Fix Plasma Cuts Not Closing Loops: Pierce Delay Guide

Tom Garrett9 min read
Motion ControlOther ManufacturerTroubleshooting
Licensed PE Working through this on a live machine? A Maine-licensed engineer can take it from here — included with IMD hardware, by the hour for everything else. Book an engineer

Problem Details

A CNC plasma table runs a closed profile (circle, slot, rectangular window, or any loop whose end point coincides with its start point) and the cut is incomplete. The typical signature is a short uncut tab at or near the start of the loop, so the slug stays attached and the part will not drop. Straight open cuts and the outer perimeter often look fine, which pushes operators toward blaming the drawing.

Two independent failure families produce the same visible symptom, and they must be separated before you change anything:

  1. Geometry family - the loop is not actually closed in the CAD/CAM file, so the toolpath ends short of the start point.
  2. Process/timing family - the geometry is closed and the motion is correct, but the torch is not producing a cutting arc at the moment motion begins. The unburned segment appears at the start of the path and is only revealed when the loop closes back onto it.
Key diagnostic: if the uncut tab is always located at the start/end junction of the loop and its length scales with feedrate, the cause is timing, not geometry. If the tab appears at an arbitrary vertex or the gap length is constant regardless of feedrate, suspect the geometry.

Root Cause Analysis

1. Torch fires late relative to motion start

The controller normally executes a sequence of torch-on, wait for pierce, then start axis motion. If motion begins before a stable transferred arc exists, the machine traverses the first portion of the path with no cut. On a closed loop that lost segment is exactly the piece that would otherwise be removed on the return to the start point, so the part stays tethered. On an open cut the same fault only shortens the lead-in and is easy to miss.

Contributors to late firing:

  • Pierce delay set shorter than the actual pierce time for the material thickness in use.
  • The controller ignores or does not wait for the arc-transfer (arc OK / arc transferred) feedback and relies purely on a fixed timer.
  • Arc-transfer signal is wired but the input filter or debounce in the controller delays recognition, or the signal is not wired at all and the timer is the only gate.
  • Torch height at pierce is too high for reliable transfer, so the plasma unit retries and the arc establishes late.
  • Worn consumables (electrode/nozzle) lengthening and destabilizing transfer time.
  • Marginal air supply pressure or flow, or wet air, causing slow or intermittent arc establishment.

2. Loop not closed in the source geometry

In most CAD and nesting packages an entity that forms part of a closed region is displayed differently from a dangling or broken entity. In the package being used here, closed outlines render red or yellow and broken/open lines render grey. A grey segment means the chain has a gap - the endpoints do not coincide within the software tolerance, or a duplicate/overlapping entity has split the chain. The post-processor then emits a path that stops short.

3. Cut-off / motion stop before the loop completes

Less common, but a controller that terminates the arc at the programmed end point without any overburn will leave a hairline of unsevered material at the junction because the kerf is centered on the path and the trailing edge of the arc lags the torch center.

Diagnostic Procedure

  1. Verify geometry first - it is free. Open the drawing in the CAM package and check the outline colour. Red or yellow = closed loop. Grey = broken. Zoom to the start point of any grey chain, and use the software's join/close-contour or gap-healing function with a tolerance appropriate to the drawing units. Re-post the program.
  2. Watch the start of a cut with the machine dry. With the plasma unit isolated (or torch-fire output disabled at the controller), single-block the program and confirm the motion actually begins only after the pierce-delay window elapses. Note whether the controller reports waiting for an arc-OK input.
  3. Observe a live start. From a safe position, watch whether the arc has transferred to the plate before the gantry moves. If the torch is still in pilot arc or still ramping when the axes start, you have confirmed late firing.
  4. Check the arc-transfer path. Confirm the plasma unit's arc-transferred/arc-OK contact is landed on a controller digital input, that the input is mapped in the controller configuration, and that the input actually toggles. Force the input logically and confirm the controller sees the change.
  5. Time the pierce. Pierce a scrap coupon of the same thickness with motion inhibited. Measure how long the arc needs to blow fully through. That measured value, plus margin, is the minimum pierce delay to program.
  6. Inspect consumables and gas. Pull the electrode and nozzle. Check electrode pit depth and nozzle orifice roundness against the plasma unit's wear criteria. Verify supply pressure and flow at the machine while the torch is firing, not at rest.
  7. Test at two feedrates. Run the same closed loop at, for example, full programmed feed and half feed. If the uncut tab length roughly halves at half feed, the missing distance is time-based - a firing/timing fault. If the tab length is unchanged, it is geometry or path.

Corrective Actions

Finding Corrective action Verification
Grey / open outline in CAM Join or heal the contour, delete duplicate entities, re-post Outline renders as a closed contour colour; simulated path returns to its start point
Pierce delay shorter than measured pierce time Increase pierce delay to the measured value plus margin, per material and thickness Arc fully through plate before axes move; no uncut tab at start
Arc-transfer feedback not used Wire and map the plasma unit's arc-OK contact to a controller input; configure motion start to gate on that input, not only on the timer Controller holds motion until the input asserts; slow transfers no longer cause short cuts
No lead-in on closed loops Add a lead-in (and lead-out where the material and part tolerance allow) so the pierce occurs off the finished edge Pierce crater and any late-fire loss land in scrap, not on the part profile
Junction hairline at loop closure Add a small overburn / path overlap past the start point Slug drops free without tapping
Worn electrode or nozzle Replace consumables as a set; re-verify transfer time Consistent, immediate transfer on every start
Low or wet air supply Correct supply pressure/flow at the torch; service dryer and filters Stable pilot-to-transfer behaviour across repeated starts
Safety: Isolate the plasma power supply and lock out before touching the torch, consumables, or any wiring. The pilot arc circuit carries high open-circuit voltage. Never defeat the torch-fire interlock to force a test.

Why Lead-Ins Matter on Closed Loops

A lead-in decouples the pierce event from the finished profile. Without one, the torch pierces directly on the part outline, and every millisecond of late firing is subtracted from the part edge at exactly the point where the loop must close. With a lead-in, the same late firing consumes lead-in length instead, and the profile is cut by a torch that is already at full cutting condition and at steady feed.

Practical rules that follow from this:

  • Make the lead-in long enough that the worst-case late start is fully absorbed. If your feedrate test showed a 5 mm uncut segment, the lead-in must exceed that by a clear margin.
  • Place the pierce point inside the scrap side of the kerf - inside the slug for internal holes, outside the part for the perimeter.
  • Prefer an arc or angled lead-in entering tangentially so the transition onto the profile does not produce a divot or a feedrate dip.
  • Add a short lead-out or overburn past the closure point when the material tends to leave a hairline bridge.

Verification and Sign-Off

  1. Cut a test coupon containing at least three internal closed loops of different sizes (small hole, medium hole, slot) plus the outer perimeter.
  2. Confirm every slug drops free under its own weight with no tapping.
  3. Inspect the closure junction of each loop with a straight edge or callipers - no step, no bridge, no double-cut gouge.
  4. Repeat the same coupon at the two extremes of the feedrate range you actually run. Both must produce free slugs.
  5. Repeat once more after a cold start of the plasma unit and compressor, where air pressure and transfer behaviour are least favourable.
  6. Record the final pierce delay, lead-in length, and consumable set in the job or machine setup sheet, indexed by material and thickness. These values are thickness-dependent and must be re-established when the material changes.

Preventive Practice

  • Build a pierce-delay table per material type and thickness rather than carrying one global value across all jobs.
  • Standardise a CAM check step: verify all cut contours display as closed before posting. Treat any grey/open entity as a hard stop.
  • Track consumable life by number of pierces, not by cutting hours - piercing is the dominant wear mechanism and directly drives transfer time.
  • Log air pressure at the machine during cutting as part of daily checks; supply degradation is gradual and shows up first as intermittent short cuts on closed loops.
  • Where the controller supports it, always gate motion start on the arc-transfer input in addition to the timer. The timer alone cannot compensate for a slow transfer.

Why does my plasma cutter leave a small uncut tab on holes but cut straight lines fine?

Because the missing material is at the start of the path. On an open cut the loss just shortens the lead-in and goes unnoticed, but on a closed loop the path returns to that same start point and the uncut segment holds the slug in place. The usual cause is the torch firing late relative to motion start.

How do I tell if the problem is my CAD file or the machine?

Check the contour colour in the CAM package - closed outlines display as red or yellow while broken lines display grey. If the outline is closed, run the same loop at half feedrate; if the uncut tab length halves, the fault is time-based firing, not geometry.

How do I set the correct pierce delay?

Pierce a scrap coupon of the same material and thickness with axis motion inhibited and measure how long the arc needs to blow fully through the plate. Program that measured time plus a safety margin, and build a table per material and thickness rather than using one global value.

Should motion start be triggered by a timer or by the arc-OK signal?

Use both. Gate axis motion on the plasma unit's arc-transferred/arc-OK input so a slow transfer cannot let the machine move without a cutting arc, and keep the pierce delay as the minimum dwell after transfer.

Will adding a lead-in fix incomplete closed-loop cuts?

It masks the symptom effectively by moving the pierce and any late-fire loss into scrap material, so parts release correctly. Still correct the underlying pierce delay, arc-transfer wiring, consumables, or air supply, because a marginal start will eventually exceed any lead-in length.

Back to blog