning, tooling selection and machine checks in Kazakhstan and the CIS, many turn to EAST CNC. The company offers not only supply of lathes and machining centers but practical service — from selection to start-up and maintenance. For recurring flip-related problems this often helps more than endlessly tweaking the zero manually.

A correct result looks simple: a base-transfer diagram, a single first-part inspection routine and a log of actual shifts. Then the cause is found quickly, not after a dozen scrapped parts.