Troubleshooting guide

CNC Crash Machine: Watch Three Times, Stop Once

Most crashes are not machine failures. They are offset, tool or fixture errors that a dry run would have caught. This guide is for engineers and setup operators who want a short, repeatable check routine before cycle start: what to look at, which symptoms point to which cause, and when stopping the spindle is the cheaper decision.

Get Instant Quote Talk to an Engineer

Dry run firstSingle blockDistance-to-goFeed hold ready

cnc crash machine setup on 5-axis machining center checking tool and fixture before cycle start

On this page

1
Six crash symptoms and what actually causes them
2
Why a cnc crash machine usually starts at the offset page
3
What a cnc crash machine does to the part, tool and spindle
4
When the part geometry makes a crash more likely
5
Signals that tell you to stop before the next block
6
Six steps to run before the spindle starts

Symptom to fix

Six crash symptoms and what actually causes them

Read the middle column before you touch the offset page. Most of these have one cause, not three.

Symptom	Likely cause	What to do
Rapid move into the fixture	Work offset set from a different corner	Re-probe the corner, verify G54 in distance-to-go
First cut is 2 mm too deep	Tool length measured on a different holder	Re-measure the tool, compare with the previous value
Alarm on the first Z rapid	Wrong work offset for the current fixture	Check the active G54 to G59 number before restart
Bore comes out tapered	Spindle or toolholder runout after a light hit	Indicate the holder, then check the spindle taper
Scraped finish on one side wall	Chip packed in the vise jaw or soft jaw	Clean the jaw, re-clamp, re-zero the part
Loud bang, no axis motion	Coolant nozzle or probe arm left in the travel path	Retract both, then re-run the dry pass

Part 1

Why a cnc crash machine usually starts at the offset page

A crash is rarely a mystery. Look at the last ten minutes before the spindle started and you will find the cause: an offset number typed by hand, a tool swapped without re-measurement, a fixture moved between jobs. The machine did exactly what the control told it to do.

The common thread is that the control has no idea what the part looks like. It only knows the coordinates you gave it. A work offset that is 3 mm off in Z becomes a 3 mm plunge at rapid feed. On a 40 mm face mill running at 8,000 rpm, that is the end of the insert and often the end of the vise.

So the routine is built around verifying coordinates, not around watching the machine run. Watching the run is the last step, not the first. By the time the tool touches metal, the decision is already made.

1
Offsets are numbers, not intentionsThe control trusts G54 exactly as entered.
2
A crash needs one wrong valueNot three. One number is enough.
3
Verification happens before spindle startAfter the first cut, you are only recording damage.

Part 2

What a cnc crash machine does to the part, tool and spindle

Damage rarely stops at the broken tool. A 12 mm carbide end mill snapping at 10,000 rpm sends fragments in unpredictable directions. The vise jaw takes the hit next, then the table surface. A deep enough crash moves the fixture, which invalidates every offset you already set.

The spindle is the expensive part. A sudden radial load can push the taper out of concentricity, and you will not see it until the next job produces a tapered bore or a poor surface finish. Replacing a spindle cartridge is a five-figure repair in most shops, plus the downtime.

The part itself is usually the cheapest loss. It is one workpiece. The setup time, the fixture alignment and the operator's confidence are what actually cost you. That is why we treat a dry run as a production step, not as caution.

1
ToolBroken inserts and snapped shanks; check the holder too.
2
MachineTaper runout, ball screw shock load, way damage.
3
SetupFixture shifts, so every offset must be re-probed.

Part 3

When the part geometry makes a crash more likely

Thin walls and deep pockets remove the operator's visual margin. The tool disappears into the cavity and the only feedback is sound and load meter. A deep pocket also traps chips, and a packed chip can deflect a 10 mm tool enough to rub the wall on the way out.

Tall parts in a three-axis vise are another risk. Anything above 150 mm of stickout amplifies every positioning error, because the tool tip travels further than the offset error suggests. On a 5-axis machine, the rotary table adds a second coordinate system, and a wrong work offset in A or C sends the part into the tool from an unexpected direction.

Hard materials change the failure mode. In 17-4PH or Ti-6Al-4V, a wrong feed does not always break the tool. It pushes the part out of the vise instead, which moves the setup and forces a full re-zero.

1
Deep pocketsChip packing deflects small tools.
2
Tall setupsStickout multiplies positioning error.
3
Rotary axesA second offset must be verified separately.

Part 4

Signals that tell you to stop before the next block

The control gives you more warning than most operators use. Distance-to-go is the most direct one. On the first Z rapid, if the remaining distance looks shorter than the clearance you measured, stop the cycle and check the offset. That single glance prevents most Z-axis crashes.

Load meter and spindle sound come next. A sudden jump in spindle load on the first engagement usually means the tool is cutting more material than the program intended. Feed hold, lift Z, and inspect. Do not let the program finish the pass to see what happens.

Chip color and shape are slower signals, but they catch problems a dry run cannot. Blue chips in aluminium mean the feed or speed is wrong for that alloy. Long stringy chips in steel mean the chip breaker is not working at that depth of cut.

1
Distance-to-goCheck it on every first rapid.
2
Load meterA spike means the cut is deeper than planned.
3
Chip colorBlue in aluminium signals wrong parameters.

Before cycle start

Six steps to run before the spindle starts

Work through these in order. Each one takes under a minute on a normal setup.

1
1. Confirm the active work offsetRead the G54 to G59 number on screen and compare it with the setup sheet. Re-probe the corner if the part was moved or re-clamped. Do not assume the last job's offset is still valid.
2
2. Re-measure every tool after a swapTool length changes with the holder, not just with the tool. Compare the new value with the previous one; a difference above 0.05 mm means something moved. Keep the old number until the new one is verified.
3
3. Check tool stickout and clearanceMeasure Z clearance from the tool tip to the top of the stock, the vise and every clamp. Allow at least 5 mm of margin on the rapid plane. Retract the probe arm and aim the coolant nozzle away from the travel path.
4
4. Run the dry pass with the spindle offUse single block, rapid override at 25% or lower, and distance-to-go on screen. Watch the Z approach on the first hole. If the remaining distance looks wrong, stop before the tool reaches the part.
5
5. Cut air on the first partRaise the work offset by 20 mm in Z and run the full program once. This catches wrong tool numbers, missing subprograms and unexpected rotary moves without touching stock. Reset the offset afterward and re-verify.
6
6. Keep a hand on feed holdRun the first real part at reduced rapid and feed override. Your hand stays on feed hold until the first tool completes its path. Speed comes back after the process is proven, not before.

FAQs

Questions we get about crash prevention

How often should tool length be re-measured?

Every time the tool leaves the holder. Pulling a tool and putting it back is not a re-measurement, and the difference can reach a few hundredths of a millimeter. On a finishing tool that is enough to scrap the part.

Is a dry run enough on its own?

No. A dry run catches coordinate and path errors, but it cannot catch wrong feed, wrong material or chip packing. Run the air cut at a raised offset as well, then cut the first part at reduced override with your hand on feed hold.

What should we check after a light crash?

Indicate the toolholder first, then the spindle taper. Check the fixture for movement and re-probe the work offset. Inspect the ball screw area for shock marks if the crash was on a rapid move. Do not restart production until the first part measures within tolerance.

Does a 5-axis machine need extra checks?

Yes. Each rotary axis carries its own offset, and a wrong A or C value sends the part into the tool from a direction the operator is not watching. Verify the rotary zero point separately, then dry run the first rotary move in single block.

How do we build this into a shift routine?

Write the six steps on a card at the machine and make the setup sheet require a signature after step 4 and step 5. A short checklist that everyone follows beats a long procedure that nobody reads.

Can the machine itself detect a crash early?

Some controls offer load monitoring and rapid override limits. They help, but they react after contact. Nothing on the control replaces verifying the offset and the tool length before the cycle starts.