
The part moves before the cutter does
A thermoformed part is a shell. It has no closed section, no ribs to speak of, and a wall of 0.8 to 2 mm. When a clamp closes on it, or a cutter pushes into it, the shell deflects, and the cut lands wherever the deflected edge happens to be.
The numbers are not subtle. Under a 200 N clamp a 0.8 mm polypropylene wall deflects 1.80 mm, a 1.2 mm wall 0.80, a 1.6 mm wall 0.45, a 2.0 mm wall 0.28 and a 3.0 mm wall 0.09. On a part with a plus or minus 0.15 mm trim tolerance, the 0.8 mm part is out by twelve times the tolerance before the cutter is even switched on.
That is why a thin part is not really cut, it is pushed. The cut edge follows the deflection, and the deflection depends on the clamp force, the clamp position and the stiffness of the shell at that point. Two parts from the same tool, clamped in slightly different places, trim to different sizes.
The practical consequence is that the fixture has to do the stiffening. A nest that supports the wall directly under the cut line turns a 0.8 mm shell into something that behaves like a 3 mm one, because the wall can no longer bend where it is supported.
Where the error actually comes from
When a trimming cell is audited by measuring the same part repeatedly, the error budget is consistent and it is not what most shops assume. Fixture location accounts for about 45 percent, part-to-part variation 30, the machine 15 and tool wear 10.
Fixture location is first because it is the largest single contributor and because it is the only one that can be zero. A part that is located on a formed feature and supported under the cut has almost no location error; a part that is located on its own trimmed edge has the tolerance of the previous operation added to this one.
Part variation is second and it is a thermoforming problem, not a trimming problem. A part that varies by 0.5 mm from the mould will trim to a size that varies by 0.5 mm, whatever the fixture does, unless the fixture locates on a feature that is formed accurately and measures from there.
The machine being only 15 percent is the reason a new router rarely fixes a tolerance problem. It is also the reason a shop that buys a better machine and keeps the same fixture sees a small improvement and concludes that the tolerance is impossible.
| Kaynak | Share | What it means | Can it be removed |
|---|---|---|---|
| Fixture location | 45 percent | The part is not where the program thinks | Yes, with a nest |
| Part variation | 30 percent | The formed part is not repeatable | Partly, by locating on a formed feature |
| Machine | 15 percent | Backlash, thermal growth, calibration | Yes, with maintenance |
| Tool wear | 10 percent | Cutter diameter changes with life | Yes, with tool life management |
Table 1 — The trim error budget, and how much of it is the fixture.
Vacuum is not a fixture
A vacuum table holds a part flat. It does not locate it. The part can slide on the table until the vacuum is applied, and it can be pulled a millimetre or two into the vacuum holes as the pressure drops, which changes both its shape and its position.
On a part with a large flat face and a 2 mm wall, vacuum is a good holding method and a poor locating method. The answer is a nest that locates on a formed feature, with vacuum used only to hold the part down onto the nest. Location and holding are two different functions and they need two different features.
On a thin part the vacuum also does structural damage. A 0.8 mm wall over a 5 mm vacuum hole deflects into the hole, and the deflection is permanent if the part is still warm. That shows up as a dimple pattern on the visible face, spaced exactly like the vacuum holes, and it is a reject that the trimming cell created.
Designing the nest
A nest is a machined support that matches the outside of the part, and its job is to do three things at once: locate the part repeatably, support the wall under the cut, and leave the trim path clear.
The location should be a formed feature rather than an edge. A moulded step, a pair of bosses, a rib or a corner radius are all better than the trim edge, because the trim edge is the output of the operation and its tolerance is what you are trying to control.
The support should extend under the cut line by at least 10 mm on both sides. A nest that supports only the inside of the cut leaves the outside unsupported, and the wall peels away from the cutter on the last few millimetres, which produces a tear rather than a cut.
The clearance has to allow for the part, not for the nominal part. A nest machined to the nominal profile will grip a part at the top of the tolerance and rattle one at the bottom. Machining to the middle of the tolerance and adding a 0.1 mm shim at the location feature is a common way to make one nest cover the whole tolerance band.
Cutter, path and the last millimetre
A single-flute cutter is the standard for thermoplastic sheet. Two flutes clear chips faster but generate more heat, and heat on a 0.8 mm wall softens it and pulls it into the cutter. The single flute also leaves a cleaner edge on PVC and polypropylene because the chip load per revolution is higher.
| What you see at the edge | Usual cause | What actually fixes it |
|---|---|---|
| White stress line along the cut | Two-flute cutter at too high an rpm | Single flute, halve the spindle speed |
| Stringy chips welded to the edge | Melt, not a cutting problem | Air blast and a slower feed per tooth |
| Cut drifts oversize on one side | Wall pushed away before the flutes engage | Ramp entry or lead-in from outside the part |
| Tear-out where the cutter exits | Nothing behind the wall at breakthrough | Leave tabs, cut the last 0.3 mm in a second pass |
| Part lifts mid-contour | Vacuum lost at breakthrough | Tabs plus a mechanical clamp, not more vacuum |
| Burr grows through a run | Edge wear past about 30 percent | Change the cutter on a count, not on feel |
Table 2 — Edge defects on trimmed thermoplastic sheet, and the lever that removes them
Climb milling, where the cutter rotation opposes the feed at the entry, gives a cleaner edge on the outside of the part and a worse one on the inside. Conventional milling does the opposite. On a part where one edge is visible and the other is hidden, the direction is a decision rather than a default.
The last millimetre is where most trimming defects happen. A cutter that exits through the wall without support tears the edge, and a part that is held only by vacuum will lift as the cutter breaks through and loses the vacuum. Tabs, small uncut bridges left at intervals, hold the part until the cut is complete and are removed by hand or in a second pass.
Entry matters as much. Plunging straight into a 1 mm wall pushes it away before the cut starts. A ramp entry, or a lead-in from outside the part, cuts the wall rather than pushing it, and it costs nothing but a line of code.
What to specify
A trimmed part is specified by the trim line and the datum, and the process is specified by what it can hold.
- Trim line dimensions referenced to a formed datum, not to an edge.
- The datum feature, and its tolerance in the forming operation.
- Whether the part is trimmed in a nest, on a vacuum table or freehand.
- Wall thickness at the trim line, which sets the deflection and therefore the achievable tolerance.
- Edge quality and burr limit, and which edge is visible.
- Tab positions, if the part needs them to stay held during the cut.
The wall thickness at the trim line is the item that predicts whether the tolerance is achievable. A drawing that specifies plus or minus 0.15 mm on a 0.8 mm wall is asking for something the process cannot do, and the answer is a nest, not a tighter tolerance.
Why is CNC trim tolerance so hard on thin parts?
Because the part deflects under clamp and cutter force. A 200 N clamp moves a 0.8 mm wall by 1.8 mm, which is far more than the tolerance, so the cut follows the deflection rather than the program.
What is the biggest source of trim error?
Fixture location, at about 45 percent. Part variation is 30, the machine 15 and tool wear 10.
Can a vacuum table locate a part?
No. It holds a part flat but does not locate it, and on a thin wall it pulls the material into the vacuum holes. Use a nest to locate and vacuum only to hold down.
Where should a nest locate?
On a formed feature such as a step, a boss or a rib, not on the trim edge. Locating on the edge stacks the previous operation's tolerance onto this one.
Why does the cutter tear the edge at the end of a cut?
Because the wall is unsupported as the cutter exits, and the part lifts when the vacuum is lost. Tabs left at intervals hold the part until the cut is complete.
Does climb or conventional milling matter?
Yes. Climb milling gives a cleaner edge on the outside of the part and a worse one on the inside; conventional does the opposite. Choose by which edge is visible.
Trim tolerance that will not hold?
Send the part drawing with the wall thickness at the trim line and the fixture concept. We will tell you what the process can hold and whether the fixture is the limit.
Continue reading
- Trimming Thermoformed Parts: Steel Rule, Router and Punch Compared
- Demoulding: Draft, Ejection and the Part That Sticks at Temperature
- Thermoforming Tolerances: What a Formed Part Can Hold
- Stacking and Nesting: Geometry That Ships Without Damage
- Custom Vacuum Forming: Tooling, Materials and Tolerances











